Audience Targeting by advanced store: The Right Combination of Common Sense and Love of Analytics

Our audience targeting is precisely aligned to the requirements of performance campaigns. Unlike branding campaigns, achieving a specific demographics is not the goal of these campaigns. Response or conversion is what counts. Therefore only methods that have a lasting effect on the response will do.

Between Intent and Interest - Finding the Right Users

We search for events based on surfing behaviour that enable us to assign a general product interest, a specific purchase intention or a demographic characteristic to the user. Interest or demographics follow the assumption that users with certain characteristics are more responsive to advertising. Hence if we advertise women's cosmetics the obvious assumption would be that women and users with an interest in cosmetics will achieve a higher response. As a result we like to use interest and demography profiles to better focus the campaign, especially at launch.

We consider the creation of intent profiles that match the campaign to be much more suspenseful for good sales performance by audience targeting.
Marc Majewski, CEO

The Point of Entry as a Starting Point

To build up specific intent segments, we work with a very simple, binary rule: We search for the "point of entry" of the user, i.e. the event that reveals us a specific purchase intention and through which we can assign that user to a specific intent segment. To put it simply this means: He is in or not. We are guided by the distinction between "general purchase intention" and "active product search".

A Specific Example

On average, one looks for a new mobile phone rate every 2.5 years. Experience shows that one is in the phase of general buying intention for 6-10 weeks. During this time the realization matures that one wants to buy a new tariff soon. A user in this phase has a particularly high response to the appropriate product advertising. But that's only 6-10 weeks every 2.5 years. For example, you buy a dishwasher every 6-8 years and the general intention to buy only lasts a few days. Intent targeting follows the goal of reaching the respective user in exactly this period and is therefore the most effective and most important method of audience targeting.

How can this time period be identified?

Only a few hours the user is on an active product search. During this time the user inquires about purchase alternatives, either by going to a shop or by checking online. For us, this is the so-called point of entry. This means that we somehow have to know when the user is actively searching, because he usually will only do so if he also wants to buy.

Audience Targeting in the Right Environments

If, for example, we mark users on a mobile phone blog, then this is not enough. Why? Because some users spend time there without the intention to buy. This environment may well ensure interest in mobile communications - certainly a suitable keyword - but it is not a sufficient indication for intent. Shopping and e-commerce environments, product tests, shopping guides or product comparison pages and every form of search are better suited. With common sense it is easy to deduce an intention to buy from suitable product pages. Then, for example, we can set an intent "smartphone" and assume that the user is in the period of the general purchase intention. Furthermore, the banner display of our smartphone offer on smartphone test pages will reach the user at the rare moment of active intrinsic product search and is therefore perfectly chosen.

Three things we intentionally don't do

1. predictive targeting
   To us, predictive targeting sounds like this: "We know something that not even the consumer himself knows." We can only put an ironic question mark behind this statement, because fortune-telling is not part of our business strategy.
2. Statistical Twin
   This procedure simply cannot complement our behavioural methodology. Either there is a "point of entry" event or it is not. Thus for us, predicting such an event on the basis of significantly similar behavior patterns, is a variant of predictive targeting. And according to our experience, it is not a robust method for CPO performance.
3. Data Modelling
   Either we know the gender of a user or we do not. Modelling as an assumption about probabilities leaves too many unclear interpretations open and, in our experience, does not provide sufficient performance uplift.

Bottom Line

Seems like ee experience significant performance upsides above all when using common sense to set the point of entry for the intent segments in the right places, especially on e-commerce or product information pages suitable for advertising. In order to optimize the media purchase and to reduce waste with a corresponding effect on the effective CPOs, this approach is absolutely profitable. Interest and demographic data can also be helpful, especially at the start of a campaign, if they are reliable. Those who set up audience targeting with a sharp eye during the campaign start-up regularly have the analytical results on their side later on.