GLP-1 patient retention: what the data says and what operators can change

GLP-1 patient retention is the number that decides whether a weight loss brand is a business or an expensive way to generate leads. The published discontinuation data is not encouraging, and reading it carefully matters, because it separates the churn that comes with the drug class from the churn an operator can do something about. This post goes through the claims studies, when patients tend to stop, the four reasons they give, and the three levers that sit inside an operator's control: the first 30 days, how titration is handled, and when refills ship.

What the claims data says

Three sources are worth knowing by name, because they get quoted loosely and the loose versions are usually wrong.

A JAMA Network Open analysis of pharmacy and medical claims, available through PubMed Central, followed adults who started a GLP-1 receptor agonist and found that 46.5% of patients with type 2 diabetes and 64.8% of patients without diabetes had discontinued within one year. The without-diabetes group is the one that matters for a weight loss brand, since almost every cash-pay telehealth patient falls into it.

A Blue Health Intelligence issue brief on GLP-1 trends, drawn from Blue Cross Blue Shield claims, found that 58% of patients stopped before they had been on treatment long enough to reach a clinically meaningful benefit, and that roughly 30% stopped within the first month.

A Cleveland Clinic study, also on PubMed Central, reported 20.4% of patients discontinuing by three months and a further 32.0% discontinuing in the three-to-twelve-month window.

All three are insured populations, mostly on branded product, and none of them measures a program that wraps the prescription in messaging, coaching and scheduled follow-up. They describe what happens when a prescription is written and the patient is left to it. That is the ceiling: an operator who does nothing should expect something close to these numbers, and an operator who does the work is trying to beat them.

Telehealth cohorts versus in-person

There is no published head-to-head comparison of GLP-1 churn on telehealth versus in a clinic that this post can cite, and the claims studies above do not split their cohorts by how the prescription was obtained. Telehealth companies publish their own retention figures in press releases and investor decks. Those are marketing numbers with no stated denominator, and they should be treated as unverified.

What can be said is that a telehealth cohort differs structurally. Patients are usually paying cash, so price sensitivity is higher and there is no insurance prior authorization to fight through, which cuts both ways. Contact is mostly asynchronous, which lowers the effort of staying but also lowers the effort of leaving; a patient who has never spoken to a human feels no social cost in letting the card lapse. Billing runs on a subscription, so a share of churn is a declined card rather than a decision. And the product itself changed for many telehealth patients over 2025, which the supply section below covers.

How well patients adhere on telehealth therefore depends mostly on how much program is built around the prescription, and only a little on the channel. A telehealth brand that sends a shipment and a rebill and nothing else is running the claims-study experiment again.

When patients quit: month one, month three, month twelve

The timing tells an operator where to spend effort. The Blue Health Intelligence figure of about 30% stopping in the first month is the single most important number in this post. Month-one quitters have not plateaued and have rarely reached the cost fatigue that shows up later. They stop because the first injections were unpleasant, because nausea arrived and nobody told them what to expect, because the pen or vial confused them, or because the results they imagined did not arrive in four weeks. Each of those can be addressed by what the clinic says in the first two weeks.

Month three is where the Cleveland Clinic figure of 20.4% lands, and it is a different patient. Weight has come off, then slowed. The dose has stepped up and side effects have returned. Three months of charges are on the statement, and the patient is now asking whether the program is worth the money at the current pace. This is the point where the plateau conversation and the cost conversation happen at the same time.

Month twelve is the one the JAMA Network Open study measures. Some patients at this point have reached a goal and stop on purpose; others have drifted. The question at twelve months is whether the brand has a maintenance offer that makes sense to a patient who is done losing, and most brands do not.

The four reasons: side effects, cost, plateau, supply

Side effects come first in time and in volume. Gastrointestinal effects are the common ones and they cluster around dose increases. The operator's exposure is that a patient with an unanswered side-effect message has a very short fuse; the clinical answer belongs to the provider, but the speed of the answer belongs to the operator.

Cost is the reason that grows over time. For cash-pay patients the monthly price is a recurring decision, and the market moved under everyone in 2025. CNBC reported in March 2025 that the manufacturer of Wegovy began offering the drug to self-pay patients at $499 a month through its own direct channel, and after an oral form was approved in December 2025 the press reported telehealth pricing of about $149 a month. A brand pricing compounded product against those reference points has less room than it did a year earlier.

Plateau is physiology plus expectation. Weight loss slows for most patients after the early months, and a patient who was promised a straight line experiences the slowdown as failure. The fix is in what was said at intake and in what the provider says at each check-in, not in the drug.

Supply is the reason specific to the last two years. FDA declared the tirzepatide shortage resolved on December 19, 2024 and the semaglutide shortage resolved on February 21, 2025. Enforcement discretion for compounders wound down on a schedule the FDA published: for tirzepatide, 503A compounding pharmacies on February 18, 2025 and 503B outsourcing facilities on March 19, 2025; for semaglutide, 503B facilities through May 22, 2025. Brands that had built a program on compounded product had to change what they shipped, and every forced change is a churn event. Some of what looked like patient churn in 2025 was product churn.

