GameAnalytics Benchmarks Hub
A metric on its own tells you very little. Retention of 18%, or a three-minute session, only means something once you can see where the rest of the market sits and how far above or below it you are. This hub gives you that reference point for both mobile and PC games, across broad percentile range, refreshed every month.
Choose a platform to see its benchmarks:
How to read percentiles
Benchmarks here are broken into percentiles. A percentile tells you where a value sits relative to the rest of the sample, not what share of games hit that exact number.
P50 is the median, the midpoint of the sample. Half the games fall below it and half above. This is usually the most representative figure to compare your own game against.
P25 and P75 mark the boundaries of the middle half. A game below the median is not unusual, a significant part of the sample sits there. A game between P50 and P75 is performing above average even if it is not yet in the top quartile.
P90 and P95 describe increasingly rare, top-tier performance. A game at P95 is outperforming 95% of the sample. These are useful for understanding what is possible at the extreme end, but they are not realistic targets for most games and should not be read as the standard to design toward.
Mobile game benchmarks
Every figure below comes from mobile games on the GameAnalytics platform, worldwide, with no regional filter applied. A game enters the sample once it passes 10,000 monthly active users (MAU), which works out to roughly 5,500 titles in a typical month, and the sample is rebuilt monthly rather than held fixed, so it tracks the live market rather than a snapshot. Figures are shown as percentiles rather than averages, because an average hides the spread: the 25th marks the bottom quarter, the 50th is the median, and the 95th is the top twentieth. Games below the 10,000 threshold are not represented and tend to behave differently, so read these as a reference for established titles.
Compare your mobile game to the industry
Copy this prompt into Claude, ChatGPT, Cursor, or any AI tool to see how your game stacks up against these benchmarks.
Compare my mobile game's performance against GameAnalytics industry benchmarks.
STEP 1: FIND MY DATA
Before asking me, check whether my game data is already available to you. Look in this order:
- The GameAnalytics MCP server, if you have it connected
- Any other connected source, such as a spreadsheet, a cloud drive, a database or files in the current project or repository
- Earlier in this conversation, or in our recent conversations if you can search them
If you find data, tell me where you found it and which month it covers before you use it, so I can confirm it is the right source.
If you find nothing, stop and tell me so. Ask me to paste the figures below, and do not continue the analysis or estimate any values until I provide them:
Game name: [ ]
Month analysed: [ ]
D1 retention: [ ]%
D7 retention: [ ]%
D30 retention: [ ]%
Stickiness (DAU/MAU): [ ]%
Average sessions per DAU: [ ]
Average session length: [ ] minutes
Playtime per DAU: [ ] minutes
ARPDAU: $[ ]
Ad ARPDAU: $[ ]
ARPPU: $[ ]
Transactions per payer: [ ]
If I have more than one game, analyse each one separately. Do not average across my portfolio or report studio-level figures. Every game gets its own comparison, its own percentile placement and its own recommendations, because a studio average hides both the games that are working and the ones that are not.
STEP 2: THE BENCHMARKS
Source: GameAnalytics mobile benchmarks. Global, games with at least 10,000 monthly active users, October 2026. Retention and engagement cover roughly 5,500 games. The four monetization metrics cover roughly 1,800, because revenue tracking is optional and only games reporting payer activity are included.
Metric P25 P50 P75 P95
D1 retention 12.42% 20.78% 28.32% 40.60%
D7 retention 2.00% 3.80% 6.90% 15.83%
D30 retention 0.35% 0.87% 2.00% 6.95%
Stickiness (DAU/MAU) 4.78% 5.86% 7.98% 20.55%
Sessions per DAU 3.08 4.51 6.34 10.42
Session length (minutes) 1.95 3.14 5.00 10.19
Playtime per DAU (minutes) 7.78 13.01 27.06 59.60
ARPDAU (USD) 0.0012 0.0106 0.0729 0.6900
Ad ARPDAU (USD) 0.0042 0.0324 0.1115 0.3212
ARPPU (USD) 4.73 10.93 25.43 212.94
Transactions per payer 1.27 2.00 3.77 17.52
P50 is the median, meaning half of all games fall below it. Treat it as the typical game, not as a pass mark.
