WO2015092430A1

WO2015092430A1 - Method, server, system and computer program product for supplying a message

Info

Publication number: WO2015092430A1
Application number: PCT/GB2014/053791
Authority: WO
Inventors: Fraser Robinson; Andrew Lucas
Original assignee: TAGGSTAR UK Ltd
Current assignee: TAGGSTAR UK Ltd
Priority date: 2013-12-19
Filing date: 2014-12-19
Publication date: 2015-06-25
Anticipated expiration: 2016-06-19

Abstract

There is provided a method of supplying an optimal message to add to a web page from a website, in which: an analysis server receives a request for an optimal message to add to a web page including a product, the request identifying the product; the analysis server fetches data from a database, the data relating to web server data in relation to the identified product; the analysis server analyses the fetched data so as to determine the optimal message to add to the web page including the product, and the analysis server sends the determined optimal message, in response to the request.

Description

METHOD, SERVER, SYSTEM AND COMPUTER PROGRAM PRODUCT FOR SUPPLYING A MESSAGE

BACKGROUND OF THE INVENTION

1. Field of the Invention

The field of the invention relates to methods, servers, systems and computer program products for supplying a message to add to a web page from a website.

2. Technical Background

For most web sites, the primary source of traffic is search engines. However, with so much competition and relatively little visual real estate on a search engine results page (SERP), it is extremely difficult to obtain a high search ranking. Even when a high search ranking has been achieved, it is still difficult to be the link selected and clicked by the individual given the number of search results on each page.

However, there are various ways to improve the chances of being noticed by a potential visitor, and thus improve the click through rate (CTR). This becomes particularly interesting to paid search advertisers who may appear further down the list, but can achieve a superior return on investment (ROI) if they receive a higher CTR than those near the top of the list. It is desirable for a website to try to modify website pages so as to obtain a higher click through rate. It is further desirable to modify internet advertisements so as to obtain a higher click through rate.

A user of a website may see a product for sale, but decide to delay purchase, for example to search for a cheaper alternative, or to attend to other matters, only to return later to find out that the product is now out of stock. It is desirable for a user to obtain some idea about if a stock may run out, so as to better inform any decision to purchase from a website. 3. Discussion of Related Art

GB2420949B, which includes prior art Figure 14, discloses a method of two-way communication between a web browser and a mobile telecommunication device (12) including steps of: accessing a web-site via a computer (1), sending a message to a mobile telecommunication device (12) from the web-site, and at a message server (4) capturing the IP address and port number of the computer (1), storing the temporary phone number, IP address of the computer (1) and port number of the computer (1) in a database (5), and sending the message to the mobile telecommunication device (12) with the temporary phone number.

SUMMARY OF THE INVENTION

According to a first aspect of the invention, there is provided a method of supplying an optimal message to add to a web page from a website, in which: an analysis server receives a request for an optimal message to add to a web page including a product, the request identifying the product; the analysis server fetches data from a database, the data relating to web server data in relation to the identified product; the analysis server analyses the fetched data so as to determine the optimal message to add to the web page including the product, and the analysis server sends the determined optimal message, in response to the request.

An advantage is that a user who receives the message can purchase a product before the stock runs out, whereas otherwise the user might not have made the purchase before stock runs out. This can save energy (eg. computer power, communications network power) expended in searching for the product if the stock from a supplier had run out, because the product can be purchased before it has run out. Furthermore, because the stock can be assessed across a plurality of websites, it is possible for a user to ascertain that the stock is likely to run out across a set of suppliers, and in response to purchase a product before the stock runs out. This will save even more energy (eg. computer power, communications network power) expended in searching for the product if the stock had run out, because if multiple suppliers have run out of the stock, it will be even harder to find the product elsewhere.

A further advantage is that a web page does not have to be redesigned in multiple ways to accommodate various messages. Instead, a web page can be designed such that an optimal message can be accommodated within a standard template. This causes the server or servers which supply the web pages to operate in a new way. The effect also operates at the level of the architecture of a system of servers. The method may be one in which the request identifies the product and a user, and the data relates to web server data in relation to the identified product and the user.

The method may be one in which the web server data is real time data. An advantage is that the message sent is as up-to-date as possible. The method may be one in which the analysis server receives the request from a first web server, the web server data is first web server data, and the analysis server sends the determined optimal message to the first web server, in response to the request. An advantage is that the system comprising the analysis server and the first web server operates in a new way, because the system provides content that is not available from the analysis server in isolation or from the first web server in isolation. A further advantage is that the message may be provided from the first web server, with minimal reconfiguration of the first web server.

The method may be one in which first web server data in relation to the identified product comprises web traffic data on the first web server in relation to the identified product and/ or current and future product price and availability data. The method may be one including the step of the analysis server receiving web server data in relation to products from the first web server, and saving the web server data in the database.

The method may be one wherein types of data being collected include one or more of: purchase, audience, inventory, review, "Add to Basket" rates, purchase rates and product price.

The method may be one in which the analysis server also tracks all user interactions, thus enabling automated content recommendations based on shopper behavior and product interests.

The method may be one in which in response to receiving the message, the first web server decides whether to bid for a paid search result, and if the bid is made and won, the first web server sends the paid search result to a search engine webserver, including the message.

The method may be one in which the analysis server receives the request from an advertising server, and the analysis server sends the determined optimal message to the advertising web server, in response to the request. An advantage is that the system comprising the advertising server and the first web server operates in a new way, because the system provides content that is not available from the advertising server in isolation or from the first web server in isolation. An advantage is that the system comprising the analysis server and the advertising server operates in a new way, because the system provides content that is not available from the analysis server in isolation or from the advertising server in isolation. A further advantage is that the message may be provided from the advertising server, with minimal reconfiguration of the advertising server.

The method may be one in which the web server data in relation to the identified product comprises web traffic data on a plurality of web servers in relation to the identified product and/ or current and future product price and availability data.

The method may be one further including the step of the analysis server receiving web server data in relation to products and users, from a plurality of web servers hosting a plurality of web sites, and saving the web server data in the database.

The method may be one in which the advertising server cookies a user.

The method may be one in which the advertising server takes the text message as supplied by the analysis server and inserts it into their ad creative before the completed creative is finalized and presented to an individual shopper in their browser.

The method may be one in which the analysis server passes relevant product and audience data to the advertising server in real time.

The method may be one in which the advertising server uses RTB exchanges and uses information as to whether or not an ad contains an analysis server message to determine how much they are willing to bid for a slot. The method may be one in which analysis server data messaging can be included in the creative to help increase CTR and therefore the return on investment (ROI) value to the advertiser of the ad. The method may be one in which the analysis server receives the request from a web browser running on a user terminal for an optimal message to add to the web page including the product, and the analysis server sends the determined optimal message to the web browser, in response to the request. An advantage is that the system comprising the analysis server and a user terminal running the browser operates in a new way, because the system provides content that is not sent out from the analysis server in isolation or which is present in the user terminal in isolation. A further advantage is that the message may be provided from the analysis server, with only minimal reconfiguration of a server which is hosting the website.

The method may be one in which the web server data in relation to the identified product comprises web traffic data on a plurality of web servers in relation to the identified product and/ or current and future product price and availability data. The method may be one further including the step of the analysis server receiving web server data in relation to products and users, from a plurality of web servers hosting a plurality of web sites, and saving the web server data in the database.

The method may be one in which the analysis server uses social proof to generate two distinct types of consumer sentiment: Urgency and Positive Validation; these types of messages can co-exist and appear on a page at the same time.

The method may be one in which the determined optimal message includes a standard width UI message, and/ or a narrow width UI message.

The method may be one in which the determined optimal message includes text and/or graphics.

The method may be one in which the analysis server includes an algorithm that identifies increases in the rate of sale of products.

The method may be one in which the web server provides messages in page. The method may be one in which the web server provides messages as informational balloons that can fade in/ out, or slide in/ out.

The method may be one in which a position of the message on the web page is varied by the analysis server based on performance.

The method may be one in which the web server displays the message at one or more of the following stages: home page, Search results /gallery page, Product page, Basket section.

The method may be one in which the message is applied in display advertising.

The method may be one in which the message is applied in search results: in natural or paid search.

The method may be one in which if the web server is able to map the visitor back to an email address, then that web server can follow up with an email to the individual that contains personalized content. The method may be one in which the method is used in personalized television advertising.

The method may be one in which code is installed in a website, to collect website data. The method may be one in which the code is JavaScript code.

The method may be one in which the optimal message includes one or more of "how many others are looking at this product", "when was the last one purchased", "how many have been booked today" and "what do others think of this".

The method may be one in which the analysis server includes machine-learning algorithms that dynamically test and optimise performance. The method may be one in which the machine-learning algorithms vary one or more of: data thresholds, message combinations, message tone, message design or colour, the number of messages, duration of a balloon on a web page, position of message on page. According to a second aspect of the invention, there is provided an analysis server, the analysis server configured to supply an optimal message to add to a web page from a website, in which: the analysis server is configured to receive a request for an optimal message to add to a web page including a product, the request identifying the product; the analysis server is configured to fetch data from a database, the data relating to web server data in relation to the identified product; the analysis server is configured to analyse the fetched data so as to determine the optimal message to add to the web page including the product, and the analysis server is configured to send the determined optimal message, in response to the request. The analysis server of the second aspect of the invention may be configured to perform a method according to any aspect of the first aspect of the invention.

According to a third aspect of the invention, there is provided a computer program product, arranged such that when running on an analysis server, the computer program product is configured to supply an optimal message to add to a web page from a website, the computer program product arranged to: receive a request for an optimal message to add to a web page including a product, the request identifying the product; fetch data from a database, the data relating to web server data in relation to the identified product; analyse the fetched data so as to determine the optimal message to add to the web page including the product, and send the determined optimal message, in response to the request.

The Computer program product according to the third aspect of the invention may be configured to perform perform a method according to any aspect of the first aspect of the invention.

There is provided a system for supplying an optimal message to add to a web page from a website, the system comprising an analysis server and a first web server in connection with the analysis server, in which: the analysis server is configured to receive from the first web server a request for an optimal message to add to a web page including a product, the request identifying the product; the analysis server is configured to fetch data from a database, the data relating to first web server data in relation to the identified product; the analysis server is configured to analyse the fetched data so as to determine the optimal message to add to the web page including the product, and the analysis server is configured to send the determined optimal message to the first web server, in response to the request.

