App growth audit: the five layers that matter

1. Presence

Can someone searching for what you do find you?

This layer covers the name, subtitle, description, screenshots, ratings, localization and the directories people check before they open the store. It is the cheapest to fix and the one most teams skip. The ASO guide covers each field.

Read three numbers separately:

• Impressions. Few means a keyword or category problem.
• Product page views. Many impressions and few views means a ranking problem.
• Installs. Views with few installs means a listing problem.

If every install comes from someone who already knows the brand, organic search is not working.

2. First session

Does the user understand the value before being asked to pay?

This layer covers onboarding, the promise, the plan, account creation and permissions. Three screens can be enough; twenty can be too many.

Measure first-session completion and time to the activation event, the moment someone gets the outcome the ad promised.

A tutorial explains where the buttons are. Onboarding should let someone finish a real task before asking for anything, so the session ends with a result instead of a tour. The first-session guide breaks down the screens.

3. Retention

Does the user come back without being asked?

This layer decides whether the other four are worth the effort. An app that loses 80 percent of its users in three days has a product or promise problem, and no campaign fixes that.

Read D1, D7 and D30 by acquisition source. Separate users who return on their own from users a notification pushed back.

The average lies. An app with 25 percent D7 can have one channel at 40 percent and another at 8 percent. The strong channel is subsidizing the weak one.

4. Monetization

Does the user who stays pay enough to fund their acquisition?

This layer covers the paywall, price, trial, in-app purchases and advertising. Score it on net revenue per user at 30 days, after refunds, platform fees and support costs, rather than on conversion alone.

Teams often rebuild the paywall when the promise is the problem. Conversion rises for a month, then refunds and one-star reviews take it back.

5. Paid amplification

Do you know what a user is worth after 30 days, per channel?

Paid acquisition multiplies what the first four layers already produce. It does not create demand. When the listing converts poorly, a bigger budget buys the same churn faster.

Compare acquisition cost per channel and per creative with 30-day net value, read in cohorts. Day-one return settles nothing for an app that plans to last.

Why the order matters

The layers multiply:

users = impressions × listing conversion
      × first-session completion
      × retention
      × payer conversion

A 10 percent gain at each layer compounds to 61 percent. A 40 percent gain at one layer stays 40 percent. That arithmetic argues for fixing the constraint rather than the layer the team enjoys most.

A layer only moves when the one before it holds. More traffic to a listing that converts poorly is wasted traffic, and better retention in an app nobody understands keeps people who never got the point.

The constraint by app type

The framework travels across categories, but the binding constraint changes.

Type	Main constraint	Deciding number	Most profitable lever
Game	The player must return without new content	Ad value per player	Difficulty, daily loop, ad inventory
Dating	Attention is scarce and local	Match rate and D7 retention	Supply density, ranking, status
Subscription	The promise must be testable	Trial completion and monthly churn	Onboarding, proof of outcome, price

A game can survive weak onboarding when the loop lands immediately. A dating app with a supply imbalance cannot, whatever its paywall, and a subscription with a vague promise cannot, whatever its campaign.

The one-week audit

The week follows Ayo Omolewa's audit method: one number per app, session recordings, then a 50/50 test.

Day	Output
1	Map the five layers and the main number for each. A missing number is the first piece of work.
2	Measure the full funnel by source: installs, completion, activation, D7, payer.
3	Watch five session recordings. Note the screens where people hesitate, not opinions.
4	Pick one bottleneck. Write down why the other four can wait.
5	Write a testable hypothesis: if I change X, Y moves by Z.
6	Instrument a 50/50 test with one variable and a duration decided in advance.
7	Keep, cut or restart. Write the decision down, even if it turns out wrong.

Sources

• Ayo Omolewa, audit method: one number per app, OpenTelemetry traces, session recordings, 50/50 experiments — instagram.com/reel/DeNKjDeyYIl
• Ayo Omolewa, organic growth and App Store economics — instagram.com/reel/DcZa_49SNwn
• Charles Liu, net app economics and the real cost of growth — instagram.com/reel/DdeYnPFx03n