Interviews, insight & analysis on digital media & marketing

When and how to use data for programmatic buying

By Kirsten Leever, Manager, Solutions Consulting, Xandr

Whilst programmatic buying offers massive scale to advertisers, it comes at the cost of advertisers needing to find and address users who are interested in their product. Data usage to granularly target users can help to solve this, allowing advertisers to reach users with the right ad at the right time.

When looking at how best to reach relevant users for their campaigns, advertisers should consider the following factors: 

  1. The type of data. Should behavioural or contextual data should be used? Behavioural data is used to target users based on their previous actions or behaviours. Meanwhile contextual data targets users browsing content on a particular topic. While there are several factors that influence why advertisers should use either behavioural or contextual data, one key distinction to bear in mind is that behavioural data is almost always cookie-based while contextual data is not.
  2. The quality of the data. Here we consider first-party versus third-party data. First-party data is data gathered by the advertiser directly. This comes with the advantage of users who have directly shown interest in the advertiser’s product. Third-party data has instead been collected by parties other than the advertiser, usually coming from companies specialising in data collection known as ‘data providers.’ They then use different methodologies and data sources to identify what interests and purchase intents users have. Since third-party data is modelled using multiple data points, there is not a guarantee that users would be interested in advertisers’ product, but rather a likelihood.
  3. The price of the data. For campaigns optimising towards price-based metrics, it is important to consider that any additional costs added on to the campaign will increase the price, so data should only be used if performance justifies the extra cost.
  4. The scalability of the data. Going back to the example of first versus third-party data, we usually see that first-party data has limited reach because of the difficulty in identifying interested users. Whereas since third-party data is collected across multiple user touchpoints, it has much more scalability than first-party data.

With these factors in mind, advertisers can make informed decisions on the right data strategy for their campaign objective. For example, campaigns with a branding-oriented goal are trying to reach as many interested users as possible. Therefore, when looking at our factors we can say that third-party data would be the better choice. Advertisers who are trying to generate interest and not direct purchases will need the scalability that third-party data provides and not the accuracy of first-party data. Additionally, as there is no need to track any type of events, cookieless traffic can be targeted together with the usage of contextual data. Finally, price only comes into play with the general price sensitivity of the advertiser, so the price of the data is less relevant.

For campaigns with a traffic generation goal, data usage is similar with one difference: price will be an important factor since advertisers are looking to drive as many users as possible to their site with the available budget.

The most nuanced types of campaigns regarding data usage have a performance-oriented goal such as CPA. The important distinction here is whether advertisers would like to use prospecting or retargeting strategies. With retargeting, first-party behavioural data should be used to readdress users that have a demonstrated interest in the brand. Due to the quality of this data, it is usually highly performant, however, it is only as scalable as the amount of data the advertiser was able to collect. Prospecting strategies on the other hand require a tricky balancing act between all four factors: the type of data should simply be based on how performant it is, however, since this is a cost-based metric, advertisers will be highly price-sensitive. Quality and scale will also need to be high since advertisers are trying to reach new customers that are willing to convert. In order to manage this balancing act, custom, third-party data such as lookalikes is best. This allows for more scalability than pure first-party data while still basing the data on proven, interested users.

Keeping these considerations in mind while selecting data for your programmatic campaigns gives advertisers the framework for running successful buying; allowing them to achieve their desired business results no matter the goal.