There is provided a system for supplying an optimal message to add to a web page from a website, the system comprising an analysis server and an advertising server in connection with the analysis server, in which: the analysis server is configured to receive from the advertising server a request for an optimal message to add to a web page including a product, the request identifying the product; the analysis server is configured to fetch data from a database, the data relating to web server data in relation to the identified product; the analysis server is configured to analyse the fetched data so as to determine the optimal message to add to the web page including the product, and the analysis server is configured to send the determined optimal message to the advertising server, in response to the request. There is provided a system for supplying an optimal message to add to a web page from a website, the system comprising an analysis server and a user terminal running a web browser, the user terminal in connection with the analysis server, in which: the analysis server is configured to receive from the web browser a request for an optimal message to add to a web page including a product, the request identifying the product; the analysis server is configured to fetch data from a database, the data relating to web server data in relation to the identified product; the analysis server is configured to analyse the fetched data so as to determine the optimal message to add to the web page including the product, and the analysis server is configured to send the determined optimal message to the web browser, in response to the request.

According to a fourth aspect of the invention, there is provided a method of supplying an optimal message to add to a web page, comprising the steps of: (i) an analysis server receiving a request from a web browser running on a user terminal for an optimal message to add to a web page including a product, the request identifying the user;

(ii) the analysis server fetching data from a database, the data relating to web server data in relation to the identified product or the user;

(iii) the analysis server analysing the fetched data so as to determine the optimal message to add to the web page including the product and being viewed by the user, and

(iv) the analysis server sending the determined optimal message to the web browser, in response to the request.

The method may be one in which the web server data is real time data.

The method may be one in which the web server data in relation to the identified product or user comprises web traffic data on a plurality of web servers in relation to the identified product or user and/ or current and future product price and availability data.

The method may be one further including the step of the analysis server receiving web site traffic data in relation to products and users, from a plurality of web servers hosting a plurality of web sites, and saving the web site traffic data in the database.

According to a fifth aspect of the invention, there is provided an analysis server configured to supply an optimal message to add to a web page, the analysis server configured to:

(i) receive a request from a web browser running on a user terminal for an optimal message to add to a web page including a product, the request identifying the product or the user;

(ii) fetch data from a database, the data relating to web server data in relation to the identified product or the user;

(iii) analyse the fetched data so as to determine the optimal message to add to the web page including the product or being viewed by the user, and

(iv) send the determined optimal message to the web browser, in response to the request. The analysis server may be further configured to perform the method according to any aspect of the fourth aspect of the invention.

According to a sixth aspect of the invention, there is provided a computer program product executable on an analysis server, the computer program product when running on the analysis server arranged to supply an optimal message to add to a web page, the computer program product configured to:

(iv) send the determined optimal message to the web browser, in response to the request.

The computer program product may be further configured to perform the method according to any aspect of the fourth aspect of the invention.

According to a seventh aspect of the invention, there is provided a method of supplying product recommendations and associated web links to add to a web page from a website, comprising the steps of:

(i) an analysis server receiving from a website server a recent activity record of a user who has been accessing the website on the website server, and an identification of the user;

(ii) the analysis server analysing the recent activity record to determine that the user is likely to leave the website soon;

(iii) in response to the determination that the user is likely to leave the website soon, the analysis server using the identification of the user to generate a set of product recommendations of third party products, together with web links to third party websites which supply the third party products, and (iv) the analysis server sending the set of product recommendations of third party products, together with the web links to the third party websites which supply the third party products, to the website server, for inclusion in the web page from the web site. The analysis server may be provided. A computer program product executable on the analysis server so as to perform the method according to the seventh aspect of the invention may be provided.

According to a eighth aspect of the invention, there is provided a method of supplying product recommendations and associated web links for inclusion in a web page from a website, comprising the steps of:

(i) an analysis server receiving from a web browser running on a user terminal a recent activity record of the user who has been accessing a website, and an identification of the user;

(iii) in response to the determination that the user is likely to leave the website soon, the analysis server using the identification of the user to generate a set of product recommendations of third party products, together with web links to third party websites which supply the third party products, and

(iv) the analysis server sending the set of product recommendations of third party products, together with the web links to the third party websites which supply the third party products, to the web browser, for inclusion in the web page from the web site. The analysis server may be provided. A computer program product executable on the analysis server so as to perform the method according to the eighth aspect of the invention may be provided.

According to a ninth aspect of the invention, there is provided a method of supplying data suitable for including in a web page including a product, comprising the steps of:

(i) configuring an analysis server such that it can send to an advertising server data relating to the product;

(i) the analysis server fetching data from a database, the data relating to web traffic in relation to the product; (iii) the analysis server analysing the fetched data so as to determine data suitable for adding to the web page including the product, and

(iv) the analysis server sending the data to an advertising web server, and an identification of the product.

The method may be one further comprising the steps of:

(v) the advertising web server including the data into a pre-configured unit relating to the data and to the product so as to create a message, and

(vi) the advertising web server including the created message in the web page.

The analysis server may be provided. A computer program product executable on the analysis server so as to perform the method according to the ninth aspect of the invention may be provided. The servers referred to in this document may be stand-alone, real, virtual or in the cloud, as would be clear to one skilled in the art.

BRIEF DESCRIPTION OF THE DRAWINGS

The above and other aspects of the invention will now be described, by way of example only, with reference to the following Figures, in which:

Figure 1 shows examples of Standard Width UI Messages (Desktop and Tablet).

Figure 2 shows examples of Narrow Width UI Messages (Mobile).

Figure 3 shows an example of an Audience Rule Table.

Figure 4 shows an example of a Purchase Rule Table.

Figure 5 shows an example in which balloon overlays fade in then out, informing the shopper without intruding and in-page, real time messaging is provided, displayed on a fixed computer.

Figure 6 shows an example in which messaging in listings pages helps drive discovery and call to action, displayed on a fixed computer.

Figure 7 shows an example of driving conversion in mobile devices too, displayed on a mobile phone.

Figure 8 shows an example in which a mobile data banner fades in then out, and in-page messaging is provided, displayed on a mobile phone.

Figure 9 shows an example in which an analysis server dynamically replaces some or all of a retailer's own product recommendations and suggests related products from 3rd party retailers instead, in the box.

Figure 10 shows an example of a system, in which fixed user terminals 1 to n are in connection with the internet, and mobile user terminals 1 to m are in connection with the internet, such as via cellular networks or via wifi networks. A website server for site A, a search engine X server and a Taggstar server are in connection with the internet.

Figure 11 shows an example of a system, in which fixed user terminals 1 to n are in connection with the internet, and mobile user terminals 1 to m are in connection with the internet, such as via cellular networks or via wifi networks. A first website server, a second website server, an advertiser server and a Taggstar server are in connection with the internet.

Figure 12 shows an example of a system, in which fixed user terminals 1 to n are in connection with the internet, and mobile user terminals 1 to m are in connection with the internet, such as via cellular networks or via wifi networks. A website server for site A, website server for site B and the Taggstar server are in connection with the internet. Figure 13 shows an example of a system, in which fixed user terminals 1 to n are in connection with the internet, and mobile user terminals 1 to m are in connection with the internet, such as via cellular networks or via wifi networks. A first website server, an advertiser server and a Taggstar server are in connection with the internet.

Figure 14 shows a prior art communication system disclosed in GB2420949B, including a mobile telecommunication device (12), a computer (1), the internet (2), a web server (3), a message server (4), and a database (5).

DETAILED DESCRIPTION

Using real time audience and product data to optimize "onsite" sales conversion, as well as the "offsite" performance of display advertising, retargeting and natural or paid search.

Taggstar improves shopping conversion and drives user engagement by providing real time social proof and persuasive messaging and content recommendations on web sites, as well as in offsite marketing. For example, informing shoppers about "how many others are looking at this product", "when was the last one purchased", "how many have been booked today" and "what do others think of this", helps to inform a buying decision while creating a sense of urgency that delivers superior shopping conversion. Taggstar also tracks all user interactions, thus enabling automated content recommendations based on shopper behavior and product interests. This type of information is often referred to as 'social proof or 'persuasive messaging' and appeals to the behavioral psychology of consumers, in examples such as:

"What if they run out of stock?

"This one seems to be selling fast. I'd better act now."

"Wow. . .there are five other people looking at this right now."

"80% of people would recommend it."

"This one always sells well. It must be good."

Product pages on Ecommerce sites are usually cluttered and the shopper is inundated with huge amounts of information, most of which is overwhelming and too difficult to absorb quickly. Taggstar analyses all the available information on a page and distills it down to key messages that help the shopper to make a fast and informed buying decision. For example, product reviews are hugely valuable, but they are often verbose and too numerous to read. For example, a product may have 35 extensive reviews, star ratings and "would/would not recommend" data points. Taggstar analyses the available review data and for example informs the shopper in a simple manner such that for example "80% of reviewers gave this 5 stars." Taggstar analyses large amounts of online data in real time and generates the message, or combination of messages, that drives the most positive outcome. In eCommerce, that outcome is typically an increase in sales conversion, or greater customer engagement, as measured by some kind of desired action such as click rate.

Taggstar uses social proof to generate two distinct types of consumer sentiment:

1. Urgency— few things are more annoying to the consumer than realizing they've missed out on a purchasing opportunity ("fear of missing out"), usually because a product is out of stock. Hesitation, the desire to price compare elsewhere or wanting to research a product further are often reasons for delaying a purchase decision. However, providing inventory information is only part of the story. Knowing when a product was last purchased, how many have sold today and how many people are looking at it right now gives greater context to the "how many are left" data point, driving faster decisionmaking on the part of the consumer, and higher sales conversion.

2. Positive Validation — social proof also inspires consumer confidence in a particular product choice by letting them know certain types of validating information: a. Perennial best seller: "Great choice! This one is a customer favourite and always sells well.

b. Summary review data: "85% of people would recommend this.

Taggstar uses both Urgency and Positive Validation to help inform the customer about the product they are considering purchasing. These types of messages can co-exist and appear on a page at the same time.

In parts of this document, the "copy", "message" or "text string" being communicated to the end user is referred to as the "Data". Data Types

There are a number of key types of data being collected in order to generate social proof and persuasive messaging in real time: 1. Purchase — tracking product purchases enables Taggstar to measure and communicate product popularity based on absolute number of purchases of a product, as well as based on rates of purchase.

2. Audience— by tracking the number of people who are looking or have looked at specific product pages, Taggstar can communicate that data back to the shopper, thus giving a sense for how much 'buzz' or activity there is around products.

