Three people at the same studio work out March's retention rate from the same booking data. The first divides the active members on March 31 by the members on March 1: 100.8%. The second takes out the people who joined during March: 93.3%. The third takes them out too and applies the studio's written rule that a short freeze is not a cancellation: 95.0%. Nobody slipped on a calculator. They answered three different questions, and the first was not about retention at all.
Below are the formula and the four decisions around it. One invented studio runs through all of it: Alturra Studios, 240 members on March 1 at $169 a month. During March 18 people joined, 12 members canceled and 4 froze, leaving 242 active members on March 31.
Retention in one minute
- Take the new members out. Retention asks whether the people you started with stayed; left in, joiners turn March into 100.8%.
- Write down who counts first. Freezes, packs, comps, staff and paid-in-full move the answer more than the formula; one freeze rule is the gap between 93.3% and 95.0%.
- Compound, never multiply. A year is a month's retention to the twelfth power: 95% a month keeps about 54% of today's members, not the 40% that 5% × 12 suggests. The churn rate calculator does the compounding for you.
| The number | The formula | Alturra's March | What it answers |
|---|---|---|---|
| Member retention | (members at end − new members) ÷ members at start | (242 + 4 − 18) ÷ 240 = 95.0% | did the members we started with stay? |
| Monthly churn | canceled ÷ members at start | 12 ÷ 240 = 5.0% | how fast are they leaving? |
| A year at this pace | monthly retention to the 12th power | 0.9512 = 54.0% | what a year of Marches keeps |
| Cohort retention | still members at day 90 ÷ that month's joiners | 12 ÷ 18 = 66.7% | are new members staying? |
| Dues retained | starting members' dues at end ÷ at start | $37,716 ÷ $40,560 = 93.0% | did the money stay with the people? |
What is the formula for member retention rate?
Members at the end of the period, minus the new members who joined during it, divided by the members at the start. As a sum: retention = (E − N) ÷ S, with every count taken under the same written definition of a member.
For Alturra's March, S is 240 and N is 18. E is the 242 active members on March 31 plus the 4 paused ones, because the studio's rule says a freeze under 60 days is not churn: (242 + 4 − 18) ÷ 240 = 228 ÷ 240 = 95.0%.
Why the new members come out. Eighteen people on the March 31 list were not members on March 1, so they cannot say whether March's members stayed. Left in, they lift a month in which 16 starting members canceled or paused to 100.8%, and a busy January can look like the best retention month of the year while it loses members.
End ÷ start is a real number with the wrong name: growth. Alturra had 2 more active members on March 31 than on March 1, or 0.8%, and that belongs on the board as growth, beside retention rather than in its place.
Is retention rate just 100 minus churn?
Yes, when both count the same members under the same rules. Alturra's March churn is 12 canceled ÷ 240 at the start = 5.0%, and 100 − 5.0 is 95.0%, exactly what the retention formula gave.
That makes the pair a free check on your counting. Both split the members you started with into stayed and left, so they must add up to 100; when they do not, someone is counted by one and not the other, and finding that person is worth more than either number.
The suspects are few: a freeze that retention counts as a loss and churn does not (the 98.3% above), a new member who joined and canceled inside the month, a plan switch recorded as a cancellation plus a join, or start and end counts taken on different days. Fix the count, not the formula: churn divides by the members who could actually cancel.
How do you turn a monthly rate into a year?
Compound it: raise monthly retention to the twelfth power. At 95% a month, 0.9512 = 0.540, so about 54% of the members Alturra has today would still be members in twelve months if every month were like March.
Multiplying is the tempting shortcut: 5% × 12 is 60%, leaving 40%, or 96 of the 240. But each month's 5% comes from a smaller group: 240 becomes 228, then about 217, then about 206, and after a year about 130 remain. The shortcut makes the year look worse than it is, more so as churn rises.
Each point of monthly churn moves the year by six to eight points, 15 to 20 of Alturra's members. The churn rate calculator runs the same compounding on your own numbers and puts the year in dollars. Give it a typical month, or an average of the last three or six, never an unusual one.
Over a year, end ÷ start flatters even more. Twelve months of joiners can lift December above January, so a studio can report retention over 100% while, at 5% a month, only about 130 of Alturra's original 240 would remain.
What is cohort retention, and why is it the truest number?
Cohort retention follows one month's joiners on their own: of the people who joined in March, what share are still members 30, 60 and 90 days after they joined. It is the truest number because nothing else gets into it: no joiners from other months, and no long-standing regulars to average the new ones out.
