# A download decline that begins outside the screenshot file

Companion to /learn/problems/downloads-dropped/

## Outcome
Investigate an acquisition decline in a sensible order before changing the listing.

Fictional example, not a client result.

## Scenario
RestSpace is a fictional sleep app. Reported acquisitions fall after a weekend. The team wants an immediate screenshot redesign, but a campaign also ended, a new app version shipped, and the analyst changed the dashboard’s default filter. Several plausible causes are present; creative quality is only one hypothesis.

## Reasoning
First verify that the decline is real under consistent definitions, filters and time windows. Then inspect availability and product incidents, source composition and relevant listing changes. Break the investigation into questions the available data can answer. Urgency should shorten the path to evidence, not eliminate it.

## Action
Restore comparable filters and inspect source-level counts. Check whether the acquisition campaign ending explains part of the movement. Ask the release owner about availability and onboarding issues. Only then investigate whether relevant listing visitors are taking the defined action less often. Preserve the original listing during diagnosis unless an accuracy or operational issue requires correction.

## Trap to avoid
Do not use a before-and-after chart alone to blame the most recent metadata edit. Temporal proximity suggests a question, not a proven cause. Record other changes and test the most plausible explanations in an order that can prevent unnecessary work.

## Advanced considerations
Use a stopping rule for the investigation. Once there is sufficient evidence for an operational fix, assign and verify it rather than waiting for a perfect explanation of every fluctuation. After the fix, review the same comparable segments. If the decline remains, reopen the next hypothesis with the improved evidence instead of quietly declaring success.

## Exercise
The decline appears only in one country after a release. What should you check before launching a global creative test?

## Suggested answer
Inspect that market’s availability, localized assets, product behaviour and source mix. A localized operational or audience change may explain the signal. A global experiment may be irrelevant and could obscure the original problem.

## Your worksheet

### Date and scope of the observed decline
Answer:
Evidence:
Uncertainty:
Owner:
Next action:

### Metric, source and comparable filters
Answer:
Evidence:
Uncertainty:
Owner:
Next action:

### Affected markets, stores and sources
Answer:
Evidence:
Uncertainty:
Owner:
Next action:

### Campaign, release and availability timeline
Answer:
Evidence:
Uncertainty:
Owner:
Next action:

### Competing hypotheses and evidence checks
Answer:
Evidence:
Uncertainty:
Owner:
Next action:

### Responsible owner, action and follow-up
Answer:
Evidence:
Uncertainty:
Owner:
Next action:

## An incident triage record for RestSpace
This fictional sequence follows the app’s campaign ending, version release and dashboard-filter change. Each check has an explicit branch. Work in parallel where owners are independent, but avoid making several speculative listing changes that obscure the original signal.

Illustrative diagnostic branches, not an automated diagnosis or a fixed order for every incident.
Investigation checkpoint | Evidence to collect | Decision branch
Measurement comparable? | Original and current event definitions, filters and complete periods | If not, repair the comparison before declaring a decline
Availability or release issue? | Affected markets, device context and release-owner findings | If confirmed, route to the operational owner and verify the fix
Traffic changed? | Source-level counts and campaign end date | If concentrated in ended activity, explain the exposure change
Comparable listing response changed? | Same event and population, with audience mix inspected | If supported, investigate a specific message or creative hypothesis
Cause still unresolved? | Completed checks and remaining evidence gaps | Assign the next discriminating check; preserve the baseline

### Preserve an incident snapshot
Save the date the decline was noticed, the period it concerns, affected segments, exact report definitions and the last known listing versions. Ask each relevant owner for a timeline rather than relying on memory in a group chat. Keep facts and interpretations separate so a convenient recent change does not become the assumed cause.

### Verify a fix against the original symptom
If an availability issue is confirmed, record what was corrected and inspect the same affected context afterwards. Do not declare resolution because the overall total rises while the original segment remains broken. Where evidence arrives with a delay, state when verification can occur and who owns that follow-up.

### Escalate uncertainty without guessing
When accessible evidence cannot distinguish a traffic change from a reporting issue, request the specific missing export or owner review. Avoid filling the gap with a general story about an algorithm update. If the next action is a creative investigation, document why operational and measurement explanations have become less plausible, while retaining unresolved alternatives.

## Continue reading
- /learn/measurement/monthly-reporting/
- /learn/problems/rankings-dropped/
- /learn/problems/low-product-page-conversion/