Selecting and reaching the right audience is a crucial factor in digital marketing as the performance of your marketing campaigns depends on it considerably. So let’s take a look at some of the most effective targeting methods available today.
Sociodemographic targeting – readily available but not very precise
Sociodemographic targeting uses – as the name implies – socio-demographic aspects such as age, gender, occupation or income to deliver content, advertisements and other messages to the right target group.
The data for this is used anonymously in most cases since otherwise, it requires the explicit consent of the respective user (opt-in). Normally, the success of this measure is very closely linked to the accuracy, completeness, and volume of the information provided. However, there are already providers in the market who can fill in any gaps in these data sets with the help of artificial intelligence (AI) and machine learning, so that even with small data sets good targeting is possible. The available user data is the basis for the ‘training of the machine’, which is then able to make extrapolations.
Such targeting can be useful if you know that the segment you are approaching is not well defined and your focus is on brand awareness. Consider sociodemographic targeting as a high-reach approach. The instrument is not very precise, so the message is not really tailor-made – but sometimes this is not necessary (think of a brand like CocaCola, for example).
Keyword targeting – considerably more precise
Keyword targeting makes sense if companies already know a little more about their target audience or want to cut scattering losses (which definitely arise in socio-demographic targeting). For this, individual categories are defined in order to describe a target group in more detail. This can be, for example, the category “winter sports”. Within this category, key terms such as “skiing”, “winter holidays” or even the names of specific mountain regions can be created. Algorithms now find in the content analysis the pages on which these keywords fall and by the user analysis those cookies or user profiles that have interacted with these pages.
Unlike keyword targeting, this form relies on semantic analysis of published content to find suitable ad space. For this, the entire website is technically crawled according to linguistic aspects and also examined for moods (positive/negative) and ambiguities. Thus, advertisements can be placed where editorial content supports or at least does not contradict the advertising message.
Behavioral targeting is the hottest technique these days and also the most controversial. It allows marketers to track users’ site “hopping” with the help of cookies to design models and behavioral patterns for targeting users. These models can be used to serve ads and other content that is relevant to those same individuals across various sites. It is probably more cost effective to use behavioral as opposed to contextual targeting because there are more contact points with the same customers.
Some service providers also offer campaign targeting. Here, users who interact with a particular campaign (clicked on an ad) can be targeted directly to a new campaign or even coordinated with the frequency of ads within a multi-platform (DSP) campaign. In practice, this approach is especially useful for those companies that have already had a successful campaign with their service provider and want to increase the ROI again in the second step.
Again, there is a downside: not every user who has clicked on an ad is really interested in the product shown. Maybe he has decided after the click that he doesn’t like the advertised item and now he is literally chased by it. Here it makes sense to match the campaign data again and again so that such cases can be sorted out quickly.
In general, the more detailed the targeting (that is: the more attributes, categories or limitation sizes), the smaller the target group. That is, if several of the above-mentioned types of targeting are combined, the scattering losses are very small, but the targeted audience may just be too small to be truly effective. However, there are already solutions to this problem. Namely, it is possible to use this audience only as a starting point (a so-called seed audience) and filter out further profiles that match those characteristics in your own data sets or in major exchanges (Look-alike modeling). This way, you can be sure that your extended audience is large but also detailed enough to cost-effectively bring very specific messages to the market.
This approach also works when an advertiser wants to use existing customers as the basis for an extended audience definition. For this, he takes the output data of let’s say 1,000 customers and calculates a target group of 500,000 potential customers in a specific product category.
Only the sky is the limit! No, wait…
The new General Data Protection Regulation (GDPR) will inevitably restrict the available and usable data for these targeting techniques. For example, sociodemographic data will not be available for broad targeting at the extent we see today. It is to be expected that companies increasingly refer to their so-called “legitimate interest” in order to be able to carry out targeting measures. Without any doubt, keyword and semantic targeting will become more and more important along with other sophisticated methods. Speaking of which, find out how a new digital platform called BehaviourExchange can help you with all your targeting requirements – just in time for your Christmas campaign!