Alturra's March cohort is the 18 people who joined that month. Counted from each person's join date, 16 were members on day 30, 13 on day 60 and 12 on day 90: 12 ÷ 18 = 66.7%. Over the same three months the studio-wide 95% a month compounds to about 86% (0.953 = 0.857). The monthly rate is held up by long-standing members, and it hides how quickly new ones leave.
| Joined in | Day 30 | Day 60 | Day 90 |
|---|---|---|---|
| January24 joined | 2187.5% | 1770.8% | 1562.5% |
| February15 joined | 1493.3% | 1280.0% | 1173.3% |
| March18 joined | 1688.9% | 1372.2% | 1266.7% |
Read the table down the columns: is this month's cohort staying better than last month's? Alturra's January brought the most joiners and kept the smallest share, a first-week problem more than a membership one, and the first week decides whether a routine forms. To build one, list the memberships that started in the last six months, group them by join month, and mark each person a member or not on day 30, 60 and 90 under the same freeze rule. Retention by join month works the same way in a gym.
Who counts in the denominator?
Whoever your written definition of a member says counts, applied the same way at the start of the period and at the end. The formula is one line; the edge cases are where two honest people get different answers, and defining a member walks through the five that matter most.
Alturra wrote its rules down in January, before anyone worked out a rate. The freeze line is the one that moved March: Alturra treats a freeze as a pause, not an exit, until it passes 60 days without a return, and then it is churn.
| Edge case | In the start count? | Lost when? | Alturra's rule |
|---|---|---|---|
| Freeze | yes, while under 60 days | at 60 days with no return | a pause, not a loss, until day 60 |
| Plan switch | yes | never: she stayed | neither a cancellation nor a new join |
| Class pack | no | not in this number | customers, not members |
| Comp | no | not in this number | a comp ends when the studio ends it |
| Staff account | no | not in this number | out of everything but class capacity |
| Paid-in-full | yes, every month of the term | the term ends with no renewal | a member all term; cannot leave mid-term |
| Intro offer | no | not in this number | a trial; enters as new when she joins |
Whatever your rules, count paused members as their own group, so the freeze decision stays in view instead of disappearing into a total that one rule quietly moves by 1.7 points.
A column of their own shows which rule was used, so anyone can recount under the other. Two reports on one studio disagree for exactly this kind of reason, and the member definitions guide sets out the rules to write down once.
What to do this week:
- Write one line for each edge case in the table, and date the page.
- Recount last month's start and end under those lines, with paused members in their own column.
- Check it against 100 minus churn, and find anyone counted differently.
Should you track dues retained as well as members?
Yes, beside the member rate, because a member who stays on a cheaper plan counts the same as one who stays at full price. Dues retained is what your starting members pay at the end divided by what they paid at the start, new members left out.
On March 1 Alturra's 240 members paid $40,560 a month between them. By March 31 the same people paid $37,716, or 93.0%, two points under the 95.0% member rate. Here is where the $2,844 went.
A member rate that holds while dues retained slips means you are keeping people and losing money. A price increase pushes the other way, so dues retained can pass 100% in a month you lost people: it sits beside the member rate, never instead of it. What one member is worth rests on her dues as much as her stay, and the price increase piece covers how to measure what an increase cost.
What is a good retention rate for your studio?
A little better than your own last six months, measured the same way each month. The comparison that tells you something is your studio against itself, because a figure from another studio arrives with somebody else's definitions of a member, a freeze and a month.
Average six months by the formula, under your written rules. If Alturra's October to March average 94.5%, March's 95.0% is a slightly better month, not a turnaround, and April's target is to hold 95. As the history grows, compare this March with last March: seasons move the numbers, and your own history is the only record of how.
Mind the noise: at 240 members one cancellation moves the rate about 0.4 points, so a point either way is weather and three months in a row is news. And the rate is counted after members have gone, while the fade shows in attendance weeks earlier. Why members leave covers the four fades to watch for, and the twelve studio KPIs show where retention sits among the numbers that run a studio.
What to do this week:
- Work out the last six months by the formula and average them: that is your baseline.
- Start a cohort row for each of those months' joiners, and fill in day 30, 60 and 90 as they come due.
- Put dues retained beside each month's member rate.
- Compound the average in the churn calculator and write the year where you will see it.
Where this lives
The formula takes a minute with a pencil; keeping its inputs honest every month is the part that slips, and the part Xyzios covers. It works with Mindbody and is an approved Mariana Tek integration, and the platform stays your system of record. You decide what counts as a member and whether a freeze counts as churn, in your studio's definitions, and the boards follow. Paused members, people with a frozen membership and no active one, are counted as their own group. Until you set goals, the boards pace against your studio's own trailing performance, and the forecast projects the next three months from your own history.
It also watches the people behind the rate. Current members whose attendance is slipping against their own pattern land on an at-risk list in High and Medium bands, and a short save note is drafted for those at High risk. Nothing reaches a member until you tap approve, and approved notes send from your studio's own email. Each approved save is logged on the Recovered Revenue Ledger and checked at 60, 90 and 120 days. It does not write to anyone who has already canceled; how the save works covers the rest. The cohort rows and the dues column stay yours, by hand.
Whatever you count it in, start with last month. Put your own numbers through the churn rate calculator to see the year, or take the two-minute studio check to see where retention sits among the numbers that decide growth.