Location Based Online Ad Targeting: Prioritize the Best Target - Suppress the Worst

Continuing the theme of the last couple of months, we will be covering how Semcasting Audience Targeting is different from the traditional online advertising of contextual or behavioral targeting methods that are current being used. We will also show how Semcasting Audience Targeting can expand and prioritize your audience for your next online campaign so you no longer are restricted to the data exchange cookie pools available today. For details on Semcasting Audience Targeting Platform now click here .

Last month we introduced the first step of the Audience Targeting process – the Analytic Targeting Model. The targeted model statistically compares current customers to the general population and identifies what makes your customers are unique. This uniqueness is based on what we know about the customer in terms of affluence, household composition, age, and consumer purchasing and ownership behavior. Using an automated predictive modeling engine, we can statistically weigh the impact of over 750 demographic variables and how they contribute in identifying the ideal audience. All of the data we use in this process is publicly available and follows all of the guidelines and requirements proposed in the pending "Do Not Track" legislation.

In Step Two we will use the model scores to locate the best target locations for your prospective customers based on the analytic model scores and an optimization process of online consumers, based on reach & frequency and specific user types.

Step Two: Detecting the Best Target Location

Using the scoring formula from the analytic model, the geographic population of all non-customers is scored. The Semcasting Data file, which is made up of over 225 million customers and over 125 million households, is used to determine where those who most look most like current customers live. The weighted impact of 750 data variables including demographic, psychographic and behavioral variables, combine to define the online propensities of the best prospects including social network participation, online user type and interest areas in key contextual categories.

The output is a list of individuals which is measured and ranked in order of percentage of lift - the likelihood to transact based on the model. The target audiences are summarized and optimized for reach and frequency across over nearly 40,000 zip codes. Each zip code location is subdivided into unique User Types, or MicroTargeted Audiences (Education, Medical, Marketing/Media, Home, Wireless, etc.). MicroTargeting provides over 15 times improvement to the analytic model and focused the optimal campaign audience on the best possible locations. The best populations are prioritized for bidding and delivery, and the worst are used to suppress from the campaign.

Next Month we cover the specifics of Audience MicroTargeting and how you can make it work for you.

Highlighted Database Variable of the Month

With each Newsletter, Semcasting introduces a data variable that has proven to be highly predictive for our customers. These variables set us apart from our peers because of the impact they have demonstrated in improving the effectiveness of models.  These variables can now be used in planning online display and social networking campaigns, just as they have been used for direct mail and email campaigns. The important thing to remember is that we built these variables based on people who have voted with their wallets for a particular product or service. They are not simply associated with a cookie or referred to by a visit to a web site. The analytic model built from these consumers are correlated and tested against known transactions over many, many campaigns. They leverage all of our findings to provide you with 100% reach to the entire potential online audience without using cookies.

Vehicle Consideration Set

In June of 2000, Congress passed the Shelby Act, prohibiting the use of vehicle ownership information- including the use of make, model and year data in direct marketing solicitations. The only permitted marketing use of vehicle ownership data is in the development of analytic models. Predictive models can be used to build marketing lists because they safely and legally reflect ownership intent.

Semcasting has designed a solution for marketers based on the belief that many, if not most, households who are in the market for a new vehicle consider more than one vehicle or brand before they buy. The vehicles they will consider generally fall into "consideration sets" defined by body style and price. Semcasting Auto Consideration Set Data is built on auto ownership information in the household over a ten year period. The results of our automated modeling process are 11 different predictive models and a rating of 1 through 5 on their likelihood to own or purchase an auto for each consideration set. Incorporated into each of the models are vehicle values, life stage, home ownership, family composition, age, income, discretionary income and other household demographics and affluence factors. Also entered into the models are distance to dealers and market share for each make model, helping to influence the model and set the trade area dynamics for a campaign.

Auto manufacturers and dealers can create campaigns based on consideration sets to identify their best prospects for targeted online advertising, postal and e-mail campaigns.