How many installs is one rating worth? Data from 252,554 Google Play apps, and a Python function

If you've ever tried to guess how many downloads a rival app has, you've probably done this: take its rating count and multiply by some number someone posted on a forum. 50? 100? 200?

We measured it instead. Google Play is the only store that shows install counts, so on Google Play the ratio between installs and ratings can be measured directly, not modeled. We took a snapshot of 252,554 Google Play apps on August 3, 2026 (apps with at least 1,000 installs and 100 ratings) and divided lifetime installs by lifetime ratings.

Short answer: the median app gets one rating per ~155 installs. But a single multiplier is the wrong tool. Category alone moves the ratio about 6×.

The ratio by category

The middle half of all apps sits between 73 and 318 installs per rating. Split by category, the extremes look like this:

Category	Median installs per rating
Educational (games)	472
Racing	361
Simulation	246
Photography	237
Tools	199
Productivity	155
Education	150
Health & Fitness	123
Puzzle	122
Finance	89
Casino	76

Kids' games barely get rated: the players are children. Apps where money is on the line (Casino, Finance, Card games) get rated constantly.

Three things move the ratio more than you'd think

1. Paid vs. free. Paid apps collect a rating about every 32 installs. Free, ad-supported apps need about 198. People who paid upfront rate roughly six times more often per install.
2. App size. Under 10,000 installs, the median is one rating per just 30 installs. Early ratings are heavily solicited: friends, launch communities, a prompt on first open. The ratio only settles near ~200 for apps between 100K and 10M installs.
3. Star rating. Apps rated 4.5★ and up earn a rating every ~92 installs; apps under 3.5★ every ~270. The likely reason is the in-app rating prompt: apps that ask at a good moment get more ratings and better ones.

A function you can actually use

Here is a small Python function that turns a rating count into an install range. It uses the category medians above, then adjusts for paid apps and small apps, because those two effects are the biggest.

# Median installs per rating, Google Play, snapshot of Aug 3, 2026.
# Source: GetAppNiche Downloads-per-Rating Study 2026 (252,554 apps).
INSTALLS_PER_RATING = {
    "educational_game": 472, "racing": 361, "simulation": 246,
    "photography": 237, "tools": 199, "productivity": 155,
    "education": 150, "health_fitness": 123, "puzzle": 122,
    "finance": 89, "casino": 76,
}
STORE_MEDIAN = 155


def estimate_installs(ratings: int, category: str | None = None,
                      paid: bool = False) -> tuple[int, int]:
    """Return a (low, high) install estimate from a rating count.

    It's a range on purpose: even inside one category, apps differ a lot.
    """
    ratio = INSTALLS_PER_RATING.get(category or "", STORE_MEDIAN)
    if paid:
        ratio = 32            # paid apps rate ~6x more often
    elif ratings < 300:
        ratio = 30            # tiny apps: early ratings are solicited
    low, high = ratio * 0.5, ratio * 2.0   # roughly the middle half of apps
    return int(ratings * low), int(ratings * high)


print(estimate_installs(12_000, "productivity"))   # (930000, 3720000)
print(estimate_installs(12_000, "casino"))         # (456000, 1824000)
print(estimate_installs(800, paid=True))           # (12800, 51200)

Same 12,000 ratings, and the Productivity app is likely about twice the size of the Casino app. A flat "× 100" would get both wrong.

A few honest limits:

• This is lifetime installs over lifetime ratings. It says nothing about this month's downloads.
• Google Play shows install counts as brackets ("1M+"), so the study uses the bracket floor. Real installs are a bit higher.
• On the App Store there are no public install counts at all, so this exact method doesn't transfer. App Store users also rate at different rates.

Where the full data is

All 47 categories, the size and star breakdowns and the method are in the Downloads-per-Rating Study 2026. The full table is also available as an Excel file there.

If you'd rather skip the math, GetAppNiche shows monthly download and revenue estimates for 3.6M+ App Store and Google Play apps, and you can query it from code through the REST API or from Claude, ChatGPT or Cursor over MCP.

What multiplier have you been using? I'm curious how far off the common rules of thumb are in your category.