3. Inventory— certain sites make product inventory available, either visibly on the page itself, or via an API. Taggstar takes this data point which, when combined with rates of purchase, enables messaging based on when a product will run out given current rates of purchase.

4. Review — certain sites include product review information as provided by customers. There is often too much review data to be easily consumed by a hurried shopper. For example, Taggstar would harvest the available review data on a product page and inform the consumer in a simple manner that "80% of reviewers gave this 5 stars", or "90% of customers would recommend this."

In addition to the above, Taggstar tracks a number of other data points for reporting purposes, "Add to Basket" rates, purchase rates and product price. Sample message strings:

Standard Width UI Messages (Desktop and Tablet): see Figure 1 for examples.

Narrow UI Messages (Mobile) (e.g. en-GB locale)

Messaging is typically abbreviated for the narrow UI (i.e. mobile) so as to reduce word count: see Figure 2 for examples.

'Hot product' messaging

When a product is selling at a faster rate than is normally seen for that product, or if the number of visitors is higher than usual, it is helpful to inform the shopper of that fact. This gives context to the social proof message already being provided and drives incremental uplift in conversion rate. For example, if a product typically sells 10 units in an average day, but on one particular day 20 have sold by 12pm, then it is useful to the shopper to inform them of this fact. If a shopper has never seen this particular product and they are simply told that '20 have sold today', it may be difficult for the consumer to evaluate if this is a high, low, or average number. By including copy and graphics in the message that emphasizes the high rate of sale, the consumer is better informed and can take action.

An example of how a graphic or visual style can denote a trending or popular product could be the use of a flame, an orange or red colour scheme, combined with copy such as "Selling fast", or "Wow! This one's popular."

Taggstar have developed an algorithm that identifies increases in the rate of sale of products, the objective being to highlight increases in rates of sale through the display of a 'selling fast' user interface (UI) message. The ultimate objective being to generate visitor add to basket uplift.

The algorithm filters out over 90% of 'low volume' products that have less than 10 purchases in a 48 hour window as these products do not large have large enough increases in rates of sale for them to be highlighted as 'hot'. E.g. An increase in the quantity purchased over 48 hours value from 2 to 3 does not make a product 'hot' despite that fact that this is a 50% increase.

The other reason why low volume product must be excluded is that low volumes in sales would make proving statistical significance difficult or impossible.

Onsite vs Offsite uses of Data

Taggstar uses social proof messaging to influence consumer behavior both on the client's web site ("Onsite"), as well as away from the web site ("Offsite").

Examples of offsite uses of data include in display advertising and search.

Onsite Messaging Taggstar delivers onsite messaging that is visible to a web site visitor in two key ways:

1. In page— messaging appears in the page much like other pieces of information, such as price or product description.

2. Balloons— Taggstar serves informational balloons that can fade in/ out, or slide in/ out. Balloons can be positioned anywhere on a page, with variable colours, message types, speed of appearance/ disappearance etc.

In both cases, the position of the message in/ on the page is varied by Taggstar based on performance (see below section on Optimisation, for example).

Onsite data can be displayed to a visitor to a web site in a number of sections of a web site. In the case of an eCommerce site, this means persuasive messaging can appear at different stages of the shopping funnel, as summarized below: a) Home page— modules displaying most recently purchased items, trending items and products getting a lot of activity help engage people who have just arrived at the site, while building a feeling of trust to those who may be new to that particular brand. Knowing that others are purchasing from a site gives a prospective customer confidence. b) Search results /gallery page — the search results page is a key area of any eCommerce site in that the customer has declared an interest in a specific product or product type. By displaying social proof messaging in the product details area of a search results page, potential customers are informed about the popularity of various products and are more likely to engage by clicking on that product. By converting more browsers to the next step of the shopping journey, the retailer is increasing the opportunity to convert to sale.

c) Product page— the product page or product detail page is where the shopper can consume the most detailed content that relates to a product. It is here that customer reviews, detailed product description, multiple images, video, detailed delivery options and so on are all available. It is here that the all-important "Buy" or "Add to basket" button typically appears. By including real time persuasive messaging on this page, the retailer is able to drive a higher conversion rate.

d) Basket section— the basket is the area of the web site where a shopper has selected the items they wish to purchase and prepares to make a payment. Basket (or cart) abandonment continues to be a significant problem for eCommerce businesses. 50% or higher drop offs are not uncommon. Using social proof messaging in the basket area of a web site helps reduce abandonment rates. Many basket sections have multiple stages, such as confirming an order, registering, providing delivery details and so on. Rather than repeat the same social proof messaging at each stage, which can cause irritation to the shopper, the type of message varies depending on how far along the basket the customer is. For example, the first step of the basket might include Validation messaging which reminds the customer that they've made a great choice. In later steps of the basket, the messaging might switch to Urgency, letting the shopper know that the product is selling well and that they should complete the transaction so as not to miss out.

Offsite messaging Taggstar's real time data messaging can be applied to all types of marketing materials, broken down into two key segments:

Display advertising: Includes standard display formats (e.g. banners, skyscrapers, mid page units (MPUs)), ad retargeting, email campaigns, television and non virtual formats such as digital billboards.

Search: Includes natural and paid search.

1. Online Display Advertising

Display advertising makes up approximately 20% of all digital marketing spend (the majority goes towards search). Standard display ad formats (as defined by the Internet Advertising Bureau) include banners, skyscrapers and MPUs. Standard display ads often target individuals based on location, demographics, subject matter of publisher where the ad appears.

Performance of display ads is based on click through rate (CTR). The price of the ads is dynamically optimized by real time bidding platforms (RTB). Taggstar data messaging can be included in the creative to help increase CTR and therefore the return on investment (ROI) value to the advertiser of the ad.

Advertisers using RTB exchanges could potentially take the information as to whether or not an ad contains Taggstar data to determine how much they are willing to bid for a slot (i.e. if you know a piece of creative contains Taggstar data that improves performance, then perhaps bid more for the media).

2. Online Ad Retargeting

Ad Retargeting is an online advertising technology that serves customized ads to people who have indicated an interest in a product on your website. This differs from standard display advertising in that retargeted individuals will see ads for a product they have already looked at when they navigate away to a third party site. For example, a shopper who views a pair of jeans on site A, might see an ad for that same pair of jeans when visiting site B. Clicking the ad would take the shopper back to the jeans product page on Site A.

These ads are managed and served by businesses that specialize in ad retargeting (e.g. Criteo, MyThings, AdRoll) using proprietary tech media trading platforms.

Taggstar aims to use the real time audience and product information it is capturing from retailers to optimize the performance of ad retargeting creative. Emphasis here is on the real-time nature of the data. This is achieved by including persuasive messaging in the ad creative itself.

Using the example of the "jeans" on "Site A" above, retargeted ad creative currently displays an image of the jeans, together with the price and retailer in hopes of triggering a click response from the shopper.

Taggstar provides the retargeting business with snippets of text to include in the ad creative. The result is that the new creative would show the thumbnail image of the jeans, the price, the retailer, and now, the Taggstar data, such as: "50 people have looked at these jeans since you did."

"25 have been purchased since you last looked."

"Only 5 left!"

"At this rate they'll be gone by the end of today."

This type of information and messaging drives an increase in ad Click Through Rate (CTR) and thus improved ROI for the retargeting business and advertiser.

3. Email retargeting

Email remains one of the most high-performing marketing channels and within that channel, email can be used in much the same way that display retargeting works.

When a visitor to an eCommerce site looks at various products, if the retailer is able to map the visitor back to an email address (typically if a visitor has been cookied and identified and if an associated email address exists), then that retailer can follow up with an email to the individual that contains personalized content. In most cases, this personalized content is based on the products in which that customer has shown an interest.

By including Taggstar social proof messaging, the engagement rates of retargeted emails can be improved.

4. Television

As television migrates online and becomes increasingly personalized, consumers will find themselves able to respond to advertising in real time. This could include engaging with a TV advertisement that leads directly to a purchase on the same screen. Retailers and brands will be able to use Taggstar's real time social proof technology to populate television advertising, much in the same way that it does standard display. This becomes possible as the television industry migrates towards ad serving technology only previously seen online. 5. Non Virtual Billboards

With the growing presence of digital screens as replacements for posters, the opportunity arises for brands to incorporate real time social proof elements into the creative to boost call to action. While the digital screens themselves may not be interactive, the mobile phones or other portable devices are, and the call to action may transfer to the device used by the individual. Furthermore, even without an immediate call to action, the very nature of the social proof messaging can provide powerful brand awareness and a halo effect relating to the perceived success and popularity of the brand.

How does the Taggstar data make its way into the advertising creative?

Note: for Taggstar to work with retargeting, the retailer/ advertiser in question must have both Taggstar's JavaScript code and the retargeter's tracking pixel installed.

When Taggstar is installed on a site (e.g. the first website server of Figure 11), data is being collected in order to deliver the persuasive messaging. The ad platform or retargeter ("Advertiser") (e.g. the Advertiser server of Figure 11) cookies the customer and collects data pertaining to which specific products they've looked at so that when that customer navigates away to a different third party site (e.g. second website server of Figure 11) within the Advertiser's ad network, they are able to serve up an ad displaying one of the products already viewed.

When the Advertiser has identified a customer, they match the correct product image (e.g. jeans) to that customer and push a bespoke advertisement. The Advertiser simultaneously calls the Taggstar application programming interface (API) (e.g. on the Taggstar server of Figure 11) with the product identification (ID). Taggstar takes the product ID and uses it to gather the relevant audience and purchase data that relates to that specific product. Once the data has been assembled by Taggstar, the most effective type of message is automatically selected, the text is assembled and then instantly passed back to the Advertiser (e.g. the Advertiser server of Figure 11) via API.

The Advertiser (e.g. the Advertiser server of Figure 11) takes the text message as supplied by Taggstar (e.g. the Taggstar server of Figure 11) and inserts it into their ad creative before the completed creative is finalized and presented to the individual shopper in their browser. An example system implementing the invention is shown in Figure 11. In Figure 11, fixed user terminals 1 to n are in connection with the internet, and mobile user terminals 1 to m are in connection with the internet, such as via cellular networks or via wifi networks. In Figure 11, a first website server, a second website server, an advertiser server and a Taggstar server are in connection with the internet.

