Why a store listing changes the calculator's inputs
On the website the default page assumes a visit, a session in analytics and an action recorded by a tag. An app store page has none of that. The store owns the page, records the view and counts the install, and the only window into the numbers is the store's own console. Web analytics cannot see the product page at all, which is why app measurement treats the store as a separate step between the ad or search and the first app open.
The store funnel also mixes very different audiences on one page. Someone who searched for the app's name already wants it; someone scrolling a category chart is browsing; someone arriving from a review site is somewhere in between. Their install rates differ so much that a single blended rate tells you more about the proportions of each than about the listing. The console splits page views and installs by source, and the calculator should follow that split.
The uplift input changes job as well. Store listing experiments test screenshots, icons, short descriptions or preview videos against the current listing, each shown to a share of visitors. Like any A/B TestControlled comparison of two variants., the uplift worth entering is the effect the experiment is designed to detect, and the extra-conversions output becomes the installs that a winning variant would add for that source.
The inputs for an app store listing
| Calculator input | On the default page | For an app store listing | Where to find it |
|---|---|---|---|
| Conversions | Conversions on the site | First-time installs or first downloads from the product page | Store console acquisition report |
| Visitors / sessions | Website visitors or sessions | Product page views, unique where the console offers it, for one traffic source | Store console, split by search, browse and referral |
| Relative CVR improvement | Improvement to a web page | The uplift a store listing experiment is designed to detect | Store listing experiment plan |
The changed input is the visitor count, and two decisions shape it. First, choose the source: store search, browse or a referral or paid source, never all of them together. Second, choose the page view definition the console provides and keep it fixed; some consoles report total views and unique viewers, and switching between them changes the rate without any change in the listing. Pair it with installs from the same source and the same definition of a first install.
Entering the store inputs step by step
Export one source from the console
In the store console's acquisition report, filter to a single traffic source, such as store search, and a fixed date range. Note which page view definition the report uses.
Enter first-time installs and page views
Enter installs as conversions and product page views as visitors. Exclude redownloads and updates so that the count reflects people new to the app.
Set the uplift to the experiment's target effect
If a listing experiment is planned, enter the effect it is designed to detect. Read the extra installs as what the winning variant would add for this source over a comparable period.
Repeat for each other source
Run browse and referral traffic separately. Write the rates side by side so that changes in the mix stay visible.
A worked store listing with illustrative inputs
A meditation app plans a screenshot experiment
Quietgrove, an invented meditation app, looks at store search traffic for one month. The console shows 9,000 product page views from search and 2,700 first-time installs, a page-view-to-install rate of 30%. The team plans a screenshot experiment powered to detect a 10% relative lift. Entered into the calculator, that lift would add about 270 installs a month from search alone. Browse traffic, which converts at a much lower rate for this app, is run separately so its share of the month does not blur the search figure.
Conversion Rate Calculator
Open the Conversion Rate Calculator →Prefilled with the example's inputs: Conversions 2,700 · Visitors / sessions 9,000 · Relative CVR improvement 10% → Conversion rate 30% · Extra conversions from uplift 270.
The numbers belong to the invented app. Replace them with your own console export, one source at a time.
Interpreting the store conversion rate
Read the conversion rate output as a page-view-to-install rate for the source you entered. For store search, it describes how well the listing persuades people who were already looking for something like it. For browse, it describes how well the icon, title and first screenshots stop a scrolling visitor. Neither is a property of the app as a whole.
Expect search traffic to convert higher than browse. That means the blended rate moves whenever the mix moves: a featuring placement that floods the page with browse visitors lowers the blended rate even if the listing performed well. Reading the sources separately is how you tell a weaker listing from a different audience.
Extra installs from the uplift size the value of a listing experiment. If the figure is small relative to your install goals, the test may still be worth running for learning, but it will not change the month on its own. The app store optimisation guide covers which listing elements tend to matter for each source.
Keep an eye on the period, too. Store traffic is shaped by releases, seasonal interest and featuring, so compare a listing variant with the control over the same days, and compare one month with another only when the source split looks similar. A rate that rises straight after a release may reflect fresh reviews or a version note rather than the screenshots you changed.
For paid installs, the rate becomes an input elsewhere. Feed it into the CPACost of each conversion. calculator's conversion slot to estimate cost per install from a cost per tap or click, as the app install campaigns guide describes. The calculator cannot see what happens after the install, such as opening the app or subscribing, so keep a post-install measure beside the store rate.
Pitfalls when store numbers go into the calculator
- Blending store search and browse traffic. One rate for both hides whether the listing or the mix changed.
- Counting redownloads as new installs. Returning users inflate the conversion count and make the listing look more persuasive than it is.
- Comparing stores with different page view definitions. A gap between stores may come from how each counts views, not from the listings.
- Changing several listing elements in one test. A new icon, new screenshots and a new subtitle tested together leave you unable to say which one moved the CVRShare of visits or clicks that convert..
- Reading a featuring week as the new normal. Short bursts of browse traffic change the blend and fade.
