Apps and software
An install is the cheapest thing to buy and the least useful thing to count. We know because we buy them for our own apps.
You run an app or a software product. People sign up or install, some of them stay, and the difference between those two groups is your whole business.
What we keep finding
Four things that are true in almost every one of these businesses.
Acquisition is measured, activation is not
Everybody knows the cost per install or per signup. Far fewer can say what a user who is still there in a week cost, which is the only one of the two worth optimising toward.
The first session is a mystery
People leave on a specific screen, on specific devices, and the reason is usually mundane and fixable. Without event level measurement it stays a discussion about the onboarding in general.
Crashes are counted, consequences are not
A crash rate is a number for engineers. The number for the business is how many of those people never opened it again.
Ad revenue and experience are argued about with opinions
One more placement earns more per session and costs some proportion of tomorrow's sessions. That trade can be measured, and almost nobody measures it.
What we measure here
The numbers this sector lives on.
Every business gets the fundamentals. These are the ones that matter here and would be a waste of money somewhere else.
- Activation defined as a real moment in the product, then measured
- Retention by cohort and by acquisition source, not as one blended curve
- Cost per retained user, alongside cost per install
- First session drop off by screen and by device model
- Crash impact measured in users lost rather than in crash rate
- Ad or paywall placement tested against next day return, not against session revenue alone
Proof
Zip VPN
Our own app, measured the way we would measure yours.
Our own app, around twenty thousand users a day, measured the way we would measure yours.
Read the case study- users a day
20,000
users a day
- events a day
300,000
events a day
- of users come back
20%
of users come back
Where this usually starts
Two pieces of work, in this order.
Questions people ask
The ones that actually arrive.
Do you work on iOS products too?
For measurement and advertising, yes. For building the app itself, Android is what we publish and maintain, so that is where our judgement is worth paying for and we will say so plainly.
We already use a product analytics tool.
Good. The usual problem is not the tool, it is that nobody agreed what activation means, so every chart is measuring something slightly different. That is a week of work and it makes the tool you already pay for worth having.
Can you help with app store listings?
Yes. We publish our own, so store listings, Data safety forms and staged rollouts are ordinary work here rather than something we would be learning on your account.
Your turn
Bring the number you do not trust.
30 minutes. Most of these conversations start with somebody describing a report they have stopped believing.