What the operator controls: titration handling, message cadence, refill timing

Titration handling. A titration protocol is the schedule by which a provider steps the dose up over the first months. The protocol belongs to the licensed provider and the operator does not write it. What the operator does control is whether the platform treats it as a fixed calendar or as a series of decisions. The retention-friendly version has a check-in before every step, a simple way for a patient to report that the current step is not tolerable, and a provider who can hold or reverse a step without the shipment schedule fighting them. A titration protocol that auto-escalates on a date, regardless of what the patient reported, produces month-three churn on a schedule.

Message cadence. The first 30 days need more contact than any other period, and it should be front-loaded: a message within 48 hours of the first shipment arriving that covers what to expect, a check-in at the end of week one, a check-in a few days after each dose step, and a reminder before each shipment. Each of these is clinically appropriate follow-up and each is also the retention mechanism, because a patient who hears from the clinic has a reason to believe the subscription is doing something. The pattern is described in more detail in from visits to recurring care.

Refill timing. A refill that arrives after the patient has run out is a forced interruption, and an interrupted patient restarts at a lower dose or quits. Ship before the last dose, put the rebill on the same day as the shipment so the charge reads as treatment continuing, and run failed-payment recovery before the shipment date rather than after. A declined card that nobody chased is indistinguishable from a patient who chose to leave, except that it was cheaper to prevent.

The published discontinuation rate is the ceiling. The operator's job is the gap between that ceiling and the number set by patients who left because nobody caught them in week two.

What retention does to lifetime value with a flat consult fee and no markup

The arithmetic below is a labelled hypothetical, not a benchmark, and the program price and wholesale cost are round numbers chosen to make the point. Take a program priced at $249 a month. Assume the medication and cold-chain shipping cost the brand $120 a month at wholesale with no markup, and that the provider bills a flat $25 per completed consult, with a consult in month one, at each of two dose steps, and quarterly after that. Assume an acquisition cost of $250, which sits inside the $180 to $320 range that one paid-media agency reports for compounded GLP-1 programs.

A patient who stays four months generates $996 in revenue against $480 of medication, $75 of consults and $250 of acquisition: about $191 of contribution. A patient who stays six months generates $1,494 against $720 of medication, $100 of consults and the same $250 of acquisition: about $424. Two more months of retention more than doubles the contribution per patient, because acquisition was paid once and every later month carries only wholesale medication and the occasional consult.

The structure of the cost matters as much as the level. When the provider fee is flat per consult and the medication carries no markup, the marginal cost of a retained month is close to the wholesale price of the drug, so retention flows almost entirely to contribution. When a platform takes a percentage of revenue, retention is shared with the platform, and the operator's incentive to invest in month-one messaging is weaker than it should be. That is why Tessic Health publishes a $25 flat fee per completed consult and 0% medication markup with no revenue share on any plan: the retention work an operator does is kept by the operator. Run the model on the unit economics calculator with real wholesale numbers before setting a price.

Numbers that predict churn

Signups and visit volume predict very little. These are the numbers to watch by cohort month.

• Share of new patients who have exchanged at least one message with the clinic by day 14
• Time from first shipment to second fill, and the share of patients who reach a second fill at all
• Share of dose steps that were held, reversed or delayed at the patient's request, which shows whether titration is being handled or just scheduled
• Median time from a side-effect message to a provider reply
• Rebill rate at month three, split into voluntary cancellations and failed payments
• Recovery rate on failed payments before the shipment date

A brand that knows these six numbers can see a month-three problem in month one. A brand that knows only revenue finds out two quarters later.

Questions operators ask

Is retention worse on compounded product than on branded product? There is no clean published comparison. The claims studies are mostly branded. What is known is that a forced switch between products, in either direction, is a churn event, so the stability of supply matters more than which product a program starts on.

Does a lower price fix retention? It lowers cost churn and does nothing for month-one churn, which is the larger pool. A brand that cuts price without fixing the first 30 days has reduced margin on the same patients it was already losing.

Is a coaching or lifestyle add-on worth building? The claims data measured programs without one and produced the discontinuation rates above. A structured check-in cadence run by the clinic is cheaper than a coaching product and addresses the same month-one failure, so it comes first.

How does this fit a brand that has not launched yet? The retention design has to exist before the first patient, because month one starts on day one. Tessic Health's weight loss launch sets a brand up with licensed providers who own the titration protocol, cold-chain delivery timed to the dose schedule, and, on the Grow tier and above, the retention automation and cohort analytics described here. A brand can also build each piece itself; the point of this post is that the pieces are known and the data says which one to build first.