# Install figures and their limits

[`/apps/metrics/installs/history`](https://appmanai.com/docs/reference/operations/appMetricsInstallsHistory.html) (and the MCP
tool `app_metrics_installs_history`) returns an app's total installs worldwide, at most one point
per day. **Google Play's number is a count; the App Store's is an estimate.** Read this before
quoting either.

## What you get

| Field | Meaning |
| --- | --- |
| `refreshed_at` | The day the figure is for. Points are in date order; days without data are absent. |
| `total` | Installs to date, worldwide. |
| `total_bucket` | `total` as a range, e.g. `500,000,000+`. On Google Play this is the range the store page itself shows; on the App Store it is the range of our estimate. |
| `daily` | The gain per day since the previous point. `null` on a first point with no earlier data and on a day the total fell. |

`from` and `to` (`YYYY-MM-DD`) pick the window. Without `from` it is the 30 days ending at
`to`; without `to` it ends today. `to` cannot be before `from`.

## Google Play: an exact count

Google keeps an exact install count for every app but shows only a rounded range on the store
page. `total` is the exact number and `total_bucket` is the range the store shows — Discord's
store page says `500,000,000+`, its `total` is 693,524,399.

## App Store: an estimate

Apple publishes no install numbers, so every App Store figure is our estimate, not a count from
Apple. Its `total_bucket` is a range taken from the estimate; the App Store itself shows none.
Keep these limits in mind:

- **Approximate for any single app.** An individual app can read noticeably high or low. The
  figure is most reliable for following one app over time and for comparing apps of the same
  kind — don't present it as an exact install count.
- **Not every country is included.** Countries without enough usable data are left out, so an
  app that sells mainly in those markets reads low.
- **Gains can arrive in steps.** The underlying store data refreshes unevenly, so growth can land
  on one day instead of spreading smoothly across several.
- **It can fall.** The estimate is sometimes revised down; `daily` is `null` on that day.
- **A day behind.** The estimate is computed once a day, so its latest point can trail Google
  Play's by a day.

## Worldwide only

There is no per-country install figure. Google publishes one number for the whole world, so both
stores report worldwide — a chart never mixes a worldwide Google Play figure with a single App
Store country.

## Reading `daily`

`daily` is the gain per day since the previous point: the change in `total` divided by the
number of days between the two. For consecutive days that is simply the difference; across a gap
it is the average over the missing days. Because it is a rate, adding up `daily` undercounts any
gap — for the growth over a window, subtract the first `total` from the last.

The first point of a window still has a `daily` when earlier data exists. It is `null` on a first
point with nothing before it, and on any day the total fell.