July 13, 2026 By Pranay Parikh, MD

Why does retention drop off a cliff after month 2?

On July 1, 2026, Medicare opened a $50 access path for eligible Part D members through the CMS Medicare GLP-1 Bridge. A cliff after month two usually means the patient's reason to continue changed: the episode ended, the expected next step never happened, recurring value weakened, or the price being compared changed. Separate appropriate completion from preventable departure before trying to bend the curve.

"Month two" can mean a second charge, a refill decision, a follow-up visit, or 60 days since intake. Pick the event that represents continued value in the actual care model. An open subscription is not proof that care continued, and a closed account is not proof that care failed.

First decide which departures are healthy

Retention is a business measure, not a clinical goal. Some patients should leave because a defined episode ended, a clinician directed them to another setting, the service was no longer appropriate, or the patient chose a documented pause. Labeling all of those outcomes as churn creates pressure to retain people who should not remain enrolled.

Start with mutually exclusive exit reasons:

Keep every patient in one primary category. If the same person appears under "price," "no appointment," and "unknown," the totals cannot identify a cause. Keep appropriate clinical exits visible, but remove them from the preventable-loss rate.

This first split often changes the shape of the problem. A care model built around short episodes should not be judged by indefinite enrollment. A recurring program that promises ongoing access should not hide an unavailable follow-up slot inside "patient choice."

Price and coverage changed in 2026

A month-two cohort can cross a policy or price boundary even when the product has not changed. The Medicare GLP-1 Bridge began July 1, 2026, offers eligible beneficiaries a $50 copay, and is scheduled through December 31, 2027. Lilly's 2026 Foundayo release listed a self-pay starting price of $149 per month at the lowest dose.

Those figures should not be treated as one blended market price. A June cash-pay cohort, a July Medicare-eligible cohort, and a commercially insured cohort can face different out-of-pocket choices. If they are averaged together, a coverage transition can look like a lifecycle failure, or a genuine price problem can disappear inside a stronger subgroup.

For any recurring telehealth service, split the month-two curve by:

Record the date of every pricing, coupon, benefit, or formulary change. A line drawn through those dates is more useful than a single company-wide retention percentage.

The next step can be harder than the first

The first encounter has a clear job. The patient wants an evaluation, an answer, or a plan. Month two asks a different question: what useful thing happens next?

The answer needs to exist in the service, not just in a reminder. Patients need to know whether follow-up is indicated, who owns it, when it should happen, what it costs, and what to do when the original path no longer fits. If the only visible event is another charge, leaving can be rational.

Operations can also create the cliff. Follow-up inventory tightens. A clinician handoff repeats the history. Support takes longer to respond. A video requirement blocks someone who could complete an allowed phone visit. A payment failure gets recorded as voluntary cancellation.

Modality deserves its own split. A 2025 Journal of Telemedicine and Telecare study indexed by PubMed examined 3,105,382 scheduled appointments: 81.2% were in person, 13.4% were video, and 5.4% were phone. That study was not a subscription-retention benchmark. Its scale and three-way mix show why "telehealth follow-up" is too broad a category for diagnosis. Phone, video, and in-person fallback paths can produce different access patterns.

Compare the month-two cliff by required modality, appointment wait, clinician handoff, support queue, and state. If departure rises where the next available slot moved out or where a phone option disappeared, fix access before rewriting messages.

Silence before departure is an early signal

The cancellation date is often the last visible event, not the first sign of trouble. Portal use, care-plan views, scheduling attempts, message replies, and completed check-ins can weaken earlier.

A JMIR analysis of eHealth dropout signals found that two weeks of inactivity was one of the most significant predictors, alongside the intervention provider and the number of coach recommendations. That single study does not set a telehealth benchmark. The useful lesson is structural: activity changed before the final exit.

Build leading indicators around actions that matter in the care model:

Do not turn every missing click into a clinical alarm. Use aggregated patterns for business diagnosis, and route individual clinical concerns through the approved care process. The purpose is to find service friction sooner, not to infer health status from behavior.