STEP 3: PLACE EACH METRIC
For every metric, and for each game separately, state where it falls: below P25, between P25 and P50, between P50 and P75, between P75 and P95, or above P95. Also estimate the approximate percentile within that band.
Calculate this rather than judging by eye, and show your working as a table with my value, the four benchmark values and the resulting placement side by side.
STEP 4: INTERPRET
Name the three strongest and three weakest metrics by percentile position, not by raw value, so metrics on different scales compare
fairly.
Flag any internal tension. Retention and engagement usually move together, so call out combinations that do not fit, such as strong
session length with weak D7, and say what that pattern typically indicates.Give three things to work on, ordered by likely impact, each tied to a specific metric and a realistic target. Moving from the median toward P75 is achievable. P95 is exceptional and should not be treated as a default goal.
RULES
Do not multiply or divide percentiles across metrics. The game at P95 for session count is not the same game at P95 for session length, so sessions multiplied by session length will not match the playtime benchmark.
Compare each metric only against its own distribution.
Skip any metric I have not supplied rather than assuming a value.
Be direct about weak numbers. I would rather know where I stand than be reassured.
Retention
D1 is the first honest read on whether a game has a hook. The median established game brings back 20.8% of its installs the next day, so even among games this size four in five are gone within 24 hours. The spread runs from 12.4% at the bottom quartile to 40.6% at the top, a little over three times. Onboarding and the first session do most of that work.
Seven days after install, the median game is holding 3.8% of the players it acquired, down from 20.8% on day one. The week costs more than the first day did. The gap between tiers widens with it, from roughly three times at D1 to nearly eight here. This is where the core loop either proves itself or quietly fails to.
A month in, the median game keeps 0.8% of its installs, roughly one player in 120. The top 5% keep 6.9%, around eighteen times more, and that ratio has climbed at every stage: three times at D1, eight at D7, eighteen here. Retention gaps compound, and it is live content rather than onboarding that separates the top.
What is retention in mobile games?
Retention is one of the most critical metrics for understanding the health and success of a mobile game. It measures the percentage of players who return to your game after their initial play session, typically tracked on key milestones like Day 1 (D1), Day 7 (D7), and Day 30 (D30). It’s widely regarded as the most critical metric in mobile gaming as it directly impacts every aspect of a game’s lifecycle: notably, user acquisition, monetization, and player engagement.
Retention is the foundation of a sustainable and prosperous game, as it measures how well your game resonates with players over time. A game with poor retention won’t survive long, regardless of how many users are acquired. Conversely, strong retention indicates players are finding value, enjoyment, or a reason to return.
Engagement
Monthly players are not daily players. In the median game, 5.9% of monthly actives open it on any given day, fewer than two days a month each, while the top 5% reach 20.6%, closer to six days. February reads high at every tier, but that is a 28-day month shrinking the denominator rather than players genuinely returning more often.
Sessions per DAU, also called session count, measures how often a player opens the game on a day they play at all. Rarely just once: the median active user starts 4.5 sessions a day, and even the bottom quartile manages three. At roughly three and a half times from P25 to P95, this is the narrowest spread on the page, which suggests frequency owes more to genre than to anything a team can tune.
Sessions are short, and in the middle of the market they are slowly getting longer. The median now runs 3.1 minutes, up from 2.97 in March, while the top 5% have drifted down from 10.76 in April to 10.19. It is the only sustained directional movement on this page, and it is narrowing the gap between typical and strong.
Playtime combines frequency and length, and separates games further than either on its own: 13 minutes a day at the median against 59.6 at the top, almost eight times. Resist multiplying the percentiles to get here. The game ranked highest for session count is not the one ranked highest for session length.
Monetization
Monetization data comes from a smaller pool. Revenue tracking is optional, so around 1,800 games report payer activity, and these charts describe that group. Three of four charts stop at the 75th percentile because revenue is concentrated enough at the top that showing the 95th would flatten every other band. That figure appears in the commentary under each chart.
