Here is a number I cannot stop thinking about this week: 26.

Campaignswell talked to 116 app and game teams about how they decide whether to scale or cut ad spend. 64 of those teams have an MMP installed. Only 26 of them make that decision in it.

I run a company that builds one. So this stings a little.

In June I said that data only matters if it drives decisions. This week an outside study put a number on how rarely that happens.

First, what this report is and is not

Campaignswell sells a prediction layer for subscription apps. The report says the sample leans toward teams already looking for better analytics. 76% of the teams are subscription or in-app purchase businesses outside games. Games are a fifth. Monthly spend ranges from $6K to $3M+.

So read it as a very good set of interviews, not as a market census. Used that way, it is gold. Teams are unusually open about what is broken, in their own words.

Three findings I would pin on the wall

1. Own BI is the most common setup (28 of 116), and it rarely tells you whether a campaign will pay back. The warehouse shows what a cohort has paid so far. To estimate what it will pay, most teams apply a multiplier from older cohorts, or wait.

2. Of 64 teams with an MMP, 38 make the call somewhere else. The MMP numbers go into a spreadsheet, a warehouse or a forecast model. The decision happens there.

3. Seven in ten teams (82 of 116) have no reliable forecast for a new cohort. They judge a week-old cohort by the ad dashboard, early ROAS, one average LTV, or a multiplier. All four assume new users behave like old ones. That stops being true when price, geo or channel changes.

One games studio co-founder in the report leaves a campaign for about two weeks before touching it, and calls it “more of an art, not a science”. It is a fair description of a lot of studios I talk to.

It is a decision problem, not a data problem

The data is mostly there. Attribution is there. Revenue is there. Spend is there. What is missing is one place where they meet, and a clear answer to the question that matters: what happens to this cohort, and what should I do next?

The report names four things a team needs to scale or cut with confidence: attribution, net revenue, spend and revenue stitched together, and a one-year LTV or ROAS prediction. I agree with all four. And I would add a fifth.

The fifth thing: what players did before they paid, or left

Everything on the report’s list is about who came in and what they paid. None of it is about what they did inside the product in between.

That is why we started talking to FastSight. We wrote about it together: a day-based number like retention day 7 is an average of players having very different experiences, and it arrives late. By then the budget behind that cohort is spent. FastSight looks at games in stages, onboarding, main loop, metagame and late game, instead of days. The answer is often in the first hours.

That connects directly to the report’s own diagnosis: multipliers break when new cohorts behave differently. The first minutes and hours inside the game are where you can see that happening, long before day 7 or day 30. Did the new geo get through onboarding? Did the new channel reach the main loop? Did anyone see the offer that actually drives purchases?

A small example from FastSight: during a LiveOps event, loading times between levels were 2 to 3 times longer for some cohorts. A certain combination of device and country got a much worse user experience. No revenue dashboard flags that. The cohort looks “worse”, the multiplier drifts, and nobody knows why.

This is our view, and we are still working out together how the layers fit. But I am convinced the direction is right: acquisition, revenue and behaviour in one picture.

Why now

  • Store data is a shrinking view of the business. Sensor Tower puts US App Store spending down 6% year on year in Q2, and estimates about 30% of revenue of the top 100 US casino titles never shows up in store data.
  • Installs are down and CPI is up (Newzoo, as I read the latest numbers). Teams that decide on retention and LTV win.
  • Hybrid games make cohorts harder to read. One studio in the report could not read young cohorts once in-app purchases grew next to ads. If the EU Digital Fairness Act changes how currency and offers work, that gets harder again.
  • And agents. MCP is quickly becoming standard. The question is what the agent can actually tell you. If your numbers disagree across five tools, an agent just disagrees faster.

Where we are heading

I do not want to bury this: this report describes the problem we are building justtrack to solve. Our mission is to help the people building apps and games know what to do next to grow, and why. In one sentence: we unify all your relevant growth metrics in one view and tell you and your agents what to do next, and why.

Where are we? Not everything is perfect. But we have come far, and the foundation is exactly what we need: attribution, in-app behaviour, revenue and spend in one data model, with forecasts and Smart Insights on top, and an MCP that agents can already read from. The next steps are the ones that matter most: one view that is clear in seconds, recommendations you can act on, and agents that can do more than read.

And a few things we do not claim yet, like net revenue after taxes and refunds, or predictions at creative level. They are the report’s own bar, and we measure ourselves against it. I would rather say that out loud than hide it.

We do not need to do everything ourselves. Our goal is to deliver answers. Sometimes the better answer comes from working with others: a behavioural layer like FastSight, a prediction specialist, a partner with data we do not have. We are open to partnerships that add value for our clients, not ones that just add cost. And we want justtrack to be the best basis for them: clean, unified data that other tools and agents can build on.

The name should follow the work, not the other way around. We will keep showing what is live, and what is next.

What I would do now

Campaignswell’s CEO suggests a simple test, and I like it. Pick a cohort from three months ago. Write down what you expected it to earn in its first week. Compare it with what it actually earned. If the two are close, your setup works. If they are far apart, or you cannot find what you expected, you have found your weak spot.

Then:

  1. Write down where you actually decide. Not where the data lives. Where the decision gets made.
  2. Separate gross from net. One studio in the report saw a 60-day payback on gross become 60 to 90 days on net.
  3. Name your multiplier. What does it assume about price, geo and channel? When did you last check?
  4. Add one behavioural signal early. Stage reach, session length, or what share of players even see your main purchase driver.
  5. Check where your tools disagree. If the same cohort looks profitable in one and loss-making in another, fix that before you scale.

What this report does not tell us

It does not tell us how studios with a strong data team differ from the rest, how the picture looks for games only, or how well each setup performs in numbers. The accuracy and spend figures in its case studies are Campaignswell’s own. The report itself says that if you pay back within a week, a simple setup can be enough for a long time.

Why I care

I have spent 16 years watching companies sit on data they cannot fully use. Gaming forced the answer first, because the economics left no room for guesswork. Subscription apps and web funnels are arriving at the same place now.

If you recognise your team in one of the six setups, I would like to hear where it breaks for you. Fifteen minutes, no pitch. Very open, and challenge welcome. Or start for free up to 10K MAU and see it on your own data.

FAQ

How many teams actually decide ad spend inside their MMP?

In a 2026 study of 116 app and game teams by Campaignswell, 64 had an MMP installed and only 26 made the scale-or-cut decision in it. The other 38 decided in a spreadsheet, warehouse or forecast model.

Why don’t teams decide in their MMP?

Not because data is missing. Attribution, revenue and spend exist, but they rarely meet in one place with a clear next step, and most teams have no reliable forecast for a new cohort.

What does it take to decide whether to scale or cut a campaign?

Attribution, net revenue, spend and revenue stitched together, and a forward-looking LTV or ROAS prediction, plus early behavioural signals, all in one view.

Built for studios that want to win. Not just measure.

Start your free trial today, or talk to us first. Either way, we’ll give you a straight answer about what justtrack looks like for a studio at your scale.

No credit card required