An alternative and potentially more efficient data flow is that the API call to Taggstar (e.g. on the Taggstar server of Figure 11) can be made by the Web Browser (e.g. running on one of the fixed user terminals or mobile user terminals of Figure 11), instead of coming from the Advertiser, and in turn, Taggstar can pass the data message directly back to the Web Browser. From a customer/ end-user perspective, the end result is the same. This is simply a different way of managing the flow of information between customer, Advertiser and Taggstar. An example system implementing the invention is shown in Figure 11.

When working with online advertising, Taggstar's challenge is to pass the relevant product and audience data to the Advertiser in real time, for them to include when ad creative is pushed to the consumer.

Operational flow: i. Shopper goes to Site A and looks at a pair of jeans (note: Taggstar and Advertiser are both installed on Site A).

ii. Advertiser cookies the Shopper as having viewed the jeans. Taggstar tracks audience and purchase data relating to those jeans.

iii. Shopper leaves site A without having made a purchase and decides instead to goes to Site B.

iv. Site B allows display advertising and is within the Advertiser's ad network.

v. Advertiser identifies the Shopper as having previously looked at the Jeans on Site A and prepares to show the Shopper a display ad (e.g. banner, skyscraper, mid page unit (MPU)) promoting the Jeans.

vi. Advertiser (or Web Browser) calls the Taggstar API, supplying the product ID for the Jeans. vii. Taggstar returns the most appropriate message directly to either the Advertiser or Web Browser. This message is made up of real time audience, purchase and review data that pertains to those jeans at the time the ad is served. Key here is that this information is not generic, but relates to the product and its real time (i.e. right now) performance. viii. Advertiser or Browser receives the optimal message from Taggstar and includes it in the ad, along side the other information that makes up the creative (e.g. thumbnail image of jeans, price, retailer etc).

ix. Shopper sees display ad and either reacts to the product by clicking the ad, or does nothing.

Search

In much the same way that Taggstar audience and product data can be applied to display advertising and retargeting to improve media performance, the same data can be used to optimize the click rates of natural and paid search results on search engines like Google and Bing.

A high search ranking in Natural (i.e. not paid) search results is typically achieved through Search Engine Optimisation (SEO). A high ranking in Paid Search results (e.g. the right column on Google SERPs) can be obtained by bidding the highest Cost Per Click (CPC) amount for key terms. For example, travel businesses might bid on searches for "flights to New York". CPC rates vary wildly based on variables that include business vertical, conversion rates, margin and so on. Those who have bid the most for traffic will appear at the top of the paid search listing. As you move down the list, the price being offered is lowered, based on the notion that those at or near the top tend to get the most clicks. However, there are various ways to improve the chances of being noticed by a potential visitor, and thus improve the CTR. This becomes particularly interesting to paid search advertisers who may appear further down the list, but can achieve a superior ROI if they receive a higher CTR than those near the top of the list.

One such technique is to include star ratings about products or web sites directly in the search result itself. Google has revealed that this sort of information can drive up to a 15% increase in CTR. Taggstar believes the CTR performance of these search results can be further improved by the inclusion of Taggstar' s audience and purchasing data.

For example, if a shopper is searching for a Samsung Galaxy S4 phone, they might see a SERP such as one from Google.

There are many retailers to choose from and each uses different promotional techniques to attract a potential customer. These include star ratings, thumbnail images (in the right column on a Google SERP), price, as well as special deals, such as "Save £5 on monthly bill."

Taggstar provides including audience and purchase information in SERPs, such as "how many people are looking at this right now", "how many have sold recently" as a means to drive customer awareness and draw them to a particular store. This kind of shopper psychology is often referred to as "social proof, and is based on the notion that consumers prefer a busy/ popular retailer (or restaurant) to one that is empty.

Taggstar believes that the inclusion of this data improves the CTR of both natural and paid search links. This in turn will increase the ROI of paid search marketing spend by online advertisers.

Search Results Integration

Similar to the Retargeting solution, Taggstar provides an API that delivers messaging into advertising to help improve the performance of the ad. Important note: sites who wish to use Taggstar data in their search results listings need to have the Taggstar JavaScript snippet installed on their site. i. Browser (for example on a user terminal of Figure 10) goes to Search Engine X (for example on search engine X server of Figure 10) and types a search e.g. "Samsung Galaxy S4 phone".

ii. Ecommerce "Site A" (for example, on website server for site A of Figure 10) sells this particular product and has bid on the search term "Samsung Galaxy S4 phone" on Search Engine X.

iii. Site A ad management platform calls the Taggstar data API (for example on Taggstar server of Figure 10), providing the product ID of the Samsung Phone.

iv. Taggstar uses that product ID to fetch the related audience and purchasing data for the Samsung Galaxy S4 phone on Ecommerce Site A.

v. Taggstar then analyses the data it has about the Samsung Galaxy 4 phone on Site A and determines what type of message will be most beneficial to improving the performance of the ad.

vi. Once the optimal message has been selected by Taggstar, Taggstar returns the data to Site A ad management platform.

vii. Site A decides if a bid will be made, and the price of the bid. If a bid is made and won, the SERP loads with Site A's link/ ad, which includes the Taggstar message.

An example system implementing the invention is shown in Figure 10. In Figure 10, fixed user terminals 1 to n are in connection with the internet, and mobile user terminals 1 to m are in connection with the internet, such as via cellular networks or via wifi networks. In Figure 10, a website server for site A, a search engine X server and a Taggstar server are in connection with the internet.

Note: the real time nature of Taggstar's data is very important.

Real-time ad pricing

All display advertising, retargeting and paid search are run on exchanges that serve as marketplaces for buying and selling advertising media. Over the last few years, most have moved to real time programmatic buying and selling of media based on the performance (CTR) of ads and upper or lower price thresholds based on availability of media as set by both advertisers and publishers. Taggstar provides and incorporates real time purchase and audience data into ad exchanges to help with the bid/ no bid decision, and the bid amount.

The ad management platform can dynamically alter the price it's willing to bid based on whether or not Taggstar data is available and performing.

Optimisation

Taggstar's performance is measured by its ability to improve shopping conversion, as well as browser return rates (e.g. if a customer typically returns 48 hours later to purchase a product they have previously looked at, Taggstar aims to shorten this to less than 48 hours. This is because the customer has a sense for a product's popularity and not wanting to miss out).

Taggstar constantly measures performance of the different message types, balloon colours and even sentence structure of messaging in order to determine which information delivers the best result.

Optimisation The key psychological effects of social proof messaging are buyer urgency ("don't miss out") and product validation ("great choice"). However, there are many different variables that go into creating the ideal persuasive message and as a result, the scope for optimization is huge. Taggstar has developed machine-learning algorithms that dynamically test and optimise performance "on the fly".

The following are a selection of the main elements that can be varied: 1. Data thresholds— a customer responds differently to a message depending on the numbers that message contains, even when the fundamental message type remains the same. For example, informing a customer that "20 people have bought one in the last 2 hours" is more compelling than saying "2 people have bought one in the last 48 hours". But with the multitude of different message types, the variance between the performance of those messages is more nuanced. This is where Taggstar continually monitors not only the message types that drive the best response, but also the point at which the performance of a message type shifts based on the numbers it contains. 2. Message combinations— as detailed elsewhere in this document, Taggstar delivers messages to the site visitor about based on audience data, purchase data and review data for individual products in an online store. However, the combination of different message types can deliver significantly improved performance. For example, telling a customer that "3 others are looking at this right now" can influence purchase behavior up to a point. But by introducing another data point, so the message now reads "3 others are looking at this right now and 5 have bought one in the last 2 hours" can significantly improve the customer's decision to purchase. Taggstar monitors the effect different combinations of message type have on consumer behavior and thus biases the messaging on those learnings.

3. Copy/ tone of voice— as in natural language, customers respond differently to the same core message expressed in a different manner, or tone of voice. Taggstar experiments with sentence structure, phrasing and individual words or synonyms to understand what generates the most positive response. All this takes place on an individual site by site basis, where customer demographics can differ wildly from client to client.

4. Design/ colour— colour and design are critical elements to product performance. In the same way that businesses continually A/B test different shapes, colours and messages of buttons, such as "Buy", Taggstar does the same with the colour and design of message balloons.

5. Number of messages — given the various message types available, Taggstar experiments with delivering multiple messages. These can all be contained in one balloon (e.g. "3 people are looking at this right now and we've sold 5 in the last 30 minutes"), or in multiple balloons, where each balloon contains a single message type (e.g. "3 people are looking at this right now" might appear in one balloon, and "We've sold 5 in the last 30 minutes" could appear as a second balloon).

6. Duration of balloon on page— Taggstar's real time messaging balloons fade in then fade out on a page. The timing of those balloons plays an important part in the effectiveness of the messages contained within those balloons. Factors such as when the first balloon appears, how long it remains before fading out, as well as when the second balloon appears and subsequently fades out influence noticeability of messaging, as well as overall feel of the dynamic nature of the product.

7. Position on page— the effectiveness of real time messaging is heavily influenced by the position of the balloons on the page. While many eCommerce sites are similar in structure, their specific layout, colour schemes and density of information can vary widely. Taggstar's message balloons can appear anywhere on a page, and performance differs from site to site. As a result, Taggstar continually tests and optimizes for the most performing position. Key verticals

Taggstar data has applications to all types of web site, whether commercial or not. However, there are certain verticals where we believe its use will be particularly effective: Retail; Travel; Property; Job search; Finance.

Installing Taggstar

Taggstar is simple to install and requires a snippet of JavaScript to be copied and pasted into the HTML of your site template, as described below:

1. Copy and paste the below JavaScript into the site template. The script may be placed anywhere within the HTML <head> or <body> elements. We recommend placing it just before the closing </head> tag. See sample script below:

<!-- Start taggstar vl— >

(function (d,t) {

var siteKey = "xxxSITEIDxxx";

var

s=d.createElement(t),e=d.getElementsByTagName(t) [0],p=d.location.protocol;s. async=true; s.src=(p= ='https:?p:'ht^:')+'V/realtime.taggstar om/static/site/''+siteKey+''/ view-statjs";

e.parentNode.insertBefore(s,e);

} ) (document,'script');

</script>

<!-- End taggstar ->

2. The above JavaScript should also be copied and pasted into the template of all confirmation pages (in any part of the HTML, preferably before the closing </head> tag). This allows Taggstar to generate data messaging pertaining to purchases.

3. Taggstar is designed for desktop and mobile web and the JavaScript should therefore be installed in both areas. An API is available for use in your native mobile application.

