How to read Google Play install counts
Google Play shows a rounded install band, but its listing data carries an exact lifetime count. Here is how to turn that counter into daily and weekly installs, and when to doubt it.

On this page
Google Play shows the public a rounded install band such as 10M+, but the listing data behind it carries an exact lifetime install count. Subtract an earlier reading of that count from a later one and you get the installs in between. Divide by the days between the two readings and you can compare any two apps, whatever their collection schedule.
Key takeaways
- The public band only moves at thresholds of 1 and 5 times a power of ten, so an app can grow up to five times without its label changing.
- The exact count is lifetime, global and cumulative. It counts installs, not active users, and Google does not document how it is defined.
- Installs in a period equal the later reading minus the earlier one. Divide by the days between readings before you compare apps.
- Counts jump for real reasons and should almost never fall. Investigate jumps and treat small dips as noise.
- Across a data gap the total stays exact, but the daily shape is lost.
The two install numbers on every listing
The public install band
Every Google Play listing shows an install band, labelled as downloads in the store, that follows a fixed ladder: 1+, 5+, 10+, 50+, 100+, 500+, 1K+, 5K+ and so on, up to 1B+, 5B+ and 10B+. Each power of ten has two steps, one at 1 and one at 5. The label tells you the app has passed a threshold and nothing more.
| Band | Exact counts it covers | Growth that stays hidden |
|---|---|---|
| 100K+ | 100,000 to 499,999 | Up to 5x |
| 500K+ | 500,000 to 999,999 | Up to 2x |
| 1M+ | 1,000,000 to 4,999,999 | Up to 5x |
| 5M+ | 5,000,000 to 9,999,999 | Up to 2x |
| 10M+ | 10,000,000 to 49,999,999 | Up to 5x |
An app that grows from 1.1 million to 4.9 million installs shows 1M+ the whole time. Bands are fine for rough sizing and useless for measuring growth.
The exact lifetime count
The listing data Google Play serves also contains an unrounded number, such as 3,482,917 rather than 1M+. Open-source scrapers expose it, but Google does not document it. Watched over time, it behaves like a lifetime counter:
- Lifetime. It accumulates from launch and does not shrink when people uninstall.
- Global. It is one number for all countries and cannot be split by country.
- Installs, not users. It counts installs, not people using the app today. Whether it treats users, devices and reinstalls the way Play Console does is not documented.
Because the definition is undocumented, compare the count with itself over time and with other apps' counts, and take care when you put it beside your own Play Console figures. AppMetis labels these numbers measured installs because they are read from the store, not modelled.
Turn a cumulative count into installs per period
The core operation is differencing. Installs between two readings equal the later count minus the earlier count. If an app read 3,482,917 on Monday and 3,489,517 on Tuesday, it added 6,600 installs in between.
From there:
- Daily installs come from consecutive daily readings.
- Weekly installs are the difference between readings seven days apart, which equals the sum of the daily differences in between.
- Monthly installs are the last reading of the month minus the last reading of the month before.
Readings are not taken at the same minute every day, so a daily difference can cover a little more or less than 24 hours. A seven-day rolling average removes most of that jitter.
Normalise uneven collection cadence
Not every app is read every day. AppMetis collects the count at each app's own cadence: daily for about 250K apps, weekly for about 580K and monthly for about 1.4M, with a full catalog pass every 14 days. Intervals also stretch when a reading is missed. Before you compare anything:
- Count the days between the two actual reading dates, not the nominal schedule.
- Divide the install difference by those days to get installs per day.
- Compare installs per day, or scale them back to a common period such as a week.
- When an interval straddles two months, split it by days if you need monthly totals.
If you need a daily line for an app read monthly, spreading the difference evenly across the days is an assumption, not a measurement. Label it as interpolated.
Why counts jump and almost never fall
Jumps
A sudden step up always has a cause worth knowing:
- Real demand spikes from featuring, a viral moment, press or a large ad campaign. These usually leave a higher plateau afterwards and bring more ratings with them.
- Pre-install and device-setup deals, where a phone maker or carrier puts the app on new devices. Installs arrive in large blocks with few ratings. Google does not document exactly which of these the counter includes.