Most games earn almost nothing per player per day. The median makes about one cent per daily active user and the bottom quartile barely registers, which is what the near-flat green line shows. This is the most skewed metric on the page because it divides revenue by every user, including the large majority who never pay. The 95th percentile sits at $0.69, roughly 65 times the median.
Once a player pays, the figures change completely. The median paying user is worth $10.93 a month, the bottom quartile $4.73 and the top quartile $25.43, a far tighter band than ARPDAU. The 95th percentile is the outlier at $212.94, around twenty times the median, and it swung between $152 and $272 across the year.
Ad revenue is the one line on this page that is clearly rising. Median ad revenue per daily active user has gone from $0.029 to $0.032 over the year, and the 75th percentile from $0.089 to $0.111. The 95th percentile has climbed the most, from $0.208 to $0.321, a 54% increase. This is a real trend rather than month-to-month noise.
The median payer buys exactly twice, and that has barely moved all year. The bottom quartile sits at 1.27, meaning most payers in most games buy once and stop. The 95th percentile, not shown, reaches 17.5 purchases, which is where subscription-like and consumable-driven economies appear. Repeat buying, not the first purchase, is what separates them.
PC game benchmarks
Every figure below comes from PC games on the GameAnalytics platform, worldwide, measured on desktop builds. A game enters the sample once it passes 3,000 monthly active users, which works out to roughly 1,200 titles in a typical month. That threshold is lower than the mobile section uses, because the PC population is smaller, so the two sets are not directly comparable and are better read side by side than against each other. Figures are shown as percentiles rather than averages: the 25th marks the bottom quarter, the 50th is the median, and the 95th the top twentieth.
Compare your PC game to the industry
Copy this prompt into Claude, ChatGPT, Cursor, or any AI tool to see how your game stacks up against these benchmarks.
Compare my PC game's performance against GameAnalytics industry benchmarks.
STEP 1: FIND MY DATA
Before asking me, check whether my game data is already available to you. Look in this order:
- The GameAnalytics MCP server, if you have it connected
- Anyy other connected source, such as a spreadsheet, a cloud drive, a database, or files in the current project or repository
- Earlier in this conversation, or in our recent conversations if you can search them
If you find data, tell me where you found it and which month it covers before you use it, so I can confirm it is the right source.
If you find nothing, stop and tell me so. Ask me to paste the figures below, and do not continue the analysis or estimate any values until I provide them:
Game name: [ ]
Month analysed: [ ]
D1 retention: [ ]%
D7 retention: [ ]%
D30 retention: [ ]%
Stickiness (DAU/MAU): [ ]%
Average sessions per DAU: [ ]
Average session length: [ ] minutes
Playtime per DAU: [ ] minutes
New users: [ ]
If I have more than one game, analyse each one separately. Do not average across my portfolio or report studio-level figures. Every game gets its own
comparison, its own percentile placement and its own recommendations, because a studio average hides both the games that are working and the ones that are not.
STEP 2: THE BENCHMARKS
Source: GameAnalytics PC benchmarks, measured on desktop builds. Global, games with at least 3,000 monthly active users, October 2026, roughly 1,200 games.
Metric P25 P50 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%
Stickiness (DAU/MAU) 3.83% 4.32% 6.43% 14.89%
Sessions per DAU 1.44 1.66 2.15 3.70
Session length (minutes) 11.60 20.82 33.83 71.79
Playtime per DAU (minutes) 17.77 37.73 69.91 159.23
New users 4,556 10,657 34,780 256,572
P50 is the median, meaning half of all games fall below it. Treat it as the typical game, not as a pass mark.
Do not compare these against mobile benchmarks. PC games are played in long sittings spread across fewer days, so day-based retention is structurally
lower and session length structurally higher. The median PC session runs 20.8 minutes against 3.1 on mobile, from 1.7 sessions a day rather than 4.5. A PC D1 of 6.5% is normal, not a failure.
STEP 3: PLACE EACH METRIC
For every metric, and for each game separately, state where it falls: below P25, between P25 and P50, between P50 and P75, between P75 and P95, or above P95. Also estimate the approximate percentile within that band.
