Well, there is no generic answer to that of course.
But let’s have a look at a specific example of how heavily this still widely used attribution model underestimates the impact of Facebook.
No need to change your attribution model today
Do you have to change your attribution model before you can achieve success on Facebook? No, but here is a bit of inspiration on how to improve the way you set KPIs to compensate the negative bias that a last-click model has towards Facebook.
What you should do, is a conversion lift test. We already wrote extensively about this here. Put simply, you compare the number of conversions in a test group that sees your Facebook campaign with the number in a control group that does not see any ads on Facebook. It will give you a scientific valuation of the impact of Facebook that is attribution model free.
Being too pessimistic won’t get you far in life and on Facebook
Let’s look into a recent test of one of our clients. In this specific test the client found out that there were 22k more conversions in the group that was exposed to their FB campaign compared to the control group who did not see any advertising on Facebook.
The client’s last-click same-session attribution model however counted only 7k conversions for that Facebook campaign. So they are a factor-3 pessimistic when evaluating what the campaign on Facebook brings them. We all know pessimism is not the key to success. This client would actually have been overestimating the cost per conversion with a factor three. And this of course would make them hesitant to scale their Facebook efforts – for the wrong reasons.
And that’s not the end of it.
Look at what happens cross-device before you decide where to put your money
Let’s look at the conversions in this test that happened on desktop. The lift study showed only 29% of the bookings came from people who actually saw the ad on desktop and then also converted on desktop. The other 71% saw the ad on mobile, then converted on desktop.
If you have a same session last click attribution model then you are completely blind to this.
You will actually try to spend more budget on desktop because that’s where you see more conversions in your model. And by doing so you’ll miss out on what in this case was the majority of people: those who would see your ad on mobile but then book from their desktop without seeing/clicking another ad on desktop.
Use more generous KPIs to compensate for attribution blindness
So do not fall into this trap. It’s essential to gather the data you need to quantify how generously you should set your KPI target on Facebook because you know your attribution model underestimates the value of Facebook. Likewise you should make sure to strike the balance between mobile and desktop placement knowing your model doesn’t take into account cross-device activity of the same user.