Lateral Targeting
Lateral targeting is a fundamentally new approach to the task of focusing advertising on consumers. The method is based on modelling consumer habits and preferences using statistical analysis of Internet traffic and recognition of individual behaviourial patterns.
The method is called ‘lateral’ after a key attribute of ‘lateral thinking’ by Edward De Bono.
• “To create flexible rather than pre-defined categories in order to achieve the desired results.”
Opportunities
Lateral targeting enables the concentration of marketing efforts on the most likely customers, irrespective of their demographics, boosts the value of second-tier web pages and generates new value from statistical data. It can be implemented as a standalone targeting service or as an upgrade to existing ad serving software.
Advantages
The traditional aproach to media campaign planning usually involves selection of one or several predefined socio-domographic groups. These are identified as the target audience and media selection is governed by relevance to this audience.
Internet advertising can achieve significantly better results through systematic targeting – but even the most precisely targeted ad campaigns miss a large proportion of the consumer opportunity, instead hitting people who belong to the target audience but are not actual consumers. Here we see efficiency lost twice.
Lateral targeting abandons all socio-demographic labels and delivers advertising to the most probable consumers regardless of sex, age, education, income, etc because we are able to calculate, with a high degree of probability, whether or not each Internet user is a potential consumer of the advertised product or service.
The Method
When people buy something they manifest aspects of their behaviour and, when they surf the web, they also generate measurable behavioural information. Given two manifestations of the behaviour of the same person, is there a chance that they are interrelated? Can we predict the probability of a purchase via analysis of web surfing ? Can we influence this probability?
The answers are yes.
A combination of traditional sampling and polling techniques allied to the sophisticated processing of hard data from user logs has led to algorithms that can do amazing things.
Our calculations require statistical data on each user's traffic, also information about related behaviour as a consumer. Traffic data are available to the banner networks, Internet Service Providers, Internet statistics systems and major online publishers. Data on consumer habits can be derived from off-line research, online surveys and from purchases made in online shops or at auctions.
Known consumers generating sufficient traffic form Reference Groups. We can extrapolate reference group behaviour to complete populations. Our mathematical method shows consistent results based on reference groups of 150 - 200 people with no significant improvement from groups of more than 500. The minimum number of unique pages required per person is 100 and optimum results are achieved when this number approaches 2000.
Users
Our product is being developed with the following types of business in mind:
• Internet Service Providers
• Internet-statistics Systems
• Internet advertisers (agencies and clients deploying banners and other Internet ad forms)
• Publishers
Implementation – Stage One
To demonstrate how Lateral Targeting can improve the efficiency of Internet advertising, we have implemented one possible approach that is aimed at increasing the ad click-through rate (CTR targeting)
We consider clicking on banners to be a form of behaviour. The first stage of campaign development is data collection from users who have clicked a banner and therefore form a reference group. The second stage is to calculate the audience potential of the given banner and deliver ads focusing on that audience.
Current Results
To date we have completed the development of a mathematical method that implements Lateral Targeting. The method is based on mathematical statistics and probability theory. The method works with all types of websites and is not based on thematic analysis of content.
The method has been alpha-tested using MediaArts Group proxy server data, TNS-Gallup Web-Index and ad-hoc research and surveys. Tests show reliability factors of 85% or better for sex-age groups, MBTI-based psychological types and even smokers and non-smokers.
We have developed a core targeting system which, in addition to CTR-targeting, implements behavioural targeting based on web page context analysis.
Context-based targeting complements CTR-targeting in cases when we have insufficient data for the user or we cannot make a precise identification (in case of blocked cookies). Context analysis alghorithms are also part of our in-house development. The core system is developed using .NET and MS SQL and is optimised for real-time performance.
SOURCE : read write web
POSTED ON CONTAGIOUS IDEAS : 2.0 strategic planning
BY : jérémy dumont
THE BEST OF OUR BLOGS PSST.FR
AN INITIATIVE OF POURQUOI TU COURS ?
2.0 STRATEGIC PLANNING AGENCY
jérémy dumont DIRECTOR OF STRATEGIC PLANNING