Calculate this rather than judging by eye, and show your working as a table with my value, the four benchmark values and the resulting placement side by side.
STEP 4: INTERPRET
Name the three strongest and three weakest metrics by percentile position, not by raw value, so metrics on different scales compare fairly.
Read retention as content consumption and lifecycle, not daily habit. Ask whether players progressed far enough to come back for updates, expansions or
multiplayer, rather than whether the game became a daily routine.
Flag any internal tension. Long sessions with weak D7 suggests the content ran out; short sessions with strong D30 suggests a routine that has not deepened. Say what each pattern typically indicates.
Give three things to work on, ordered by likely impact, each tied to a specific metric and a realistic target. Moving from the median toward P75 is achievable. P95 is exceptional and should not be treated as a default goal.
RULES
Do not multiply or divide percentiles across metrics. The game at P95 for session count is not the same game at P95 for session length, so sessions
multiplied by session length will not match the playtime benchmark.
Compare each metric only against its own distribution.
New users is an absolute count, not a rate, so it reflects marketing spend and store placement rather than game design. Place it, but do not treat a low figure as a quality problem or include it in the three things to work on unless I ask about acquisition.
A D30 P25 of 0.00% is real, not missing data. At least a quarter of PC games retain essentially nobody at thirty days.
Skip any metric I have not supplied rather than assuming a value.
Be direct about weak numbers. I would rather know where I stand than be reassured.
Retention
Day one on PC looks nothing like mobile. The median desktop game brings back 6.5% of its installs the next day, against 20.8% on mobile, and that is a difference in how the games are played rather than in quality. A PC player who buys deliberately and sits down for an hour is not checking in the next morning. The spread runs from 4.2% to 32.4%.
By day seven the median desktop game holds 1.4% of its installs, down from 6.5% on day one. The proportional fall is steeper than mobile's, but it means something different: a PC player can go a week without playing and still not have churned. Read this as whether the content pulled them back at all, not as a measure of weekly habit.
A month out, the median desktop game holds 0.19% of its installs while the top 5% hold 3.85%, around twenty times more. The bottom quarter sits at zero in October, meaning at least a quarter of PC games keep essentially nobody thirty days on. For PC this is the clearest read on lifecycle design: the games still holding players are generally the ones shipping updates, seasons or multiplayer content.
Why is PC game retention different from mobile game retention?
Retention in PC games reflects a different engagement model than on mobile. While mobile retention often measures how successfully a game becomes a daily habit, PC retention is more closely tied to content consumption, progression depth, and lifecycle design. Many PC games are designed for long, immersive sessions spread across fewer days, making lowerday-based retention a natural outcome rather than a sign of underperformance.
In this context, retention should be interpreted as the time players continue to engage with a game after their first experience, not how frequently they return each day. It provides insight into whether players find enough value to progress through content, return for updates, or re-engage after major milestones such as expansions, seasonal updates, or multiplayer events.
Engagement
PC players return on fewer days than mobile players. In the median desktop game, 4.3% of monthly actives appear on any given day, around 1.3 days a month each, against 5.9% on mobile. The top 5% reach 14.9%, closer to four and a half days. February reads high at every tier here too, the same 28-day denominator effect seen in the mobile section.
Sessions per DAU, also called session count, is the tightest metric in the PC set. The median desktop player opens the game 1.7 times on a day they play, against 4.5 on mobile, and even the top 5% only reach 3.7. PC is close to one sitting per day. Where mobile engagement is assembled from many short visits, PC is built from a single long one.
This is where PC separates from mobile completely. The median desktop session runs 20.8 minutes against 3.1 on mobile, nearly seven times longer, and the top 5% reach 71.8 minutes. One PC session outlasts an entire day of mobile play in most games, which is the strongest evidence for reading PC retention as content consumption rather than daily habit.
Daily playtime on PC comes from long sessions rather than frequent ones. The median desktop player spends 37.7 minutes a day, close to three times mobile's 13, assembled from 1.7 sessions instead of 4.5. The top 5% reach 159 minutes. Resist multiplying the percentiles: the game ranked highest for session count is not the one ranked highest for length.