4. Data messaging will only appear on your site when Taggstar activates your account, which we will do once you have confirmed you are ready to go live.

Note: installing the Taggstar JavaScript does not mean data messaging will instantly appear on your site.

Third Party White Label Partners If any parts of your site work with Third Party White Label Partners, the above JavaScript needs to be copied and pasted by them. Please forward these instructions directly to the relevant Third Party for installation.

Multiple/international sites

For businesses with multiple sites (e.g. international), Taggstar will provide separately configured snippets of JavaScript. This allows for local language messaging, as well as distinct reporting.

Bespoke UI

Taggstar's in-house UI team will create and implement all bespoke designs to suit the look and feel of the particular brand.

Platform Scalability

Taggstar's platform can be hosted in the Amazon EC2 Cloud and can be scaled horizontally to manage any traffic spikes or seasonal trends seamlessly.

Some examples of the data points Taggstar is capturing when installed on a web site.

1. Audience— how many people are looking/have looked.

2. Purchase history— how recently was this product last purchased. How many have been purchased, rates of purchase and so on.

3. Inventory— how many are remaining.

4. Time— relevant for real-time Audience measurement, as well as Purchase history.

5. Location— "this product was last purchased by someone in [London] 23 minutes ago."

• How many people are looking at this item right now?

• How many people have looked at this item in the last X period of time?

• When was it last purchased, and where are they? (e.g. Last purchased at 10:23am by someone in London, UK).

• Where are those people (e.g. London, UK)? • People looking at this destination (e.g. a hotel in New York) also looked at these hotels in New York.

• There are [25] people looking at NYC right now (e.g. in a flights or hotels category) .

· [50] people have booked a [hotel] in [NY] in the last hour.

• Travel dates: there are [20] people looking at this hotel right now, [5] of whom overlap with your chosen dates.

• Popularity: in-page module that displays most viewed or purchased right now/today/last 7 days, overall and by category. E.g. Most purchased destinations, dresses/red dresses, comedy shows today. Or most viewed. One example is a feed of latest purchases e.g. by flights vs hotels vs theatre etc.

• What's being purchased in a particular category? (e.g. shoes or dresses).

• What are the top [10] purchases [today/last 7 days] overall? By section?

• People who bought this also bought these.

· Velocity: number in stock / rate of purchase means we could estimate when it could go out of stock eg going fast "at this rate, they'll be gone by Thursday."

• Personalisation/recommendations: if a shopper looks at a particular hotel in New York, show them that same hotel the next time they visit, perhaps with a voucher code for a discount, tell them how many bookings of that hotel have been made since they were last on the site. Here are some other hotels that people booked who also booked the hotel you looked at.

• Summarise product review data and communicate that data clearly and simply. For example, if a particular product has been given a five star rating by 80% of reviewers, then deliver that message to help drive a purchase decision.

· Extend time horizon so we can say how many people have booked this hotel in the last 7 days.

• Map of the world showing real time purchases (real-time could be delayed for this kind of information).

• Use data messaging in retargeting creative— if a shopper looks at a hotel in New York, retarget them later on a third party site (e.g. via Criteo) saying how many people have booked that hotel in the last 24 hours (for example)

• Can we take learnings about customers for retargeting by competing businesses? E.g. Someone is looking at red dresses on Top Shop, can we retarget red dresses from Selfridges? • Is there something interesting to explore with Google around displaying product popularity info alongside natural and paid search results?

Technical Overview

System Actors 1. Retailer.

An organisation or business operating an E-Commerce web site.

2. Taggstar.

A software technology company.

3. Real Time Message System.

A Software -as-a-Service system providing a HTTP interface to JavaScript tag code executed by web browser applications having a method of storing data in a database and processing data.

4. Retailer Web Server.

A HTTP web server making available an E-Commerce web site where purchases can be made for goods or services.

5. Visitor.

An Internet user using a web browser application to interact with a Retailer's E- commerce web site

6. Advert System.

A software system maintained and operated by Taggstar for the purpose of optimising adverts and having a method of storing data in a database and processing data.

7. Ad Platform.

Any service operated for the purpose of displaying Internet advertising.

System Function The Taggstar Real Time Message System (the 'system' hereafter) displays 'persuasive' messages on the product pages of a Retailer's E-commerce web site. The objective of the system is to increase the Retailer revenues, when measuring revenue from those Visitors who view persuasive messages compared to those Visitors that do not. This measurement is achieved using A/B testing methods. The system must be integrated with two Visitor use cases that occur on the Retailer's E- commerce site. Use Case 1: Visit Product Web Page Use Case

In this use case the goal of the Visitor is to obtain information about a product and, or, to add the product to a basket/ cart. The main mechanisms of the system in this use case are: i ) record the visit event for the purpose of displaying the data point, either singularly, or aggregated with other events of the same type, within a persuasive message. ii) display one or more persuasive messages within the product web page to the Visitor to influence the visitor to purchase the product.

Use Case 2 : Order Confirmation Web Page Use Case The goal of the Retailer in this use case is to inform the Visitor that their order been completed successfully and no further steps are required by the Visitor. This is generally achieved by displaying an 'Order Confirmation Web Page' containing information such as an order identifier. The main mechanism of the system in this use case is: i ) record the order event for the purpose of displaying, either singularly, or aggregated with other events of the same type, within a persuasive message. Integration Model

The Real Time Message System integrates with the Retailer's web site through the use of JavaScript tag code, supplied by Taggstar to the Retailer. The JavaScript tag code is placed into Web Page content either directly, i.e. by the Retailer's Web Server including the tag in the HTML response body of a Web Page, or indirectly, by a Tag Management application (e.g. Google Tag Manager). The JavaScript tag code is customised by Taggstar for each Retailer by the inclusion of an unique identifier, the 'Retailer Web Site ID' that is transmitted to the Real Time Message System to enable identification of the Retailer's web site.

The JavaScript tag code uses the 'bootloader' pattern, i.e. it is a relatively small amount of code designed to load additional JavaScript depending on the execution environment, including, but not limited to :

(i) the type of web page (e.g. product page or order confirmation page)

(ii) the device (e.g. mobile device or desktop). The function of the JavaScript tag code is :

(i) to load additional JavaScript that then extracts information from its execution environment and sends this information to the Real Time Message System

(ii) in the case of the 'Visit Product Page Use Case' to display persuasive messaging, using HTML and CSS, to increase the probability of a purchase being made by a Visitor.

Additional functions of the JavaScript tag code include :

(i) managing visitor state information (including but not limited to, an experiment group, a list of product ID's viewed, a session ID) using cookies and or local storage

(ii) sending debug information about the web page content to the Real Time Message System

Use Cases Visit Product Page

1. A Product Web Page on a Retailer Web Site is requested by a Visitor using a web browser application. 2. The Product Web Page content is processed by the Visitor's web browser and the Taggstar JavaScript tag code is executed that a) downloads any additional JavaScript required and b) makes a request to the Real Time Message System for persuasive messages strings and c) displays in the web page any message strings returned by the Real Time Message System.

3. The Taggstar JavaScript tag code monitors the UI element that triggers an 'add to basket' action and the Visitor triggers this action, makes a request to the Real Time Message System to log this information along with the details of the Visitor. Order Confirmation Page

1. An Order Confirmation Page on a Retailer Web Site is requested by a Visitor using a web browser application.

2. The Order Confirmation Page content is processed by the Visitor's web browser and one or more Taggstar JavaScript files are requested and then executed that then extracts order and product information, such as identifier and price, from the web page and makes a request to the Real Time Message System to record the order event data along with the details of the Visitor (including but not limited to a Visitor ID, a session ID).

Real Time Message System

Concepts

The system continuously calculates 'audience' measures for products on a Retailer's web site. An audience measure is an integer value that is equal to the size of the set of sessions related to a product at a point in time. To create this relationship in the system a session ID is created as a composite ID, comprising a Visitor ID and a Product ID, allowing the set of sessions for a given product ID to be easily discovered. It is possible for the set of sessions to be an empty set.

Session has its normal meaning in the context of a software application, i.e. a means of identifying a time ordered series of events, where no event in the session is separated from any other event in the session by more than N units of time, where N is a duration referred to as the 'session expiry time'. A session is said to have expired when the time between the last event in the session and the current time is greater than the session expiry time. The audience measure used by the system excludes expired sessions.

The system creates and maintains two sets of sessions. One to calculate a 'current' audience measure and one to calculate a 'recent' audience measure. The current audience measure uses sessions with an expiry time of 20 minutes and the recent audience measure uses sessions with an expiry time of 2 hours.

In Visitor Use Case 1 : Creation and Update of an Audience Session

Trigger : A Visitor requests a given product page a) IF the Visitor [1] has not requested the product page with the last N minutes THEN i) Create a session timer [2], initialised to N minutes, and identified by the combination of the visitor ID and the product ID. ii) Increment the audience counter for the product by 1. b) ELSE IF the Visitor has requested the product page within the last N minutes THEN i) Lookup the session timer identified by the visitor ID and product ID

ii) If the session timer exists reset it to N minutes. END

[1] The Visitor is identified by a unique ID stored in the Visitor's web browser using cookies or local storage. [2] A session timer ID is a composite ID, comprising a Visitor ID and a Product ID. A session timer counts down from the value it is initialised to, or reset to, and upon reaching zero decrements by 1 the audience counter for the related product (identified by the product ID obtained from the timer's composite ID). After reaching zero the timer has no further use in the system and at some point is destroyed by the system application.

Quantity of Purchases

The system records purchases for Retailer products that occur during the Order Confirmation Page Use Case. The system calculates the quantity of purchases in the last N days for all products at frequent intervals as a background process. Message Rule Set

In the Visit Product Page Use Case in step 2 a request is made to the Real Time Message System and the response may include zero, or one or two persuasive messages strings that are the result of an execution of a set of rules that determine the most persuasive message to be displayed to a visitor.

The set of rules comprises of two tables, one for each category of message, audience and purchase. No more than one message per category is returned on the response. Rule trigger conditions are evaluated in ascending order by rule priority within a table. Rule trigger condition evaluation within a table stops when evaluation of a trigger condition returns true, then at that time, the persuasive message string corresponding to the trigger, is generated by replacing a placeholder in the message template, shown as { {variable} } , with the variable used in the trigger condition, and the resulting string is added to the response (e.g. for display in the Visitor web browser).

