FunnelHound / Guides / Metrics explained
App Store Connect metrics explained: every App Analytics number in one table
App Store Connect reports about twenty metrics across two sections, and they fall into three families: what happened in the store (impressions, page views, downloads), what happened on devices (installs, deletions, sessions, crashes), and what happened to money (units, proceeds, paying users). The store family counts everyone. The other two count only users who opted in to share data, which is the single fact that explains most "the numbers don't add up" questions.
The table
| Metric | Counts | Population | The trap |
|---|---|---|---|
| Impressions | Times your icon was on screen for at least a second in search, browse, Today or a product page's "you might also like" | All devices | Reported as total and as unique devices; the dashboard's conversion rate uses unique |
| Product page views | Opens of your product page, any source | All devices | Installs from search results skip the page, so page views can be lower than downloads |
| Conversion rate | Total downloads ÷ unique device impressions | All devices | Not page views, not purchases. Compare only against impression-based benchmarks |
| Total downloads | First-time downloads plus redownloads | All devices | Includes people re-installing on a new phone; first-time downloads is the growth number |
| First-time downloads | A device's first download of the app under that Apple account | All devices | The closest thing to "new users" Apple gives you |
| Redownloads | Downloads by accounts that had the app before | All devices | Spikes after a feature or a viral moment are often old users returning |
| Updates | Manual and automatic updates installed | All devices | Reflects install base, not interest; a big number means many existing users |
| Pre-orders | Pre-order placements and cancellations before release | All devices | Converted pre-orders appear as downloads on release day, not before |
| Installations | The app installed on a device, including restores and family sharing | Opt-in only | Different population from downloads; never divide one by the other |
| Deletions | The app removed from a device | Opt-in only | Undercounts by the opt-in share; use the ratio to installations, which shares the population |
| Sessions | Foreground use of at least two seconds | Opt-in only | Short widget or notification interactions do not count |
| Active devices | Devices with at least one session in the period | Opt-in only | Devices, not people; one person with an iPhone and an iPad is two |
| Active last 30 days | Rolling 30-day active devices | Opt-in only | Rolling window, so a daily chart moves slowly by design |
| Crashes | Crash reports collected | Opt-in only | A rate per session is the useful form; the raw count follows the install base |
| Paying users | Users who completed at least one paid transaction in the period | Opt-in only (App Analytics) | For true totals use Sales and Trends, which counts every transaction |
| Proceeds | Your share after Apple's commission and taxes | Opt-in in Analytics, complete in Sales and Trends | Analytics proceeds will be lower than your payout; that is the opt-in gap, not a fee |
| Sales | Customer-paid amounts before Apple's cut | Same split as proceeds | The number customers see; proceeds is the number you see |
| In-app purchases | Count of in-app purchase transactions | Opt-in in Analytics, complete in Sales and Trends | Subscriptions renewals count each period |
| Units, App units | Transactions in Sales and Trends, by product type | Complete | Free downloads, paid purchases and in-app purchases are all units; filter by product type |
| Subscriptions | Active, new, churned, trial, by product and territory | Complete | The Subscriptions report, not App Analytics, is the one with trial-to-paid conversion |
Downloads, units, installations: three numbers for "how many got it"
Total downloads is the store's view: how many times the app was fetched. Units is the ledger's view: how many transactions happened, where a free download and a $4.99 subscription are each one unit of their type. Installations is the device's view, but only from devices that opted in. Downloads and units can be reconciled, since both see everyone. Installations can only be compared with its siblings, deletions and sessions, which share its population. The units guide works through the reconciliation with real rows.
Which numbers are ratios of which
- Tap-through rate: product page views ÷ impressions. Both store family. Safe.
- Conversion rate (Apple's): total downloads ÷ unique impressions. Both store family. Safe, and the benchmarks view compares it with your category's percentiles.
- Page conversion: downloads ÷ page views. Both store family. Can exceed 100% because of search installs.
- Deletion rate: deletions ÷ installations. Both opt-in. Safe as a ratio, undercounted as absolutes.
- Purchase rate: paying users ÷ total downloads. Mixed populations. Directionally useful, numerically low; the purchase conversion guide explains how to read it anyway.
- Deletions ÷ downloads: mixed populations. Do not. This one produces the "a third of my users delete on day one" panic that is really a 30% opt-in share.
Two sections, two refresh rates
App Analytics (Analytics → Metrics) updates once a day with a two to three day lag and holds the store and device families plus the opt-in commerce numbers. Sales and Trends updates daily by the next morning and holds the complete ledger: units, sales, proceeds, subscriptions. Payments and Financial Reports, the third section, is monthly and is what actually reaches your bank. Three dates for one sale is normal.
Where FunnelHound reads each
The app takes the store family from the Analytics Reports API (impressions, page views, downloads), deletions and installations from the same source with the opt-in caveat shown, and the money from the Sales and Trends reports so proceeds match your payout rather than the opt-in undercount. The funnel pills are built only from same-population pairs, which is a long way of saying the arithmetic above is baked in.
Twenty metrics, one funnel
FunnelHound reads the store, device and money families from App Store Connect and draws the four stages that matter, impressions to purchases, on your iPhone, with every ratio computed from matching populations.
Get FunnelHoundData notes: definitions follow Apple's App Store Connect Help, "View metrics and benchmarks" and the Sales and Trends glossary, as of 2026. Opt-in share varies by app and region; a quarter to a third is a common range, not a rule. The 1,000 / 1,180 / 310 example is illustrative. Apple renames metrics occasionally; "App Units" in older documentation corresponds to today's Units filtered to the app product type.