What is ad set splitting?

Ad set splitting is simply the process of breaking down larger audiences into smaller audiences based on demographics or other factors such as placement. Ad set splitting is extremely beneficial and allows you to zone in and shift budgets to top performing audiences, while pausing underperforming audiences. From a creative perspective, ad set splitting also allows you to discover the creative that is performing well for a particular segment, while pausing underperforming ads.

Here’s a list of some of the variables that you could decide to split:

  • Age (splitting 21 – 45 into 4 age brackets)
  • Gender
  • Geo-location
  • Placement (Newsfeed on desktop and mobile, right column on desktop and Instagram)
  • Interest based targeting
  • Behavioural based targeting
  • Further demographic targeting (such as relationship, work, education)

Important Considerations

It’s very easy to get carried away with splitting your audiences at campaign launch based on the above, which can rapidly increase the number of ad sets within your campaign. This then extends the amount of time it will take to make campaign optimisation decisions based on insufficient spend and limited audience reach.

As an example, let’s use a theoretical campaign where:

  • you have 12 creative for testing
  • a budget of R50 000
  • over a campaign duration of 1 month
  • testing gender splits
  • and splitting over 5 age brackets
  • over all 3 Facebook placements

This looks like a reasonable campaign structure. However, based on this, it would result in a total of 30 ad sets, each with a daily spend of R53. Each ad set has a total of 12 creative for testing, which effectively gives each creative a daily spend of R4. Each audience needs time to be tested, reaching a sufficient percentage of the audience. Similarly, each creative would also need time to be tested and over splitting does not allow for this. Based on your KPIs and expected results, this may be over splitting a campaign if your target is a cost per acquisition of, for example, R400. It will very likely take too long to determine any significant action points.

To overcome this, rather narrow the splits on campaign launch. For example, instead of running 5 age splits, test 2 and refine further once sufficient reach has been achieved, and then split the most successful from the first test.

Did this answer your question?