Learning about your users through Facebook ads

Did you review your mobile app install campaigns from a research perspective already?


Facebook’s mobile app install ads are extremely popular amongst app developers these days. They have become the user acquisition channel for many developers who want to drive quality users to their applications. The main reason for that is the very granular and efficient targeting criteria that empower you to reach your campaign goal. More relevance for the user means better performance in terms of click-trough rates, conversion rates, cost per install and retention. But besides the best possible CPI there is another thing that you can leverage Facebook’s powerful targeting capabilities for: user research.

Although advanced tracking technologies collect more data from users than ever before startups often don’t have an exact idea of what their (potential) target group looks like. Especially smaller teams have more urgent challenges to manage than setting up extensive market/target group/user research. Running a new marketing campaign, shipping the next update with that cool new feature or fixing this annoying bug is usually more pressing than setting up user interviews, in-app surveys or other activities that can teach you more about your user.

There are good news though: If you are already running mobile app install ads on Facebook you can learn a lot about your users by applying some tweaks to your campaigns. You also might want to consider them when setting up your very first campaigns.

1. Create targeting clusters

If you already use app install ads to acquire users you‘ve probably already created targeting clusters — these are basically just the combination of targeting criteria that you use for your campaigns. By defining those clusters more structured and more thoughtful you will be able to get some valueable insights from your campaigns in the future.

Try to come up with ideas on what a stereotypical user of your application could look like and then think of stuff that this person might be interested in (= keywords). What does he do in his spare time? Which services, websites or apps does he use? Create a list with these interests/keywords (I usually do this in Excel), which you can then use to find corresponding interests and behaviors in the Facebook ads manager when setting up the targeting of your campaign. Give each user group you come up with a catchy name (e. g. Console Gamers, Basketball Addicts, Startup Hipsters), which will be the name of your targeting cluster.

This is what your “Console Gamers” keyword column in Excel could look like:

Console Gamers
Playstation
Wii
XBOX
Assassin’s Creed
Call of Duty
FIFA Football 15
Far Cry

Once you have defined your user groups by interests you just have to apply some splits. Split the user groups by whatever targeting criteria makes sense for your product, e. g. by geo, age, gender or education. Et voilà: You have pretty granular targeting clusters that can now act as ad sets in your campaign setup. Here is an example:

Spain | Console Gamers | Male | 25 - 34 | College degree

2. Analyze by cluster

If you have collected a relevant data set and you are ready to start analyzing the performance of each cluster with regards to your main conversion KPIs (Cost per Install, Cost per Registration, Cost per Acquisiton, Cost per Order, etc.)

The new structure of your campaign enables you to calculate the performance for each of the splits you applied. In our example you would be able to calculate the Cost per Order for each

  • Country (Spain)
  • Interest Group (Console Gamers)
  • Gender (Male)
  • Age group (25 - 34)
  • Education (College degree)

3. Look at the whole funnel

While looking at the end of your conversion funnel you should not forget the conversion steps that happen before goals like registrations or orders. Especially the click-through rate can be an interesting metric if you apply the concept of targeting clusters in the described way.

If you test various creatives (image + text) against each other the click performance per targeting cluster can be an interesting indicator on what each target group may or may not like about your product, marketing copy, app store screenshots, welcome screen, website, ...

4. Draw conclusions

Besides allowing you a (hopefully) more efficient optimzation of your app install campaigns this setup will enable you to gain interesting insights on your target group and your users.

Let’s stick to the example above: Imagine a company that develops a casual sports game. The company wants to see how their game peforms in different markets before they decide where they should expand to with further marketing efforts. Running app install ads with properly structured targeting clusters is an ideal tool to do that. They could compare campaign KPIs …

… by age groups in different markets to set the right tone of voice in their local landing page.
… by level of education to get an idea of the monetization potential (per market).
... by interest group of highly vs. poorly engaged users to find out which advertising enviroment to look at when planning TV ads.

Of course the conclusions you could come up with are highly individual to your business and the insights you can get cannot replace a proper user research. I still find this campaign setup helpful to get more out of Facebook campaigns than just installs and hope you will too.