Start for $1

Retention demo on sample data

Who comes back after their first join on Demo network, and which regulars went quiet. All servers, cohorts from the last 8 weeks.

Day 1 return16.3%

+0.6 pp vs the 8 weeks before

450 of 2,769 came back, 100 still open

Day 7 return6.2%

+0.7 pp vs the 8 weeks before

146 of 2,372 came back, 497 still open

Day 30 return2.8%

+0.2 pp vs the 8 weeks before

32 of 1,139 came back, 1,730 still open

Players in cohorts2,869

+3.4% vs the 8 weeks before

First joins in the last 8 weeks

Return

How many first joins came back, by day and by week.

Return curve

Share of new players who played on each day after their own first join, pooled over the cohorts of the last 8 weeks. The dashed line is the 8 weeks before.

Return on Day 1, 3, 7, 14 and 30 for the cohorts of the last 8 weeks and the 8 weeks before.
CohortsDay 1Day 3Day 7Day 14Day 30
These 8 weeks16.3%9.4%6.2%3.8%2.8%
The 8 weeks before15.6%9.2%5.5%4.7%2.6%
Return on every day after first join, the cohorts of the last 8 weeks against the 8 weeks before.
DayThese 8 weeksThe 8 weeks before
Day 1 after first join16.3%, 450 of 2,76915.6%, 434 of 2,774
Day 2 after first join11.1%, 299 of 2,69510.8%, 300 of 2,774
Day 3 after first join9.4%, 246 of 2,6169.2%, 254 of 2,774
Day 4 after first join7.4%, 187 of 2,5407.5%, 208 of 2,774
Day 5 after first join6.4%, 159 of 2,4796.3%, 174 of 2,774
Day 6 after first join6.1%, 147 of 2,4165.7%, 157 of 2,774
Day 7 after first join6.2%, 146 of 2,3725.5%, 152 of 2,774
Day 8 after first join5.6%, 129 of 2,3174.8%, 132 of 2,774
Day 9 after first join5.4%, 123 of 2,2655.1%, 141 of 2,774
Day 10 after first join5.0%, 111 of 2,2104.6%, 128 of 2,774
Day 11 after first join4.2%, 91 of 2,1544.2%, 117 of 2,774
Day 12 after first join4.5%, 95 of 2,1004.2%, 116 of 2,774
Day 13 after first join4.3%, 87 of 2,0354.4%, 122 of 2,774
Day 14 after first join3.8%, 75 of 1,9804.7%, 130 of 2,774
Day 15 after first join4.1%, 79 of 1,9274.2%, 117 of 2,774
Day 16 after first join4.0%, 75 of 1,8794.4%, 121 of 2,774
Day 17 after first join4.5%, 83 of 1,8293.2%, 89 of 2,774
Day 18 after first join3.5%, 62 of 1,7733.3%, 92 of 2,774
Day 19 after first join3.7%, 64 of 1,7153.2%, 89 of 2,774
Day 20 after first join3.1%, 51 of 1,6563.4%, 95 of 2,774
Day 21 after first join3.5%, 56 of 1,6073.1%, 85 of 2,774
Day 22 after first join3.1%, 48 of 1,5573.5%, 96 of 2,774
Day 23 after first join3.6%, 55 of 1,5153.1%, 87 of 2,774
Day 24 after first join2.9%, 42 of 1,4653.0%, 84 of 2,774
Day 25 after first join3.7%, 53 of 1,4172.7%, 75 of 2,774
Day 26 after first join3.2%, 44 of 1,3632.8%, 78 of 2,774
Day 27 after first join3.1%, 40 of 1,3042.6%, 71 of 2,774
Day 28 after first join2.7%, 34 of 1,2502.9%, 80 of 2,774
Day 29 after first join3.0%, 36 of 1,1982.7%, 74 of 2,774
Day 30 after first join2.8%, 32 of 1,1392.6%, 72 of 2,774

Cohorts by week

Each row is one week of first joins. A cell is the share of that week's players who played on that day after their own first join.

Weekly cohorts of first joins, by day after each player's first join.
Cohort weekPlayersDay 1Day 3Day 7Day 14Day 30
Oct 1231516.7%9.7%PendingPendingPending
Oct 538613.5%9.3%5.9%PendingPending
Sep 2838415.4%8.9%6.3%2.6%Pending
Sep 2135516.6%10.1%5.1%3.4%Pending
Sep 1437917.2%10.6%6.1%4.5%0.0%
Sep 733820.4%9.5%7.7%3.8%3.0%
Aug 3136115.2%10.0%6.1%4.2%3.0%
Aug 2435115.7%7.4%6.0%3.7%3.1%

Pending is never counted as a failure.

Cohorts

Day 1, Day 7 and Day 30 return by day, week or month.

Return by cohort

Each row is the players who first joined in that day, week or month. A cell is the share who played on that numbered day after their own first join.

