Most app campaigns are run on one metric: cost per install. It is easy, it shows up straight in the platform dashboard, and it can be pushed down quickly. That is exactly why it is dangerous.

The case: a services app in Saudi Arabia

I took over a services app's account running on Meta and TikTok. The team was judged on cost per install alone:

  • Cost per install: SAR 21.
  • Activation rate (installed, then opened an account): 22%.
  • So a real user cost SAR 95, and nobody was counting that number.

After 150 days:

  • 46,000 installs at SAR 8.6 per install.
  • Activation rose to 41%.
  • Cost per active user fell from SAR 95 to SAR 21.

The right metric: cost per activated user

By that formula, a SAR 10 install with 10% activation (SAR 100 per user) is far worse than a SAR 20 install with 50% activation (SAR 40 per user). Anyone judging by cost per install will always choose the first.

"Activation" depends on the app: opening an account, a first order, completing a profile, a first booking. What matters is that it is the event that means the user has started to get value from the app.

Step one: change the event campaigns learn from

The algorithm looks for more people who perform the event you optimise for. When that event is the install, it finds you the cheapest installers, and they often install everything and delete it.

We moved optimisation to the activation event itself and wired it to the platforms through the SDK. The algorithm started looking for people who use the app, not people who download it. Cost per install rose slightly at first, as expected. Then cost per active user started to fall clearly.

What you need technically

  1. Platform SDKs (Meta, TikTok, Google) or a mobile measurement partner (MMP) such as AppsFlyer or Adjust.
  2. A clear activation event sent from the app with the same name on every platform.
  3. Enough weekly events for the campaign to learn. If the event is very rare, start with a nearer event in the same direction.

Step two: look for drop-off outside the ads

Once campaigns improved, funnel analysis showed most of the drop-off was in the first two signup screens, not in the ads. Users install, open the app, hit a long signup and leave.

We removed two steps from signup, and activation rose before a single campaign was touched. That is one of the biggest lessons in app advertising: a media buyer who only sees the ads dashboard will try to solve a signup problem with more budget.

Tools such as Mixpanel or Firebase show exactly where users leave. Check them before judging campaigns.

Common mistakes in app campaigns

  • Celebrating a lower cost per install without looking at activation and retention.
  • Launching campaigns before app events are connected to the platforms.
  • Different event names on each platform, so they cannot be compared.
  • Ignoring the first minute in the app, usually the biggest source of drop-off.
  • Judging campaigns a few days after changing the event, before the algorithm has learned.

What to measure next

After activation, the next step is retention: how many users are still active after 7 and 30 days? And if the app earns revenue, the final metric is acquisition cost against user value over a set period. It is the same logic as break-even ROAS, applied to users instead of orders.

The bottom line

In app advertising, what you measure is what you get. Measure installs and you get installs. Measure activation and you get users. In this case study the difference between the two was SAR 74 per user.

If your app gets plenty of installs and few users, get in touch through the freelance media buyer page, or see the full case study.