Pseudo Code product = retrieve product from database using ID included in the request made by the JavaScript tag. The product has associated with it variables current_audience, recent audience and quantity of purchases, as described in this document. Also last_purchase_date, that is the date of the last purchase of the product, and quantity purchased that is the quantity of purchases within the last N days. A trigger condition requires various 'threshold' parameters that are initialised before each rule set execution : time_now = current_time_secondsQ

current_audience_threshold = 2

current_audience_hot_threshold = 10

recent_audience_threshold = 5

recent_audience_hot_threshold = 20

last_purchased_seconds_threshold = time_now - 60 seconds

last_purchased_minute_threshold = time_now - 120 seconds

last_purchased_minutes_threshold = time_now - 1200 seconds

last_purchased_oldest_threshold = time_now - 7200 seconds

qty_purchases_threshold = 1 Audience Rule Table

See Figure 3 for example.

Purchase Rule Table

See Figure 4 for example. Advert System Introduction

The goal of the Advert System is to increase the Click Through Rate (CTR) of Retailer adverts by providing to an Ad Platform persuasive messages and data generated by the Real Time Message System.

The function of the Advert System is act as interface between the Real Time Message Server and an Ad Platform.

Ad Platform Integration In this section current audience, recent audience and quantity of purchases have the meanings already provided in this document. Batch Integration

In this integration mode the Advert System (for example, the Taggstar server of Figure 13) regularly connects to an Ad Platform API (for example, hosted in the advertiser server of Figure 13) and updates the content of pre-configured Advert creatives, for a specified Retailer product, with data points including current audience, recent audience and quantity of purchases. Updates may occur regularly or be triggered by changes in the product data, for example, by current audience increasing by X %, or by a change in quantity of purchase rate for a product. The data points are incorporated within natural language sentences or graphic images by the Ad Platform (for example, hosted in the advertiser server of Figure 13), into the pre- configured advert units created by the Retailer (for example, whose website is hosted on the first website server of Figure 13), and served by the Ad Platform to Internet users (for example, users of user terminals shown in Figure 13) who meet the advert targeting criteria specified by the Retailer.

A pre-configured advert must contain meta data specifying a product ID from the Retailer web site. The Advert System uses the Advert meta data to match an advert with a Retailer product. This also allows the Advert System to obtain from the Real Time Message System data points for a specific product. Once the data points are obtained for a product, the Ad Platform API is used to update the pre-configured Advert for the product.

An example system implementing the invention is shown in Figure 13. In Figure 13, fixed user terminals 1 to n are in connection with the internet, and mobile user terminals 1 to m are in connection with the internet, such as via cellular networks or via wifi networks. In Figure 13, a first website server, an advertiser server and a Taggstar server are in connection with the internet. Real Time Integration

In this integration mode the Ad Platform connects to the Advert System when an Internet user who meets the targeting criteria for the advert created by the Retailer, and the Advert System returns data points including current audience, recent audience and quantity of purchases, that are then incorporated as text or graphic images into the advert unit served to the Internet user.

In an API made by the Ad Platform to the Advert System, a product ID must be specified to enable the data points to be obtained from the Advert System and returned to the Ad Platform.

Taggstar: sell more Our goal is to help you inspire your customers, convert better and sell more. 1. Persuasive messaging

Increasing shopping conversion. Creating awareness, urgency and trust with persuasive messaging can increase conversion by up to 10%.

How it can look on product pages

Balloon overlays fade in then out, informing the shopper without intruding. In-page, real time messaging is provided. See Figure 5 for example, displayed on a fixed computer.

Listing pages

Messaging in listings pages helps drive discovery and call to action. See Figure 6 for example, displayed on a fixed computer.

Driving conversion is provided in mobile devices too. See Figure 7 for example, displayed on a mobile phone. Mobile data banner fades in then out. In-page messaging is provided. See Figure 8 for example, displayed on a mobile phone.

What are the benefits?

1. Increase sales conversion by up to 10%.

2. Product validation

3. Trust Testing and reporting

1. We do our own A/B split testing for you.

2. Reports includes traffic, product popularity, purchase rates, conversion uplift, messaging coverage.

2. Recommendations

Increasing basket value

• Frequently bought together

· Customers who bought this also bought. . .

• Customers who looked at this bought. . .

• Highly rated items

What are the benefits?

Increase basket value by 2-3x

• Higher basket value.

• Longer site time, more data.

• Deeper customer "buy-in".

3. Exit traffic monetisation On average, 2% of shoppers will make a purchase, which means 98% of them will leave empty-handed. (That's a lot of money walking out the door.) Taggstar can monetise the 98%. Influencing the non-buyers

-98% of traffic leaves without making a purchase. Most go to other retailers, or back to Google to do another search. Influence the non-buyers and sell the qualified leads to those willing to pay for it.

What about lifetime value?

The data shows that browsers are sawy, promiscuous and already know where else to look. If they're leaving your site to shop elsewhere anyway, why not influence and monetise that journey?

How?

1. Taggstar's customer "scoring" algorithm identifies:

· Visitors to your site who won't make a purchase.

• When they'll leave.

2. Our product recommendation engine suggests third party paid products to those most likely to leave your site empty handed, just before they leave.

3. When visitors do leave your site empty-handed, you are paid per click.

Using traditional cross-sell slots

Traditional product pages include "Customers also viewed" modules.

"More from our partners .

Most shoppers become low conversion prospects, and they are close to leaving your site. At this point, Taggstar dynamically replaces some or all of your own product recommendations and suggests related products from 3rd party retailers instead. See box in Figure 9 for example, which may include: "You may also like"

"More from the web"

"Ideas from our partners"

Technical implementation

· All products are powered by pasting one snippet of JavaScript into your site HTML. • Robust, scalable, cloud-based platform (Amazon EC2).

In an example, a user of a terminal, such as a fixed terminal (e.g. fixed user terminal 1 of Figure 12) or a mobile terminal (e.g. mobile user terminal 1 of Figure 12), views items for purchase on website A (eg. on website server for site A of Figure 12). The process flow may be as follows. i. Browser (for example on a user terminal of Figure 12) goes to website A (for example on website server for site A of Figure 12) and provides views of products for sale.

ii. Website A management platform calls the Taggstar data API (for example on Taggstar server of Figure 12), sending the recent activity record of the user.

iii. Taggstar uses that activity record to determine that the user is likely to leave website A soon.

iv. the Taggstar product recommendation engine generates third party paid product recommendations suitable for the user who is likely to leave website A soon.

v. The Taggstar server sends the generated third party paid product recommendations suitable for the user who is likely to leave website A soon to the website A management platform, including links to third party websites (eg. website server for site B of Figure 12) which sell the recommended third party paid products. vi. Site A sends to the browser the received third party paid product recommendations suitable for the user who is likely to leave website A soon, including links to third party websites which sell the recommended third party paid products. An example system implementing the invention is shown in Figure 12. In Figure 12, fixed user terminals 1 to n are in connection with the internet, and mobile user terminals 1 to m are in connection with the internet, such as via cellular networks or via wifi networks. In Figure 12, website server for site A, website server for site B and the Taggstar server are in connection with the internet.

Why us?

Fast: Get going quickly and easily.

Simple: No fuss. Quick and easy to implement.

Effective: Sophisticated technology.

Affordable: No setup costs. Flexible pricing based on scale. FAQs

Q. Do I need a lot of traffic for this to be effective?

No, you don't. Taggstar's intelligent platform decides what kind of message works best for a product, depending on volume and frequency of views, purchases and we can provide recommendations as well. There are many ways to construct persuasive messaging, even around lower volume products.

Q. What happens if a product isn't being viewed at the time?

It is unusual that a product has no history whatsoever, but if need be, we will show no messaging at all. Our goal is to add value wherever possible. For example, if no-one else is looking at a product, then we might instead inform the shopper that "this product has been purchased 5 times today."

Q. Will this affect my site's load time?

No. All scripts and assets are served remotely from our cloud-based servers. Additionally, our JavaScript is tiny and loads asynchronously, meaning it loads last.

Q. Who controls the messaging and tone of voice?

You do. We will work with you to determine the best tone of voice for your brand, which can be changed at any time.

Q. Can I pick and choose from the various features and products?

Yes, absolutely. All our products are modular, meaning you only use what you feel will work best for your business. ExampleCompany Test Plan

Key objectives

Taggstar aims to drive higher customer engagement and shopping conversion uplift by displaying real time messaging about products directly to shoppers on your site.

Test Sites

Firstwebsite.com (web and mobile web)

Secondwebsite.com (web and mobile web)

Installation

Taggstar's JavaScript has already been supplied and installed by ExampleCompany. Our preference is to be installed as broadly as possible, to allow for a large data set.

Please confirm in which categories the JS has been installed.

A/B traffic split

We recommend splitting the traffic randomly, as follows:

Control: 10%

Treatment: 90%

Taggstar will track data independently using your A/B split.

1. It is critical that the two experiment groups are named "A" (Control) or "B" (Treatment) in order that Taggstar can interpret the data correctly. A JavaScript global variable called 'taggUserExperimentGroup' should be created in each page containing either 'control' or 'treatment' string values and this variable must be present on every page impression where Taggstar messaging could be displayed.

2. It is critical that a visitor be mapped to an experiment group using either a long-lived cookie or browser local storage so the group assignment is persistent across sessions.

Real Time Shopper Data

For each product, ideally we would want to see at least 1,000 impressions before we could deliver statistically significant results. Achieving 1,000 impressions will take varying amount of time depending on the product and we will want results for at least 30% of your products in order to start drawing any conclusions.

Key test criteria

We will look for results in the following key performance indicators (KPIs):

1. Add to basket CTR uplift. Broken down by category and by product.

2. Add to wish list CTR uplift. 3. Product page bounce rates.

4. Products pages viewed per session.

5. Time spent per page per session.

6. Time to return (how long before a customer returns to re -visit a product).

Key variables

There are a number of variables we will track that can influence the results. We will want to understand if persuasive messaging is more/less effective based on variables such as:

1. Time of day/ day of week

2. Location of purchaser

3. Product type (if ExampleCompany has it's own internal product category breakdown, please provide).

Reporting

Taggstar will gather data points and report weekly on a number of different queries, including:

1. Product views by product/ category.

2. Rates of purchase by product/ category

3. Filter by message type/ style. "Hot or not".

4. Time/ day.

5. Location of shoppers.

6. Product type.

Questions for ExampleCompany

1. Do you want our log files so you can break that down against your own internal data? e.g. demographic

2. How many products do you have?

3. Please provide your internal product category breakdown.

Project Hotcake

Project Hotcake is a business to business (B2B) product that creates a sense of shopper urgency on e-commerce sites, thereby driving greater engagement and shopping conversion.