Apply
A numbered day is a 24 hour step from each player's own first join. A cell counts only players whose day has ended, so a window nobody has finished reads pending, never 0%.
Cohort weekPlayersDay 1Day 7Day 30
Oct 1231516.7%PendingPending
Oct 538613.5%5.9%Pending
Sep 2838415.4%6.3%Pending
Sep 2135516.6%5.1%Pending
Sep 1437917.2%6.1%0.0%
Sep 733820.4%7.7%3.0%
Aug 3136115.2%6.1%3.0%
Aug 2435115.7%6.0%3.1%
Aug 1736714.4%6.0%1.9%
Aug 1034615.9%6.9%3.8%
Aug 333817.2%2.4%1.8%
Jul 2732513.2%6.2%1.8%
All shown4,24515.9%5.9%2.5%

At risk

Regulars who went quiet and may be worth a message.

Players at risk

Regulars who look like they are leaving, with the reasons. Each rule adds points, and Medium starts at 30 and High at 55. A score ranks players for you to look at. It does not predict who will leave.

At risk
134

of 476 regulars

High risk
74

Score 55 or more

Payers at risk
44

With a paid, completed purchase

Showing the top 100 of 134 at risk. Export CSV holds the same 100.

Players at risk, highest score first.
PlayerRiskWhyAwayPayer
ThornWolfPremium Java, United StatesHigh80
  • Away 23 days, usually plays every 2.5 days+40
  • Played on 0 days in the last 14, about 4 before+20
  • Sessions shortened from 27 min to 16 min+10
  • Has never left spawn+10
23 daysNo
driftraven28Cracked Java, GermanyHigh70
  • Away 15 days, usually plays every 3.5 days+40
  • Played on 0 days in the last 14, about 2 before+20
  • Has never left spawn+10
15 daysYes
driftraven33Cracked Java, FranceHigh70
  • Away 16 days, usually plays every 3.3 days+40
  • Played on 0 days in the last 14, about 4 before+20
  • Sessions shortened from 64 min to 35 min+10
16 daysYes
BirchLynx20Premium Java, United StatesHigh70
  • Away 14 days, usually plays every 1 day+40
  • Played on 0 days in the last 14, about 2 before+20
  • Has never left spawn+10
14 daysNo
EchoSprout96Premium Java, BrazilHigh70
  • Away 16 days, usually plays every 2 days+40
  • Played on 0 days in the last 14, about 4 before+20
  • Sessions shortened from 51 min to 25 min+10
16 daysNo
CinderRavenPremium Java, United StatesHigh70
  • Away 16 days, usually plays every 2 days+40
  • Played on 0 days in the last 14, about 2 before+20
  • Has never left spawn+10
16 daysNo
CinderGolemPremium Java, BrazilHigh70
  • Away 16 days, usually plays every 1 day+40
  • Played on 0 days in the last 14, about 2 before+20
  • Has never left spawn+10
16 daysNo
.MossCrafter38Bedrock, United KingdomHigh70
  • Away 17 days, usually plays every 2 days+40
  • Played on 0 days in the last 14, about 2 before+20
  • Has never left spawn+10
17 daysNo
TidalFox14Premium Java, NetherlandsHigh70
  • Away 18 days, usually plays every 2 days+40
  • Played on 0 days in the last 14, about 2 before+20
  • Has never left spawn+10
18 daysNo
NovaKnight56Premium Java, GermanyHigh70
  • Away 19 days, usually plays every 2 days+40
  • Played on 0 days in the last 14, about 2 before+20
  • Has never left spawn+10
19 daysNo
How the score works
  • Away longer than usual (up to 40). Days since the last session divided by the player's usual gap between play days. 1.5 times adds 10, 2 times 22, 3 times 32, 4 times or more 40. Needs at least 2 days away.
  • Playing on fewer days (up to 20). Play days in the last 14 days against the 14 day average of the 28 days before. A drop of 25 percent adds 7, 50 percent 14, 75 percent 20. Needs at least 3 play days before.
  • Shorter sessions (up to 15). The newest 3 sessions against the sessions before them. Down to 80 percent adds 5, 60 percent 10, 40 percent 15. Needs at least 3 earlier sessions.
  • Never left spawn (up to 10). WorldGuard events in the last 60 days show only the spawn region. Adds 10. Skipped when the player has no region events.
  • Played through lag (up to 10). Sessions in the last 14 days that overlapped a lag event on the same server. 1 session adds 5, 2 or more add 10.

Payer status adds no points. A payer is not more likely to leave, they are worth more to keep, so payers sort first among equal scores. Weights were set by hand and checked on sample data. They are not fitted to your network and their accuracy has not been measured. Read the full method

Regulars gone quiet

Each point is the end of a week. It counts regulars, who played on 3 or more days in the 30 before, that are away at least 2 times (and 4 times) their usual gap. It uses the away rule only, so it can read lower than the list.

Regulars gone quiet in this range: latest, peak, med for each line.
LineLatestPeakMed
Away 2 times usual or more152152106
Away 4 times usual or more768650
Payers among the first505436
Regulars gone quiet per week. Table of the plotted values.
BucketAway 2 times usual or moreAway 4 times usual or morePayers among the first
Week ending Aug 21, 2026854619
Week ending Aug 28, 2026682317
Week ending Sep 4, 2026843721
Week ending Sep 11, 2026995032
Week ending Sep 18, 20261064936
Week ending Sep 25, 20261415946
Week ending Oct 2, 20261447054
Week ending Oct 9, 20261468646
Week ending Oct 16, 20261527650

A numbered day is a 24 hour step from each player's own first join. Cohort weeks start on Monday, UTC.

Sample data