Jump to a section
Leave with a decision
Explain percentage points, relative change and the limits of the comparison.
Start with the visual explanation, work through the example, then try the exercise on your app.
The same rate change has two descriptions
Illustrative data: both periods have 10,000 comparable listing visitors.
Conversion rate · 0–100% · equal scale for both bars
Before · 20%
2,000 installs from 10,000 visitors.
After · 22%
2,200 installs from 10,000 visitors.
Work through it · fictional example
Separate a better rate from a different audience.
A fictional listing gets 10,000 comparable visitors. Its conversion changes from 20% to 22%. That is 2,000 versus 2,200 installs in the scenario.
Why this changes the decision
Two percentage points is the subtraction of the rates. Ten percent relative improvement is 2 divided by the original 20. The extra 200 installs assumes the same 10,000 visitors. If the second period contains more branded or paid traffic, the aggregate change may reflect that mix instead of your new screenshots.
What to check before calling it a win
Break down the periods by available source and market information, annotate releases and promotions, and record uncertainty. A before/after report can describe what happened without proving which change caused it.
Your turn · a five-minute start
Make one useful decision for your app.
Write one result with its denominator, source, period and a competing explanation. Calculate both percentage-point and relative change.
What a useful answer looks like
Moving from 20% to 22% is not a 2% relative lift. Do not confuse the two in a client report.
Begin with metric definitions
Record the source, denominator, time period and audience for every metric. A conversion rate is incomplete without knowing what is being converted. Keep iOS and Play reports separate unless their definitions genuinely align. Use console values as defined rather than renaming every source-filtered download “organic.” Apple explicitly notes that Apple Ads performance can appear in its search reporting.
Create a baseline and change log
Save a comparable period before an update, with notes about releases, campaigns, featuring and seasonal events. A chart without those annotations is easy to misread. Baselines should reflect the decision and available traffic, not an arbitrary fixed duration. Keep original exports so a later investigation can reproduce the comparison.
Check the context before celebrating
A rising chart is a question, not a verdict.
Definition
Use the same metric and denominator in both periods.
Audience
Check whether source, country or campaign mix changed.
Timing
Account for seasonality, releases and other simultaneous changes.
Decision
State what the data supports, what is uncertain and what to investigate next.
Distinguish reach from persuasion
Rankings, impressions, listing visits and downloads answer different questions. A visibility change can be valuable without immediately changing conversion; a conversion improvement can be hidden by a less relevant traffic mix. Break down results where the console permits and avoid comparing a paid campaign audience with a different organic audience as though the groups were equivalent.
Interpret experiments and before/after changes differently
A native controlled experiment and a simple time comparison provide different kinds of evidence. For a controlled test, read the platform’s uncertainty and eligibility guidance. For before/after work, document alternative explanations and keep claims modest. Do not declare success based solely on the day the chart moves upward.
Connect installs with business value
Where privacy and data access allow, investigate activation, retention and monetization of acquired cohorts. The listing may attract more people who are poorly matched to the product. Record whether the business outcome improved alongside the acquisition metric. A hypothetical value-per-install calculation is a planning aid, not measured incremental revenue.
A conversion change, calculated carefully
Illustrative scenario: 10,000 comparable listing visitors generate 2,000 downloads at a 20% conversion rate. At 22%, the same number of visitors would generate 2,200 downloads. That is 200 additional downloads in the scenario, a two-percentage-point change and a 10% relative improvement.
| Measure | Baseline scenario | Alternative scenario |
|---|---|---|
| Comparable visitors | 10,000 | 10,000 |
| Conversion rate | 20% | 22% |
| Downloads | 2,000 | 2,200 |
| Additional downloads | — | 200 |
These are invented inputs, not a forecast or a measured result. A before-and-after report with different traffic sources cannot assume that a listing change caused the difference. Keep the denominator, audience and measurement definition visible.
A short report that supports a decision
- What changed, where and when?
- Which metric and audience are being compared?
- What else changed during the same period?
- What can the evidence support, and what remains uncertain?
- What should the team keep, test or investigate next?
Your next steps
- Define the metrics and periods before changing a listing.
- Annotate paid campaigns, releases and other plausible explanations.
- Record outcomes and uncertainty, including inconclusive tests.
- Review whether new users reach the intended product value.
Sources and platform guidance
Platform documentation checked 29 September 2026. Verify the applicable console settings before publishing.
Questions and answers
Can I claim organic lift from a search-source report alone?
Check the console’s definition and paid traffic treatment first. A source label is not a causal attribution method.
What if the test is inconclusive?
Record it honestly. You can keep the control, refine the hypothesis or gather more appropriate evidence without manufacturing a winner.
Work directly with Gabriel
Find the problem. Leave with the priorities.
Send your listing and the decision you are stuck on. I’ll scope an audit with annotated evidence, prioritized actions and a handover your team can use. A report without a next action is an expensive obituary.
Help me interpret my ASO results →Or email gabriel@asoprofessional.com. Bring your app link, market and the decision you need help with.
Want a first opinion? Get a free video teardown of your listing.