Which metrics matter most for an indie game?
Six numbers cover most decisions: installs by source, D1 retention, D7 retention, session length, the fail rate of your hardest early level, and crash-free sessions. Each one answers a different question. Installs tell you whether marketing works. D1 tells you whether the first session works. D7 tells you whether the game holds people. Session length tells you how deep a typical visit goes. The fail rate finds the wall. Crash-free sessions tell you whether the build itself pushes people out.
Small teams do not need fifty charts. They need a short list that they read every week and that they can act on. Add a metric only when you know which decision it changes.
| Group | Metric | Question it answers | Events it reads |
|---|---|---|---|
| Acquisition | Installs (first opens) by source | Which channels bring players? | game_open with is_first_open |
| First session | Tutorial and first level completion | Do new players reach the fun? | level_start, level_complete, custom steps |
| Retention | D1, D7, D30 | Do players come back? | game_open, session_start |
| Engagement | Session length, sessions per DAU, stickiness | How much do they play? | session_start, session_end |
| Stability | Crash-free sessions, errors per session by build | Does the build push players out? | error, app_unclean_exit |
Sources: GameAnalytics Benchmarks Hub; Manalens metrics documentation.
What is the formula for each game metric?
Each metric is one count divided by another count. The table gives the formula, a small worked example and the documentation that defines it. The example numbers are invented to show the arithmetic. They are not benchmarks and they are not data from a real game.
Check two things before you compare a number with another tool. The first is the day boundary: GameAnalytics and Manalens both use UTC days. The second is who counts as active: Unity counts a player in DAU when the player starts a session.
| Metric | Formula | Worked example (example numbers) | Defined by |
|---|---|---|---|
| Installs by source | First launches from one source ÷ all first launches | 120 ÷ 400 = 30% from creator links | Manalens install attribution |
| D1 retention | New players who play on day 1 ÷ players whose first launch was on day 0 | 40 ÷ 500 = 8% | GameAnalytics, Unity |
| D7 retention | New players who play on day 7 ÷ the same day 0 group | 9 ÷ 500 = 1.8% | GameAnalytics, Unity |
| D30 retention | New players who play on day 30 ÷ the same day 0 group | 2 ÷ 500 = 0.4% | GameAnalytics, Unity |
| Churn | 1 - retention for the same day | 1 - 8% = 92% on day 1 | Unity Manual |
| DAU | Players who start a session on one calendar day | 300 players on Saturday | Unity |
| MAU | Players who start a session in the last 30 days | 5,000 players | Unity, GameAnalytics |
| Stickiness | DAU ÷ MAU | 300 ÷ 5,000 = 6% | Unity, GameAnalytics |
| Sessions per DAU | Sessions on a day ÷ DAU | 510 ÷ 300 = 1.7 | Unity |
| Session length | Total session time ÷ sessions | 10,200 minutes ÷ 510 = 20 minutes | Unity, Manalens |
| Playtime per DAU | Total session time on a day ÷ DAU | 10,200 minutes ÷ 300 = 34 minutes | GameAnalytics Benchmarks Hub |
| Level fail rate | Fails ÷ (completes + fails) | 45 ÷ (105 + 45) = 30% | Manalens |
| Funnel conversion | Players who reach the last step ÷ players who enter the first step | 540 ÷ 1,000 = 54% | Manalens |
| Crash-free sessions | 1 - crashed sessions ÷ all sessions | 1 - 30 ÷ 2,000 = 98.5% | Firebase Crashlytics, Manalens |
| Refund rate | Units returned in a period ÷ units purchased in that period | 12 ÷ 200 = 6% | Steamworks |
| ARPDAU | Revenue on a day ÷ DAU | $45 ÷ 300 = $0.15 | Unity |
| ARPPU | Revenue in a period ÷ players who paid in that period | $450 ÷ 50 = $9.00 | GameAnalytics |
| ARPU | Revenue in a period ÷ active players in that period | $450 ÷ 5,000 = $0.09 | GameAnalytics |
| Paying share | Players who paid on a day ÷ DAU | 6 ÷ 300 = 2% | GameAnalytics, Unity |
| LTV per player | Up-front net revenue + ARPDAU × expected days played | $7.00 + $0.15 × 4 days = $7.60 | Our Player LTV calculator |
Sources: GameAnalytics Metrics documentation; Unity Data Explorer V2, Analytics dashboards and Manual; Firebase crash-free metrics documentation; Steamworks Developer Refund Reporting; Manalens metrics documentation.
How do I measure acquisition for a PC game?
Count first launches, not downloads, and split them by source. A download or a key activation does not prove that anyone played. A first game_open does. The Godot and Unity SDKs send game_open with is_first_open on the first launch, so an install is a real start of the game.
