App makers · Store conversion
Why did your App Store conversion rate drop?
Recalculate downloads using last week’s rate for each source. The worked example explains how overall conversion falls from 10.4% to 7.2% while both channels hold steady.
GimmeTheSauce editorial ·
Imagine opening App Store Connect on Monday. Downloads are down by 320. You remember changing screenshot two on Wednesday. Before you spend the afternoon reverting it, answer a cheaper question. What would last week’s visitors have done if each source had converted exactly as it did the week before?
Rebuild the number from two rows
Here’s an illustrative example for one territory, one platform and an app with no pre-orders. Each cell shows unique impressions, then downloads. Assume the source groups in this example do not overlap.
| Source | Week A | Week B | Downloads per impression |
|---|---|---|---|
| App Store search | 8,000 → 960 | 4,000 → 480 | 12% both weeks |
| App referrer | 2,000 → 80 | 6,000 → 240 | 4% both weeks |
| Overall | 10,000 → 1,040 | 10,000 → 720 | 10.4% → 7.2% |
The share of impressions coming from search halved. Apply the old rates to the new volumes: 4,000 × 12% + 6,000 × 4% = 720 expected downloads. That matches what happened. The aggregate drop supplies no evidence that screenshot two hurt conversion. It also tells you nothing about whether the extra referral users were worth acquiring.
Use Apple’s definitions when you copy the figures. Its store conversion metric uses unique device impressions as the denominator. Total downloads include redownloads, and pre-orders have their own counting rule. Dividing first-time downloads by product page views answers a different question. Apple’s metric definitions
Do the comparison on your own app
- Pick two complete periods of equal length with the same weekdays. Keep territory and platform fixed for the first pass. Note the dates of releases, featuring and campaign launches.
- In Acquisition, break each period down by source type. Record the numerator and denominator for each row. Keep small sources together if a handful of downloads would reverse your conclusion. Apple’s acquisition guide
- For each source, multiply this period’s impressions by last period’s rate. Subtract that expected count from the downloads you actually saw. Rank sources by the number of missing downloads, rather than by the biggest percentage swing. A new source has no historical rate. Keep it separate.
- Check your reconstruction against Apple’s reported total. Unique devices can show up in more than one segment. If the totals do not reconcile, use the rows to locate a change and resist claiming an exact decomposition.
Make the next decision in downloads
Suppose search delivered 360 downloads from those 4,000 impressions instead of 480. That leaves 120 fewer search downloads than the old rate predicts. Now there’s a specific audience to investigate. Look at what search visitors actually see, which terms or markets changed where that information is available, and whether the release changed the promise on the page.
If the leading explanation is a screenshot, write the hypothesis before designing it. For a receipt scanner, try “show the exported expense report first” against the current opening image. Hold the icon and the other screenshots steady. Apple’s product page optimization report gives you a baseline, estimated lift and confidence for a treatment. A single Monday total gives you none of those controls. Apple’s product page optimization guide