Spots the regular who has quietly stopped coming and reaches out in week three, not month three, with a specific class rather than a discount.
An AI agent for lapsed yoga students watches attendance patterns, notices when somebody has broken theirs, and sends a message while the break is still a break. Almost nobody cancels a yoga membership after a bad experience. What happens is a busy fortnight, then a holiday, then a sense that they have lost their fitness and will start again in January. By the time the direct debit gets canceled the decision was made two months earlier. The studios that hold on to people are the ones that notice at the three-week mark, and the message that works at three weeks is not a discount — it is somebody naming the class they used to come to and saying it is still there.
Turns a silent gap into a conversation while it is still recoverable.
Learns each student's normal pattern rather than applying one rule to everybody
Flags a break against that pattern, so a twice-weekly regular is noticed after two weeks
Messages with the specific class and teacher they used to come to
Offers to book them straight back in from the reply
Recognizes a member paying and not attending, which is the highest cancellation risk you have
Stops after one or two attempts rather than pursuing somebody who has moved on
Every studio owner can name the students who stopped coming, but they name them weeks late, usually when somebody asks after them. That lag is the whole problem. A message in week three lands on somebody who still thinks of themselves as a person who does yoga at your studio; the same message in month three lands on somebody who has to restart. It is also worth being careful about the instinct to lead with a discount, which teaches regulars that leaving is how you get a better price and signals that the studio thinks price was the reason. Usually it was not — it was a schedule change, an injury, or three busy weeks.
A simple three-step flow from broken pattern to rebooked class.
The agent builds a picture of how often each student comes and which classes they favor, so it can tell a gap from a normal rhythm.
When somebody's gap gets meaningfully longer than their own normal, they are flagged — sooner for a regular, later for an occasional visitor.
It sends a short message naming their usual class, offers to book them in, and passes anyone who replies with a reason worth hearing to you.
A realistic case of a twice-weekly member going quiet.
Scenario: a studio was discovering lapsed members at the point of cancellation, roughly two months after they had actually stopped coming. A member has come to Tuesday 6:30 and Saturday 9:30 almost every week for a year. She misses a Saturday, then the following Tuesday, then the Saturday after. On the fifteenth day the agent flags the break — not because fifteen days is long in general, but because it is long for her — and sends one message: we have not seen you in a couple of weeks, Tuesday 6:30 with Sam is on as usual, want me to save you a spot? She replies that she has hurt her wrist. The agent does not try to book her; it acknowledges it, offers to check back when she is ready, and passes the thread to the owner, who messages her about the wrist-friendly Yin class. She is back on Thursday.
Studios whose growth is capped by the back door rather than the front.
You notice absences from the front of the room but have no way to act on them systematically.
A member paying and not attending is the clearest cancellation signal there is, and it is invisible in revenue.
Replaces a quarterly retention review with something that acts inside the month.
Students lapse when a favorite teacher's slot moves, and that is fixable if you catch it early.
Timetable changes produce a wave of quiet lapses from people whose class no longer fits.
Summer gaps become permanent unless somebody makes contact before September.
Reads attendance and membership status to tell a gap from a goodbye.
Supplies attendance history, class preferences and membership status, and takes the rebooking.
Carries the message, because a win-back only works if it can be answered in one line.
Shows who is still being billed while not attending, which is the priority group.
Used for students who prefer email, and for anything longer than a rebooking offer.
An alternative source of attendance and membership status.
Records outreach and outcomes so nobody is contacted twice from two directions.
The gaps worth catching early.
Questions about messaging people who have gone quiet.
An AI agent for lapsed yoga students watches attendance patterns, notices when somebody has broken theirs, and sends a message while the break is still a break. It works against each student's own rhythm rather than one studio-wide rule.
It should be relative to the student, not fixed. Two weeks is alarming for somebody who came twice a week and meaningless for somebody who came monthly. A single studio-wide rule either spams the occasional visitors or misses the regulars, and usually does both.
Usually not on the first attempt. Leading with money implies price was the problem, when the reason is more often a schedule change, an injury or a busy month. It also teaches regulars that going quiet is rewarded. Keep a discount in reserve for the second contact if you use one at all.
It can be, and the difference is specificity and restraint. A message naming their class and offering a spot reads as a studio that noticed. Three messages in ten days reads as a system. One attempt, maybe a second, then stop.
The agent stops selling, acknowledges it, offers to check back when they are ready, and hands the thread to you. Injury conversations are exactly where a studio can build loyalty, and exactly where an automated rebooking prompt does damage.
It should. A member who is not attending is losing money every month and will eventually cancel; a pass holder with credits left has a deadline instead. Those are different messages, and sending the membership one to a pass holder is the fastest way to sound like you are not paying attention.
Only from what people reply, which is a partial picture but a useful one. Studios that run this for a few months usually find one or two recurring reasons — a class that moved, a teacher who left — that were visible in the replies long before they were visible in the numbers.
Spots the regular who has quietly stopped coming and reaches out in week three, not month three, with a specific class rather than a discount.