Then inspect the expectation and follow-up gap

If appropriate completion, coverage, price, access, and payment do not explain the cliff, inspect what the patient expected after the first encounter and what the service actually delivered.

The cause can begin at signup. A page promises ongoing support, but the first-visit summary never defines it. A clinician documents follow-up, but the scheduling path does not surface it. A renewal notice names the charge, but not the service available during the next period. A reminder asks the patient to return without explaining why the next step matters.

That is the missing link. The visible departure happens in month two, while the preventable cause sits earlier in expectation setting, follow-up ownership, or the sequence of useful communications. More reminders will not fix missing capacity or a weak recurring product. One clear message at the right handoff can outperform a high-volume sequence that points nowhere.

The month-two exit most teams misread

From Pranay Parikh, MD:

The most common month-two exit is not a clinical one. The patient stops because they have not lost weight, and concludes the medication does not work for them.

But the first month's dose is not really there to produce weight loss. It is there to let the body get used to the medication. The starting dose is tolerance, not treatment. The patient was never going to see much on the scale, and nobody told them that.

That is not an appropriate discontinuation. It is a preventable service failure, and it happened before the first injection. It happened in the ad, in the welcome email, in whatever set the expectation. If your onboarding does not say plainly that month one is titration and the scale is not the month-one scorecard, you will lose patients who were responding exactly as expected.

This is why month-two retention is a marketing problem before it is a clinical one. The dose schedule is not going to change. What the patient was told to expect from it can.

Find the first point where continuing stopped making sense

First changed signal Leading explanation Evidence to inspect
Planned completion or referral rose Appropriate clinical exit Aggregated disposition and follow-up plan
Price or coverage exits rose on a known date Payment-path change Cohort, payer, copay shown, and effective date
Due follow-ups rose but completed visits fell Access or handoff failure Modality, wait time, state, clinician, and support queue
Meaningful activity faded before cancellation Recurring-value or service gap Scheduling attempts, replies, care-plan actions, and unresolved requests
Operations held but unknown exits rose Expectation and communication gap Signup promise, first-visit summary, renewal notice, and exit interview

Then make six decisions:

  1. Define a valid retained patient. Use an indicated follow-up, an appropriate active-care status, or another care-model event. Do not use a login alone.
  2. Separate healthy exits. Keep completion and referral visible without treating indefinite enrollment as success.
  3. Split price and coverage cohorts. Do not blend cash pay, commercial coverage, and Medicare when their choices differ.
  4. Track the first missing action. A failed payment, unavailable slot, unread message, and silent cancellation need different owners.
  5. Inspect modality and state. Video, phone, and in-person fallback access should not share one unexplained total.
  6. Repair the earliest break. Clarify the plan before renewal, restore a usable next step before adding reminders, and fix support ownership before asking the patient to reengage.

The output should be a named exit reason, a date, a cohort, and an owner. "Month-two churn" is only the symptom.

Get the Marketing Triage diagnostic by email to sort the cliff into appropriate completion, recurring value, price, access, payment, or follow-up.

If the result remains mixed, use marketing triage to identify which retention review should come first.

Use why retention drops off after month 2 to examine the earliest preventable exits, then map each handoff with the telehealth growth metrics reference. The Marketing Triage diagnostic turns exit reasons into a prioritized review, while the telehealth growth metrics reference connects the full patient relationship.

FAQ

Is every month-two departure a retention problem?

No. Planned completion, a clinician-directed referral, or the end of an episodic need can be appropriate. Separate those outcomes from price confusion, payment failure, unavailable follow-up, service problems, and silent departure before deciding what to change.

How should a telehealth service measure retention for episodic care?

Define the next clinically and operationally meaningful event for that care model, such as an indicated follow-up or an appropriate active-care status. Report planned completion and referral separately so indefinite enrollment does not become the default definition of success.

What should we inspect first when many patients leave after month two?

First verify whether continued care was expected, then find the earliest break across price and coverage, recurring value, appointment access, payment, follow-up clarity, and handoff ownership. Fix that earliest break before adding reminders.