Before launch, wishlists are the acquisition metric on Steam. Tag every link to your store page and read the Steamworks UTM report, which Manalens can import. After launch, install attribution links a first launch to a recent website visit from the same network, and the Attribution page shows it on the Installs tab. Treat it as a strong hint, not a perfect join: a player who goes straight from a video to Steam has no website visit to match.
Sources: Manalens Steam and install attribution documentation.
Which retention numbers should I watch?
Watch D1 and D7 first, and D30 only when the game has enough content to fill a month. D1 is the share of new players who start a session on the day after their first launch. D7 and D30 use the same rule for day 7 and day 30. These are strict day numbers: a player who returns on day 6 but not on day 7 does not count for D7. GameAnalytics documents the same strict rule, and Unity calls it N-Day Retention.
PC numbers are much lower than the mobile numbers that most blog posts quote. In the GameAnalytics Benchmarks Hub, the median PC game had 6.54% D1, 1.44% D7 and 0.19% D30 in October 2026. The median mobile game had 20.78% D1 in the same table. Compare your game with the PC rows, and then with your own earlier builds.
- D1, D7 and D30 retention benchmarks for PC and indie games →
- Glossary: D1 retention →
- Glossary: D7 retention →
- Glossary: D30 retention →
- Glossary: Churn →
- D1, D7, D30 retention calculator →
| PC metric (desktop builds, October 2026) | P25 | P50 (median) | P75 | P95 |
|---|---|---|---|---|
| D1 retention | 4.23% | 6.54% | 12.24% | 32.38% |
| D7 retention | 0.83% | 1.44% | 3.05% | 10.46% |
| D30 retention | 0.00% | 0.19% | 0.75% | 3.85% |
| Session length (minutes) | 11.60 | 20.82 | 33.83 | 71.79 |
| Playtime per DAU (minutes) | 17.77 | 37.73 | 69.91 | 159.23 |
Sources: GameAnalytics Benchmarks Hub and Metrics documentation; Unity Manual; Manalens metrics documentation.
How do I measure engagement?
Use session length, sessions per daily active player, and stickiness. Session length is the average duration_s of session_end events. Sessions per DAU is the number of game sessions per player per active day. Stickiness is DAU divided by MAU, the share of your monthly players who play on a typical day.
PC sessions are long and few. The median PC game in the GameAnalytics table had a session length of 20.82 minutes, 1.66 sessions per DAU, 37.73 minutes of playtime per daily active user and a stickiness of 4.32%. A premium game with a story often has long sessions and low stickiness, and that can be healthy. Read engagement next to retention, not alone.
A session ends after 30 minutes without focus or events, so a player who leaves the game open in the background does not inflate the number forever. Steam shows concurrent players (CCU), which is a different number from DAU. Manalens has no CCU chart. To turn a public peak CCU into a DAU range, use the estimator below.
- Glossary: DAU →
- Glossary: MAU →
- Glossary: DAU/MAU stickiness →
- Glossary: Sessions per day →
- Glossary: Session duration →
- Glossary: Playtime →
- Glossary: CCU →
- CCU to DAU and MAU estimator →
Sources: GameAnalytics Benchmarks Hub; Unity Data Explorer V2 documentation; Manalens Godot SDK and metrics documentation.
How do I find where players struggle?
Send level_start, level_complete and level_fail with the level number, then read the fail rate per level. The fail rate is fails divided by completes plus fails. A level with a fail rate far above its neighbours is a difficulty spike, and difficulty spikes are a common place for players to stop.
Next, build a funnel from the first launch to the end of the first real goal. The Level funnel in the Game KPIs template goes from game_open to level_complete for levels 1 to 4, and you can edit its steps. The step with the largest drop is your first fix.
- First session funnel and FTUE analysis →
- Glossary: Funnel →
- Glossary: Drop-off →
- Glossary: Completion rate →
Sources: Manalens metrics and funnels documentation.
Which stability metrics should I track?
Track crash-free sessions and errors per session by build. A crash in the first hour costs more on PC than on most platforms, because Steam refunds a game bought in the last 14 days with less than two hours of playtime. A player who crashes twice in the tutorial can ask for the money back.
Crash-free sessions is 1 minus the share of sessions that crashed. Firebase Crashlytics uses the same formula and says the metric matters most early in a new release, when a crash in the first session can make a user leave. In Manalens, a crashed session is one with app_unclean_exit or an unhandled critical error. The Godot and Unity SDKs send app_unclean_exit on the next launch after a crash. The Game health template adds crash-free players, errored sessions and the top errors.
