App Store Conversion Rate, Defined by Apple

A workout logger can have a busy search result and a quiet install button. The number that describes that gap has a definition, and it is not “whatever percentage felt good in a Twitter thread.” Apple publishes the arithmetic. Use that arithmetic. If you divide downloads by product page views and call the result Apple’s conversion rate, you are tracking a cousin of the metric, and you will argue with App Analytics every Monday.

The definition is on Metric definitions : total downloads and pre-orders divided by unique device impressions. A pre-order counts. The later download of that pre-order does not count again. How impressions and page views differ is in how to read App Store Connect acquisition. The frames that usually move the rate are covered in the first three screenshots and product page optimization.

The formula, with nothing added

Numerator: total downloads and pre-orders. Denominator: unique device impressions. Impressions are times the app was viewed on Today, Games, Apps, and Search for more than one second, and that count includes product page views. Unique device impressions count devices, not every repeat glance. Conversion uses the unique-device figure, so one phone that sees you five times in a day does not inflate the denominator five times.

Pre-orders are in the numerator on purpose. If you are taking pre-orders for a new logger, those orders count toward conversion. When the app later lands on the device, Apple does not count that download again for this rate. If you add pre-orders and first-time downloads yourself from two exports and forget that rule, you will double-count the same person and declare a miracle.

Why your spreadsheet disagrees with the console

The homemade number usually divides installs by product page views, sometimes by unique page views, sometimes by “sessions” from a different tool. Apple’s denominator is unique device impressions, and impressions include views that never opened the product page: a search row that stayed on screen for more than a second, a Today card, a Games or Apps placement. A logger that is seen and not tapped has a conversion event in Apple’s math and no page view at all.

Apple divides downloads and pre-orders by unique device impressions. A ratio built from product page views answers a narrower question and will not match the console.

Keep the homemade ratio if you want it, and call it page-view conversion. Use it when people already opened the page and you want to know whether the page closed. Use Apple’s rate for the path from “a device saw the app” to a download or a pre-order. Mixing the names is how two different fractions get described as the same conversion.

There is no official “good” percentage

Apple does not publish a target. A sentence like “healthy apps convert at 30%” is not from the metric definition, and this post will not invent a replacement target. App Analytics is where you compare yourself: your rate over time, and your rate by source type, such as App Store search versus App Store browse versus referrals. Peer group benchmarks live in that analytics area too. Read your comparison there. Do not paste someone else’s percentile into a plan.

The metrics are available once the app has at least five first-time downloads or pre-orders. A brand-new logger will show a noisy rate for a while because the denominator is small. A single featuring day, or a single bad screenshot test, can swing it. Judge a change across a stretch of ordinary days, split by source, not across the afternoon you emailed your friends.

What actually moves it on a workout logger

Four things on the result and the page do most of the work, and none of them is a secret rank coefficient. The icon is the object in the row. The name and subtitle are the only words in that row besides the rating. The first screenshots are what search shows when no preview is playing, and they are what the page leads with. The star average sits in the same row. Apple does not publish a ranking weight for stars, but a low average still changes the decision to download, and that decision is this rate.

For the logger, “Track workouts” as a subtitle is weaker than the job a tired person typed: log sets, rest timer, history. The first screenshot should be a filled workout, not an empty chart and a logo. If ratings mention a subscription that starts during the first set, the page has to say that before the tap, or the rate you “gain” from a prettier icon comes back as refunds and 1-star notes. Promotional text, 170 characters, can change without a version and does not affect ranking. It can announce a program. It is not the denominator.

• Icon: recognizable at row size. A dumbbell the size of a speck does not convert.
• Subtitle: the job, inside 30 characters, not a slogan.
• First screenshots: one completed set, the rest timer, the history. Same promise as the subtitle.
• Ratings: read the distribution. A low average is a conversion objection, not a published rank weight.

Where to look when the rate falls

Split the fall before you redesign everything. If impressions are high and product page views are low, the row is the problem: icon, name, subtitle, and the stars beside them. People saw the logger and did not open it. If product page views are high and downloads are low, the page is the problem: screenshots, preview, description’s first lines, and the rating text. If both are fine and the rate still looks ugly, check whether you are comparing a search-heavy week to a browse-heavy week. Those sources convert differently, and Apple lets you see that split.

Metadata other than promotional text ships with a version. A subtitle test waits on a binary. Screenshot tests on the default page wait on a version too, unless you are running product page optimization, which is a separate test of icons, screenshots, and previews. Do not “fix conversion” by stuffing the keyword field. Keywords change who shows up. Conversion is what happens after a device has already seen you.

A single-device walkthrough, not a benchmark

One phone, one day, labeled as an illustration of the definitions, not as this logger’s analytics export. The app stays on screen in search for more than a second. That is an impression. The person opens the product page. Impressions include product page views, so that is another impression, and it is one product page view. Unique device impressions for that phone that day are still one. If they download, Apple’s rate for that device is a download over one unique impression. If they only saw the search row and left, the unique impression is in the denominator and the numerator is empty. Your page-view spreadsheet never recorded them.