- Corrections on Google's side that move the counter in one step.
- Long intervals. A monthly reading looks like a jump if you forget to divide by days.
Dips
The count is cumulative, so in principle it only rises. Small dips do appear now and then, usually from a correction or a stale read. Do not record negative installs. Carry the previous highest value forward and flag the interval. A large fall means something changed with the listing itself, so treat that interval as unreliable until you know why.
How to treat a data gap
AppMetis has no data from 9 September to 7 November 2026. We show that period as a gap and never fill it silently. Because the count is cumulative, the gap costs less than it seems:
- The total is exact. The first reading after the gap minus the last reading before it gives the precise number of installs during the gap.
- The shape is lost. You cannot tell whether those installs arrived evenly or in one spike in October.
- An average is only an average. Dividing the gap total by its days gives average installs per day. Do not draw it as a daily series or call it measured daily data.
- Growth rates inside the gap do not exist. Compare whole periods on either side of it instead.
iOS chart ranks are missing for the same dates, so chart cross-checks are unavailable inside the gap too. The gap in our history explains what happened.
Installs, downloads and active users
These three numbers get mixed up constantly:
- Installs, the Google Play counter, measure lifetime reach across all countries.
- Downloads are what App Store Connect reports for your own apps, split into first-time downloads and redownloads by day and country. Apple publishes no public count for any app, and Google Play labels its public band as downloads, which adds to the confusion.
- Active users count people who opened the app in a day or a month. Only the owner's analytics or console data shows daily active users and their monthly equivalent.
A lifetime count of 10 million says nothing about whether 100,000 or 3 million people still use the app.
Sanity checks before you trust a number
- Installs per new rating. Divide the install difference by the change in rating count over the same interval, then compare with the app's own history and with similar apps. If installs jump tenfold while ratings barely move, suspect pre-installs or a correction. If ratings jump while installs do not, the app probably changed its rating prompt.
- Chart positions. A genuine surge usually shows up in the store's top free charts at the same time. For apps on both stores, the App Store version's chart movement is a useful second opinion.
- Band crossings. When the exact count passes a threshold such as 5,000,000, the public band should change soon after.
- Owner statements. Some developers announce install milestones. Check that the count agrees.
A worked example
The numbers below are hypothetical. Picture a budgeting app read weekly, with one missed reading:
| Reading date | Lifetime count | Days | Installs | Per day |
|---|---|---|---|---|
| 4 Aug 2026 | 1,840,000 | |||
| 11 Aug 2026 | 1,882,000 | 7 | 42,000 | 6,000 |
| 18 Aug 2026 | 1,931,000 | 7 | 49,000 | 7,000 |
| 1 Sep 2026 | 2,029,000 | 14 | 98,000 | 7,000 |
| 8 Sep 2026 | 2,239,000 | 7 | 210,000 | 30,000 |
| 8 Nov 2026 | 2,666,000 | 61 | 427,000 | 7,000 on average |
How to read it:
- The 1 September row covers 14 days because a weekly reading was missed. Its 98,000 installs look like a doubling, but the pace per day is unchanged at 7,000.
- The week to 8 September is a real jump, more than four times the normal pace. If new ratings rose by a similar multiple, it is probably genuine demand such as featuring. If ratings barely moved, look for a pre-install deal or a correction.
- The 8 November row spans the gap. The 427,000 installs over 61 days are exact, the 7,000 a day is only an average, and a second spike in October would be invisible.
- The public band showed 1M+ for the whole period, while the app added 826,000 installs.
How AppMetis helps
AppMetis reads the exact count for about 2.25 million Google Play apps and turns it into measured installs per day, week and month at each app's own cadence, with the September to November 2026 gap shown as a gap. Explore any app's series with measured installs, see which apps are adding the most installs a day on trending Google Play apps, and read how we collect and label the data in our methodology.
Next steps
- Pick three competitors and convert their counts into installs per day for the periods before and after the gap.
- Put your own Play Console installs beside your public count to learn how the two relate for your app.
- Note a likely cause for every jump, and check it against ratings and charts before you report it.
- When a period crosses the 2026 gap, report one total for the gap instead of daily figures.