Read the refund rate next to it. Steamworks defines it as the units returned in a period divided by the units purchased in that period, and shows the reasons that players gave.
Sources: Steam Refunds policy; Steamworks Developer Refund Reporting; Firebase crash-free metrics documentation; Manalens error tracking documentation.
Which of these metrics does Manalens show?
Most of them, from two board templates and the automatic Steam sync (in early access). The table lists each metric and where it is. A "No" means that no built-in card or chart shows it today.
| Metric | In Manalens | Where |
|---|---|---|
| Installs (first opens) by source | Yes | Attribution page, Installs tab |
| D1, D7, D30 retention | Yes | Game KPIs board |
| Cohort retention curve | Yes | Install cohorts chart, or any Retention chart |
| DAU, MAU, stickiness | Yes | Game KPIs board |
| Sessions per DAU, session length | Yes | Game KPIs board |
| Playtime per DAU | No card | Write a SQL chart on session_end |
| Level fail rate, level funnel | Yes | Levels table and Level funnel, Game KPIs board |
| Crash-free sessions and players, errors by build | Yes | Game health board |
| ARPDAU, ARPPU | Yes, in one currency | Game KPIs board |
| ARPU, LTV | No card | Write a SQL chart, or use the Player LTV calculator |
| Steam wishlists, units sold and returned | Early access, with a Steamworks Financial API key | Daily Steam sync |
| Refund rate | Not as a rate | Divide units returned by units sold |
| CCU | No | Read it on Steam |
Sources: Manalens dashboards, metrics, error tracking, Steam and install attribution documentation; the Game KPIs and Game health board templates.
How often should I review game metrics?
Once a week, at the same time, with the same table. Daily numbers move too much for a small game. A weekly review with written notes shows trends and keeps you honest about what changed.
- Note last week's builds, patches, sales and marketing beats.
- Read installs by source and the D1 of last week's new players.
- Read D7 for the players who started two weeks ago.
- Check the level fail rates and the first session funnel for a new spike.
- Check crash-free sessions by build.
- Write one decision for next week and one metric that will show whether it worked.
Sources: Manalens retention guide.
How do I get these metrics without building them myself?
Manalens is the analytics and growth engine built for games. Install the Unity or Godot SDK, create the Game KPIs board and the Game health board, and read installs, D1, D7, session length, level fail rates and crash-free sessions in one project.
Sources: Manalens product documentation.
Frequently asked questions
What is the single most important metric for an indie game?
Before launch, Steam wishlists. After launch, D1 retention of new players, because it shows whether the first session works. Read it next to installs by source, so you know whether a drop comes from the game or from a change in the players that marketing brought in.
Are mobile retention benchmarks useful for a PC game?
Not as targets. In the GameAnalytics Benchmarks Hub for October 2026, median D1 was 6.54% for PC and 20.78% for mobile. The samples also differ: PC games enter at 3,000 monthly active users and mobile games at 10,000. Compare a PC game with PC data and with itself.
What is a good session length for a PC game?
It depends on the genre. The median PC game in the GameAnalytics Benchmarks Hub had sessions of 20.82 minutes in October 2026, and the top quarter had 33.83 minutes or more. Use the number to spot changes between builds, not to judge a short arcade game against a long strategy game.
What is the difference between ARPU, ARPDAU and ARPPU?
The number they divide revenue by. ARPU divides by all active players in a period. ARPDAU divides the revenue of one day by the players of that day. ARPPU divides by the players who paid. With $450 from 50 payers among 5,000 active players, ARPU is $0.09 and ARPPU is $9.00. These are example numbers.
Do I need a separate tool for crashes?
Not always. The Godot and Unity SDKs send errors and app_unclean_exit events, and the Game health board shows crash-free sessions, errors by build and the top errors. A dedicated crash reporter adds stack symbols and native crash dumps, which large teams may still need.
Sources and implementation
- GameAnalytics Benchmarks Hub: PC and mobile benchmarks (October 2026 data, read 19 November 2026) ↗
- GameAnalytics documentation: Metrics (read 19 November 2026) ↗
- Unity documentation: Data Explorer V2 measures and metrics (read 19 November 2026) ↗
- Unity documentation: Analytics dashboards (read 19 November 2026) ↗
- Unity Manual 2022.2: Analytics Metrics, Segments, and Terminology (read 19 November 2026) ↗
- Firebase documentation: Understand crash-free metrics (Google, last updated 17 November 2026, read 19 November 2026) ↗
- Steam Refunds policy (Valve, last updated 23 April 2024, read 19 November 2026) ↗
- Steamworks documentation: Developer Refund Reporting (Valve, read 19 November 2026) ↗