"What if they run out of stock?

"This one seems to be selling fast" "Wow. . .there are five other people looking at this right now."

Urgency is generated by providing real-time information about a product to the shopper, such as how many other people are looking at the same product right now, how many purchases have been made recently, and how many are left.

Key data points

1. Audience— how many people are looking right now

2. Purchase history— how recently was this product last purchased (+ variants)

3. Inventory— how many are left

5. Location— "this product was last purchased by someone in Knightsbridge 23 minutes ago."

Number of people looking at product right now/last hour/last day

1. Figure out what product is

2. Decide when they started looking vs when they finished. Did it time out.

3. Keep track of page entry vs exit. Assume timeframe for active viewer (e.g. if page open for more than 10 mins, are they actually viewing?)

Last time product was purchased How it may appear:

1. Text integrated into html via API

2. Overlay somewhere on page that appears and then disappears (e.g. fade in/ out bubble or slideout) via JavaScript.

Where it may appear:

1. Product page 2. Search results page— this would most likely need to fade in to a defined area as we would not be able to deliver the results as quickly as the search results (we're starting our search when the initial search is finished).

3. Inform the shopper the time the product was last purchased, and where the purchaser was (e.g. London, Birmingham). Alternatives should include: a. "Last purchased at 10:15am by someone in London, UK"

b. "10 people have purchased this in the last hour." Need to be able to set some simple rules that identify if a product is selling quickly. Ie if a product has sold 10 times in the last few hours, we might want to adjust the messaging to say "Wow. . .this one's going fast. 10 sold in the last 2 hours."

How would we do this?

a. Need a javascript (JS) tracking pixel on the confirmation page and we point to certain bits of information. We would prefer this as it's less work for the retailer, but there could be a security issue if the page is secure.

b. Alternatively, they would provide us with the information via a request on tracking pixel. They create the tacking pixel but it points to us. This means a little more work for the retailer, but no security issue.

c. Or polling or a reporting API if they have it available, but options A and B are simplest/best.

Inventory (ie number of products remaining)

1. Requires real-time inventory information that we display differently.

2. No way for us to generate that info if they don't already have it.

Languages

Messaging may be delivered in the native language of the site on which we are running. It does not need to adapt to the language of where the shopper is located.

Channels 1. Web

2. Mobile/tablet

3. Native mobile e-commerce apps

Reporting

Clients may require monthly reporting that shows the following daily data: 1. Total page impressions.

2. Total page impressions where we display a piece of data. No data displayed, no impression.

3. "Buy" button clicks

4. Products listed by number of purchases.

Implementation

1. Install JS in all relevant areas of site.

2. Install tracking JS on confirmation page (unless client provides info with their own tracking pixel).

Cost criteria

1. Number of page impressions

2. Number of visits

3. Number of uniques

4. Number of products

Self sign up ecommerce businesses could create an account, self sign up, generate the JS, customize the UI with colour and font, enter payment details and then install.

Other ideas Connect to facebook for which of your friends has looked at this, or other items from that same retailer.

Most-viewed in-page module— can be used on home pages to showcase the products that most people are looking at right now. "Hot right now"

Recommendations/Personalisation— if a shopper looks at a particular hotel in New York, show them that same hotel the next time they visit, perhaps with a voucher code for a discount.

Third party product links - CPC

Retargeting data— if a shopper looks at a hotel in New York, retarget them later on a third party site (via Criteo etc) saying how many people have booked that hotel in the last 24 hours (for example)

Product heat maps for internal display

Cloud widget

Geographical map for travel destination searches Messaging— "scoring urgency" Availability

This is contingent upon retailer providing this info already.

There's only 1 left

There are 10 left

If there are lots remaining, do we show nothing, or do we say, loads left.

There are none left, but more coming next week/ month etc or due date.

Purchasing 24 hours is our definition of real-time.

1 purchase in the last 10 minutes

5 purchased in the last hour

If zero purchases in last 24 hours, then no messaging

Start messaging at: If 1 purchase in last 24 hours, we say "last purchased at" How do we calculate velocity?

Number in stock / rate of purchase means we could estimate when it could go out of stock eg going fast at this rate, they'll be gone by Thursday.

Audience

No-one's looking, but if 10 people have looked in the last hour, then say that.

We can define what right now means eg could be a 10 minute window.

If one person's looking then say "1 person's looking"

Customer Stories

As a customer I want to display real time product information messages in product pages because I believe it will result in an uplift of add to basket and add to wish list conversion.

As a customer I want to do A/B testing so I can be sure I am measuring % uplift caused by the product and compared to not using the product.

As a customer I want to be able to turn the product on or off quickly in case any problems occur with the messaging being displayed on my web site.

As a customer I want to get reports on % uplift made available to me at regular intervals so I can monitor the effectiveness of the product.

Key Features V0.1 1. Product Messaging

- Display real time messaging on customer's product pages

- display audience message in balloon

- display product purchased message in balloon

- display last purchased message in balloon

- A/B testing - - split product page traffic 80% control group, 20% treatment group. - customer assigns experiment groups via Javascript variable.

- Display insert -

- Display real time messaging on search result pages

- Display low stock message

2. Reporting & Analytics

- Provide web page analytics in CSV format daily

- Provide A/B test uplift Analytics in CSV format daily

i) provide uplift report with % uplift and % confidence level on checkout conversions

- Customer dashboard

Promised to ExampleCompany

Taggstar will gather the data points and report weekly based on a number of different queries, including:

1. Product views by product/ category.

2. Rates of purchase by product/ category

3. Filter by message type/ style. "Hot or not".

4. Time/ day

5. Location of shoppers

6. Product type. KPI's promised to ExampleComapny

1. Add to basket CTR uplift. Broken down by category and by product.

2. Add to wish list CTR uplift.

3. Product page bounce rates.

4. Products pages viewed per session.

5. Time spent per page per session.

6. Time to return (how long before a customer returns to re -visit a product).

Message Types

1. Balloon message Types a) current audience (with hot style)

b) recent audience

c) last purchased

d) recently purchased (with hot style)

No tag needed.

2. In page message

Displays on a product page. Requires a tag like this - <div data- taggstar-product= "XXX" ></div> Note: you should replace XXX with your product SKU or product ID. A/B testing

The customer may wish to split traffic into control and treatment experiment groups. Information provided to ExampleComapny -

1. It is critical that the two experiment groups are named "A" (Control) or "B" (Treatment) in order that Taggstar can interpret the data correctly. A Javascript global variable called 'taggUserExperimentGroup' must be created in each page containing either 'control' or 'treatment' string values and this variable must be present on every page impression where Taggstar messaging could be displayed. Message Display Rules

Configuration Variables current_audience_count = unique visitor count for a product within last 20 minutes - 1 (-1 to avoid counting the user currently viewing the page). The 20 minute window should be a global variable.

recent_audience_count = unique visitor count for a product within last 2 hours - 1. The 2 hours window should be a global variable.

current_audience_hot_threshold - when unique visitor count for a product within last 20 minutes is over this threshold the hot style is added to an audience message. This variable will be per site.

audience_display_threshold - when current_audience_count is > 1 then display the audience message.

purchases_display_threshold - if number of purchases with N hours is less then this threshold then the purchase message is not displayed (as showing low numbers is not a persuasive message). The purchase_display_threshold is hard coded as '3' in version 0.1 but will be per site in later versions. N, the number of hours, will be a per site variable. purchases_count - number of checkouts for a product within N hours. This will be site specific.

purchases_hot_threshold - when purchases_count is over this number then the purchase message is displayed with the hot style. This variable will be per site and hardcoded in version 0.1, and could be dynamically set in later versions.

Audience Message Rule if current_audience_count is > audience_display_threshold then

display current audience message -

E.g. "2 other people are looking at this product right now"

If current_audience_count > current_audience_hot_threshold then

add 'hot' style to message

else if recent_audience_count > audience_display_threshold then

display recent audience message - E.g. "2 people have viewed this product in the last N hours"

Purchase Message Rule

If purchases_count > purchases_display_threshold then

display "X people have purchased in last 24 hours" message according to - If purchases_count > purchases_hot_threshold add hot style to message else

display last purchase message according to - If product purchased within last 60 seconds then

display "product last purchased X seconds ago"

else if product purchased within last hour then

display "product last purchased in X minutes ago"

else if product purchased over one hour ago then [1]

display "product last purchase at [time]"

[1] ExampleComapny do not want to display a last purchase message when the purchase was made over one hour ago. Review Message Rules

ExampleComapny may make this message be displayed as follows - In preference to the audience message on secondwebsite.com where review data available and display rule evaluates to true

In preference to the last purchase mesage on firstwebsite.com where review data available and display rule evaluates to true

Display Rule Set for Firstwebsite.com & Secondwebsite.com execute audience message rule (seconds and minutes only)

If % of firstwebsite.com reviewers that would recommend > = 90%

(or for secondwebsite.com If % 5 star reviews >= 80% and total number reviews > 1) then

display review message

else

execute purchase message rule

ExampleComapny General Information 1. Product messaging for reviewed items are displayed on firstwebsite.com and secondwebsite.com.

Example - http:/ /www.firstwebsite. com.com/ spear-and-jackson-elements-dutch- hoe/1042190685.prd?cmtag=o&_requestid=6688&prdToken=/p/prodl 5990983- sku26901387-CL&browseToken=/q/hoes&totalResults=l

The review message on firstwebsite.com is X% of reviewers recommended this item

The review message on secondwebsite.com is X% of reviewers rated this item Y stars

2. The 0.1 messaging must include a review message type delivered by taggstar a) Scraping review data -

Reviews HTML have different syntax on secondwebsite.com vs firstwebsite.com as the reviews HTML generated by different providers. Secondwebsite.com has the review syntax (powered by pluck) that will be used throughout ExampleComapny sites in the future.

Firstwebsite.com has Old' review system. b) When to display reviews

Waiting on rule site from customer. c) Logging

We need to add another message type name.

3. Style of messaging Messaging UI style should match the existing style, as shown on the firstwebsite.com pages.

4. We need to pull the experiment group from qubit from their pages Waiting on detail from ExampleComapny.

5. Rules on when to show last purchased They suggested only display last purchased when time window is seconds and minutes ago. i.e. less than 1 hour.