New users is the only metric here counted in absolute numbers rather than rates, so it tracks marketing spend and store placement more than design. The median desktop game added 10,657 players in October, the bottom quartile 4,556 and the top quartile 34,780. This chart stops at the 75th percentile: the top 5% added 256,572, which would flatten every other band.
Roblox benchmarks
These figures come from 538 Roblox titles that passed 1 million monthly active users between August 2025 and July 2026, together accounting for 4.76 billion sessions. That is a deliberately high bar. It measures games already working at scale rather than the platform as a whole, so beating the median here means beating titles that have already proven themselves. Unlike the mobile and PC tabs, these describe the whole period rather than month by month, because the Roblox sample shifts with which titles happen to be large and live at any moment.
Compare your Roblox game to the industry
Copy this prompt into Claude, ChatGPT, Cursor, or any AI tool to see how your game stacks up against these benchmarks.
Compare my Roblox experience against GameAnalytics industry benchmarks. STEP 1: FIND MY DATA Before asking me, check whether my data is already available to you. Look in this order: - The GameAnalytics MCP server, if you have it connected - Any other connected source, such as a spreadsheet, a cloud drive, a database or files in the current project or repository - Earlier in this conversation, or in our recent conversations if you can search them If you find data, tell me where you found it and which period it covers before you use it, so I can confirm it is the right source. If you find nothing, stop and tell me so. Ask me to paste the figures below, and do not continue the analysis or estimate any values until I provide them: Experience name: [ ] Period analysed: [ ] D1 retention: [ ]% D7 retention: [ ]% D30 retention: [ ]% Average session length: [ ] minutes Average daily session count: [ ] Revenue per paying user: $[ ] If I have more than one experience, analyse each one separately. Do not average across them or report studio-level figures, because an average hides both the experiences that are working and the ones that are not. STEP 2: THE BENCHMARKS Source: the 2026 GameAnalytics Roblox Benchmark Report. 538 Roblox experiences that passed 1 million monthly active users between August 2025 and July 2026, together accounting for 4.76 billion sessions. These describe the whole period rather than a single month. Compare my experience against this table, and this table only. Metric P25 P50 P75 P90 P95 D1 retention 8.4% 10.3% 12.9% 15.9% 18.0% D7 retention 1.0% 1.5% 2.5% 3.9% 5.3% D30 retention 0.3% 0.5% 0.9% 2.0% 3.0% Session length (minutes) 7.3 9.8 12.0 14.7 16.6 Daily session count 1.34 1.56 2.07 3.22 4.06 Revenue per paying user (USD) 0.32 0.70 1.25 2.33 4.80 The figures below describe Roblox players across every experience they play, not individual experiences. Use them as context for how much attention and spend exists on the platform. Do not place my experience against them. Metric P25 P50 P75 P90 P95 Daily sessions per player 1 3 6 11 17 Session length per player (min) 2.93 7.28 14.85 26.30 37.18 Daily playtime per player (min) 6.58 22.40 58.70 123.85 186.47 Revenue per paying player (USD) 0.13 0.81 2.75 8.88 20.31 P50 is the median, meaning half of all experiences fall below it. Treat it as typical, not as a pass mark. STEP 3: PLACE EACH METRIC For every metric, and for each experience separately, state where it falls: below P25, between P25 and P50, between P50 and P75, between P75 and P90, between P90 and P95, or above P95. Also estimate the approximate percentile within that band. Calculate this rather than judging by eye, and show your working as a table with my value, the five benchmark values and the resulting placement side by side. STEP 4: INTERPRET Name the three strongest and three weakest metrics by percentile position, not by raw value, so metrics on different scales compare fairly. Flag any internal tension. Long sessions with weak D7 suggests the content ran out. Strong D1 with weak D30 suggests a good first impression without the live content to sustain it. Say what each pattern typically indicates. Give three things to work on, ordered by likely impact, each tied to a specific metric and a realistic target. Moving from the median toward P75 is achievable. P95 is exceptional and should not be treated as a default goal. RULES Do not compare these against mobile or PC benchmarks. Roblox retention is structurally lower because the typical player chooses between millions of free, one-click experiences rather than committing to something they installed. A median D1 of 10.3% is normal for the platform, not a failing experience. Every experience in this sample already has 1 million or more monthly active users. If mine is smaller, beating the low end of these benchmarks is a strong signal rather than a low bar, and falling short of the median is not a warning sign. Do not multiply or divide percentiles across metrics. The experience at P95 for session count is not the one at P95 for session length, so the two will not combine into a playtime figure. Retention and monetization now shape discovery together. Roblox guidance from August 2026 signals that experiences strong in both may see broader Home recommendation than those strong in only one, so treat them as related rather than separate goals. Skip any metric I have not supplied rather than assuming a value. Be direct about weak numbers. I would rather know where I stand than be reassured.
