Over a third of men who claim to hate Justin Bieber… actually love him.
A good campaign setup on Facebook will have a mix of interest, lookalike and custom audience targeting, but have you ever considered how much your audiences overlap?
If audiences in different campaigns are overlapping then you have 2 or more campaigns competing with each other to reach the same user – driving up your own costs and negatively impacting your ROI.
Facebook has now released a helpful tool so you can easily check your audience overlap between ‘Saved Audiences’, ‘Lookalike Audiences’ and ‘Custom Audiences’ . Simply go into your Audience Manager (in Ads Manager or Power Editor) → select two audiences → Actions → Show Audience Overlap.
Let’s take a really simple example – a high-street fashion brand has launched a campaign to advertise their new spring range. They have set up one campaign targeting F18+ who like Topshop, and another campaign targeting F18+ who like ASOS.
The tool shows that there is a 51% overlap between these two audiences, therefore they are driving up their own costs on over 50% of the overall target audience! Have a play around with the tool and see how much your audiences overlap. Interestingly, we found out:
- 38% of men who like the page ‘I hate Justin Bieber’ also like Justin Bieber on Facebook #indenial
- 66% of people who like Pepsi also like Coca Cola #ShareAPepsi
I’ve got Audience overlap… what should I do!?
What you do depends on which audience types are overlapping…and what % the overlap is. Here are a few options you have:
If you have a lookalike overlapping with interest targeting, simply exclude the lookalike from the interest targeting. This is best practice for all campaign setups
If it is either interest targeting overlapping with another interest target or a lookalike overlapping with another lookalike, consider combining the two overlapping audiences into one adset. If both of these targeting options are high performers, the combined performance of both types of targeting in one adset should improve performance as you won’t be competing against yourself for the overlapped audience.
When you have a high overlap of 75% between two audiences, with ‘Audience 1’ performing significantly better than ‘Audience 2’. Consider cutting out the low performer entirely. The most value is found in ‘Audience 1’. By cutting out ‘Audience 2’ the 25% of lower value traffic isn’t targeted anymore. You can improve performance in ‘Audience 1’ as 75% of the audience won’t be targeted in multiple adsets.
Is the overlap less than 25%? It would be great if you could solve this by combining audiences or exclusions, but I wouldn’t worry TOO much if the overlap is this low.
Take a custom audience that you want to make a lookalike out of and split it into two groups of the same size (A & B). Build lookalikes of both audiences and check the audience overlap.
If the audiences have a high overlap, then this is a high quality seed audience as it suggests there are similar people within the audience. However, if you have low overlap it suggest that the seed is not defined well enough.
So there you have it! Reducing audience overlap to 0% would be a near impossible task, but minimising should be an easy way to help drive higher ROI on Facebook. At quickthink we expect the addition of this tool to positively impact the day to day running of campaigns for our clients.
Alan Frankton is head of digital media at quickthink.