6. Two messages per page maximum.

Note

It is to be understood that the above-referenced arrangements are only illustrative of the application for the principles of the present invention. Numerous modifications and alternative arrangements can be devised without departing from the spirit and scope of the present invention. While the present invention has been shown in the drawings and fully described above with particularity and detail in connection with what is presently deemed to be the most practical and preferred example(s) of the invention, it will be apparent to those of ordinary skill in the art that numerous modifications can be made without departing from the principles and concepts of the invention as set forth herein.

Claims

1. A method of supplying an optimal message to add to a web page from a website, in which: an analysis server receives a request for an optimal message to add to a web age including a product, the request identifying the product; the analysis server fetches data from a database, the data relating to web server data in relation to the identified product; the analysis server analyses the fetched data so as to determine the optimal message to add to the web page including the product, and the analysis server sends the determined optimal message, in response to the request.

2. Method of Claim 1, in which the request identifies the product and a user, and the data relates to web server data in relation to the identified product and the user.

3. Method of any previous Claim, in which the web server data is real time data.

4. Method of any previous Claim, in which the analysis server receives the request from a first web server, the web server data is first web server data, and the analysis server sends the determined optimal message to the first web server, in response to the request.

5. Method of Claim 4, in which first web server data in relation to the identified product comprises web traffic data on the first web server in relation to the identified product and/ or current and future product price and availability data.

6. Method of Claims 4 or 5, including the step of the analysis server receiving web server data in relation to products from the first web server, and saving the web server data in the database.

7. Method of Claim 6, wherein types of data being collected include one or more of: purchase, audience, inventory, review, "Add to Basket" rates, purchase rates and product price.

8. Method of Claims 6 or 7, in which the analysis server also tracks all user interactions, thus enabling automated content recommendations based on shopper behavior and product interests.

9. Method of any of Claims 4 to 8, in which in response to receiving the message, the first web server decides whether to bid for a paid search result, and if the bid is made and won, the first web server sends the paid search result to a search engine webserver, including the message.

10. Method of any of Claims 1 to 3, in which the analysis server receives the request from an advertising server, and the analysis server sends the determined optimal message to the advertising web server, in response to the request.

11. Method of Claim 10, in which the web server data in relation to the identified product comprises web traffic data on a plurality of web servers in relation to the identified product and/ or current and future product price and availability data.

12. Method of Claims 10 or 11, further including the step of the analysis server receiving web server data in relation to products and users, from a plurality of web servers hosting a plurality of web sites, and saving the web server data in the database.

13. Method of any of Claims 10 to 12, in which the advertising server cookies a user.

14. Method of any of Claims 10 to 13 in which the advertising server takes the text message as supplied by the analysis server and inserts it into their ad creative before the completed creative is finalized and presented to an individual shopper in their browser.

15. Method of any of Claims 10 to 14 in which the analysis server passes relevant product and audience data to the advertising server in real time.

16. Method of any of Claims 10 to 15 in which the advertising server uses RTB exchanges and uses information as to whether or not an ad contains an analysis server message to determine how much they are willing to bid for a slot.

17. Method of any of Claims 10 to 16, in which analysis server data messaging can be included in the creative to help increase CTR and therefore the return on investment (ROI) value to the advertiser of the ad.

18. Method of any of Claims 1 to 3, in which the analysis server receives the request from a web browser running on a user terminal for an optimal message to add to the web page including the product, and the analysis server sends the determined optimal message to the web browser, in response to the request.

19. Method of Claim 18, in which the web server data in relation to the identified product comprises web traffic data on a plurality of web servers in relation to the identified product and/ or current and future product price and availability data.

20. Method of Claims 18 or 19, further including the step of the analysis server receiving web server data in relation to products and users, from a plurality of web servers hosting a plurality of web sites, and saving the web server data in the database.

21. Method of any previous Claim, in which the analysis server uses social proof to generate two distinct types of consumer sentiment: Urgency and Positive Validation; these types of messages can co-exist and appear on a page at the same time.

22. Method of any previous Claim, in which the determined optimal message includes a standard width UI message, and/or a narrow width UI message.

23. Method of any previous Claim, in which the determined optimal message includes text and/ or graphics.

24. Method of any previous Claim, in which the analysis server includes an algorithm that identifies increases in the rate of sale of products.

25. Method of any previous Claim, in which the web server provides messages in page.

26. Method of any previous Claim, in which the web server provides messages as informational balloons that can fade in/ out, or slide in/ out.

27. Method of any previous Claim, in which a position of the message on the web page is varied by the analysis server based on performance.

28. Method of any previous Claim, in which the web server displays the message at one or more of the following stages: home page, Search results /gallery page, Product page, Basket section.

29. Method of any previous Claim, in which the message is applied in display advertising.

30. Method of any previous Claim, in which the message is applied in search results: in natural or paid search.

31. Method of any previous Claim, in which if the web server is able to map the visitor back to an email address, then that web server can follow up with an email to the individual that contains personalized content.

32. Method of any previous Claim, in which the method is used in personalized television advertising.

33. Method of any previous Claim, in which code is installed in a website, to collect website data.

34. Method of Claim 33, in which the code is JavaScript code.

35. Method of any previous Claim, in which the optimal message includes one or more of "how many others are looking at this product", "when was the last one purchased", "how many have been booked today" and "what do others think of this".

36. Method of any previous Claim, in which the analysis server includes machine- learning algorithms that dynamically test and optimise performance.

37. Method of Claim 36, in which the machine-learning algorithms vary one or more of: data thresholds, message combinations, message tone, message design or colour, the number of messages, duration of a balloon on a web page, position of message on page.

38. An analysis server, the analysis server configured to supply an optimal message to add to a web page from a website, in which: the analysis server is configured to receive a request for an optimal message to add to a web page including a product, the request identifying the product; the analysis server is configured to fetch data from a database, the data relating to web server data in relation to the identified product; the analysis server is configured to analyse the fetched data so as to determine the optimal message to add to the web page including the product, and the analysis server is configured to send the determined optimal message, in response to the request.

39. Analysis server of Claim 38, configured to perform the method of any of Claims 1 to 37.

40. Computer program product, arranged such that when running on an analysis server, the computer program product is configured to supply an optimal message to add to a web page from a website, the computer program product arranged to: receive a request for an optimal message to add to a web page including a product, the request identifying the product; fetch data from a database, the data relating to web server data in relation to the identified product; analyse the fetched data so as to determine the optimal message to add to the web page including the product, and send the determined optimal message, in response to the request.

41. Computer program product of Claim 40, configured to perform the method of any of Claims 1 to 37.

42. System for supplying an optimal message to add to a web page from a website, the system comprising an analysis server and a first web server in connection with the analysis server, in which: the analysis server is configured to receive from the first web server a request for an optimal message to add to a web page including a product, the request identifying the product; the analysis server is configured to fetch data from a database, the data relating to first web server data in relation to the identified product; the analysis server is configured to analyse the fetched data so as to determine the optimal message to add to the web page including the product, and the analysis server is configured to send the determined optimal message to the first web server, in response to the request.

43. System for supplying an optimal message to add to a web page from a website, the system comprising an analysis server and an advertising server in connection with the analysis server, in which: the analysis server is configured to receive from the advertising server a request for an optimal message to add to a web page including a product, the request identifying the product; the analysis server is configured to fetch data from a database, the data relating to web server data in relation to the identified product; the analysis server is configured to analyse the fetched data so as to determine the optimal message to add to the web page including the product, and the analysis server is configured to send the determined optimal message to the advertising server, in response to the request.

44. System for supplying an optimal message to add to a web page from a website, the system comprising an analysis server and a user terminal running a web browser, the user terminal in connection with the analysis server, in which: the analysis server is configured to receive from the web browser a request for an optimal message to add to a web page including a product, the request identifying the product; the analysis server is configured to fetch data from a database, the data relating to web server data in relation to the identified product; the analysis server is configured to analyse the fetched data so as to determine the optimal message to add to the web page including the product, and the analysis server is configured to send the determined optimal message to the web browser, in response to the request.

45. A method of supplying an optimal message to add to a web page, comprising the steps of:

(i) an analysis server receiving a request from a web browser running on a user terminal for an optimal message to add to a web page including a product, the request identifying the user; (ii) the analysis server fetching data from a database, the data relating to web server data in relation to the identified product or the user;

(iii) the analysis server analysing the fetched data so as to determine the optimal message to add to the web page including the product and being viewed by the user, and (iv) the analysis server sending the determined optimal message to the web browser, in response to the request.

46. Method of Claim 45, in which the web server data is real time data.

47. Method of Claims 45 or 46, in which the web server data in relation to the identified product or user comprises web traffic data on a plurality of web servers in relation to the identified product or user and/or current and future product price and availability data.

48. Method of any of Claims 45 to 47, further including the step of the analysis server receiving web site traffic data in relation to products and users, from a plurality of web servers hosting a plurality of web sites, and saving the web site traffic data in the database.

49. Analysis server configured to supply an optimal message to add to a web page, the analysis server configured to:

50. Analysis server of Claim 49, further configured to perform the method of any of Claims 45 to 48.

51. Computer program product executable on an analysis server, the computer program product when running on the analysis server arranged to supply an optimal message to add to a web page, the computer program product configured to:

52. Computer program product of Claim 51, further configured to perform the method of any of Claims 45 to 48.

53. Method of supplying product recommendations and associated web links to add to a web page from a website, comprising the steps of:

(iv) the analysis server sending the set of product recommendations of third party products, together with the web links to the third party websites which supply the third party products, to the website server, for inclusion in the web page from the web site.

54. Method of supplying product recommendations and associated web links for inclusion in a web page from a website, comprising the steps of: (i) an analysis server receiving from a web browser running on a user terminal a recent activity record of the user who has been accessing a website, and an identification of the user;

(iv) the analysis server sending the set of product recommendations of third party products, together with the web links to the third party websites which supply the third party products, to the web browser, for inclusion in the web page from the web site.

55. A method of supplying data suitable for including in a web page including a product, comprising the steps of:

(i) the analysis server fetching data from a database, the data relating to web traffic in relation to the product;

(iii) the analysis server analysing the fetched data so as to determine data suitable for adding to the web page including the product, and

56. Method of Claim 55, further comprising the steps of:

(vi) the advertising web server including the created message in the web page.