Player behavior
This part looks at players rather than games, aggregated across every GameAnalytics-tracked Roblox game a person played. It describes the total pool of attention and spend available on the platform, which is the ceiling any single game competes for a share of. No game creates more of either, it can only capture more of what already exists.
Most Roblox players move between devices. The largest group, 44%, plays on both mobile and PC, with PC-only second at 40.76% and mobile-only at 11.55%. Console in any combination accounts for under 4%. Build for both PC and mobile as a baseline. This is the one chart drawn from the full pool of 17,000+ integrated games rather than the 1M MAU sample, because it measures coverage rather than behavior.
The median player opens Roblox three times a day across every game they touch, reaching six at the 75th percentile and 17 at the 95th. Read these as the ceiling rather than a target for your own game, since they count sessions across all titles a player used. If you build idle timers or offline progression, the high end shows what that kind of design produces.
The median Roblox session runs 7.28 minutes, roughly doubling to 14.85 at the 75th percentile and reaching 37.18 at the 95th. These are pooled across whatever titles a player used in that session, so they are not directly comparable to a single game's dashboard figure. Deep progression, co-op and competitive loops are what hold a player inside one experience rather than letting them drift to the next.
Across everything they play in a day, the median Roblox player spends 22.4 minutes, rising to 58.7 at the 75th percentile and 186.5 at the 95th. Use this as a diagnostic rather than a target. If a player's platform-wide playtime runs to hours but your own data shows a few minutes, the attention exists and your game is not the one holding it.
Among players who spend anything at all, most spend very little. The median paying player puts down $0.81 across the period, reaching $2.75 at the 75th percentile and $20.31 at the 95th. Only 3.80% of tracked players paid anything. This is the player's ceiling rather than your game's: a paying player has already decided to spend on Roblox, and the question is what share of that comes to you.
Game-level benchmarks
This part ranks individual games. Every metric is calculated and ranked independently across the sample, so where a game sits on one says nothing about where it sits on another. These are separate distributions, not a combined score.
This is where session length becomes something a game controls rather than a player's pooled activity. The median game holds players 9.8 minutes per session, 12.0 at the 75th percentile and 16.6 at the 95th. Do not read a session under 10 minutes as underperformance, it is close to the median at this scale. Even the strongest titles work within a narrow band rather than reaching 30-minute averages.
The median game sees 1.56 visits per player per day, just above a single session, rising to 2.07 at the 75th percentile and 4.06 at the 95th. Returning once or twice a day is normal even among large titles. The move from roughly two sessions to three or four is where daily quests, timers and rotating rewards earn their place, and it compounds with session length rather than competing with it.
Retention on Roblox behaves unlike almost any other platform. At the median, even among these large titles, 10.3% of players return the day after their first session, 1.5% a week later and 0.5% by day 30. That is not a sign of broken games. It is what retention looks like where the typical player is choosing between millions of free, one-click experiences rather than committing to something they installed.
This ranks games by what their paying players spend, not players by how much they spend. The median game sees $0.70 per payer, $1.25 at the 75th percentile and $4.80 at the 95th. If your own figure is under a dollar you are in line with the median large game, not behind it. Reaching 1 million monthly actives does not by itself lift per-payer spend, which only moves at the very top.