A health subscription behaves differently from a streaming or meal-kit subscription. The product has a physical or clinical rhythm: a bottle lasts 30 days, a monthly insulin pen needs a refill on a known date, a nutrition plan has a review at week six. Retention is not only a marketing outcome. It is a function of whether the product fits that rhythm, whether the customer feels a reason to continue, and whether the reminder arrives before the supply runs out.
This article is for growth and operations leads at supplement, diagnostics-plus-coaching and chronic-care brands that already have a few months of subscription data. It shows how to build a cohort table, how to read the curve, how to add refill-timing analysis that most dashboards miss, and how to run one experiment at a time. The numbers below are hypothetical and labelled so; use them to check your own spreadsheet logic, not as benchmarks. Published retention benchmarks for health subscriptions are thin and vary by category, so your own curve is the reference that matters.
Why average churn hides the problem
Many dashboards show a single monthly churn figure. That number mixes brand-new customers, who leave in predictable waves, with long-standing customers, who rarely do. A healthy-looking blended figure can hide a business that loses most new subscribers in month two and survives only because older cohorts stay. The reverse also happens: a bad blended figure may come from one poorly performing acquisition campaign.
A cohort groups customers by the month they started. A cohort curve tracks how many of each group are still active in month one, two, three and so on. Reading cohorts side by side lets you see whether a change you made in March improved the March cohort compared with February, which a blended average cannot tell you.
Build the cohort table
You need three fields per subscriber, extracted from your subscription system or order table:
- Start month: the month of the first paid subscription order
- Order dates: every subsequent paid renewal
- End date: the date of cancellation, or the date you decide the customer has lapsed
Define "active in month N" precisely. For a monthly bottle, a customer is active in month N if a paid order shipped in that period. Pause and skip states need their own rule. Many brands count a paused customer as active, which flatters the curve; track pauses as a separate state.
A hypothetical table, counted in customers:
| Start cohort | Started | Month 2 | Month 3 | Month 4 | Month 6 | Month 9 |
|---|
| Jan | 1,000 | 640 | 520 | 470 | 410 | 380 |
| Feb | 900 | 560 | 430 | 380 | 330 | n/a |
| Mar | 1,100 | 780 | 650 | 590 | n/a | n/a |
Divide each cell by the starting customers so cohorts of different sizes are comparable. In the hypothetical table, the steepest fall is between the first and second months in all three rows. That is the first finding: the loss is front-loaded, so effort belongs in the first renewal, not in a general loyalty scheme.
The curve for older customers flattens. In the January row, the move from month six to month nine is small compared with month one to two. A flattening curve indicates a core group for whom the product has become a routine. The commercial question is how to make a larger share of each cohort reach that flat part.
Read three shapes
A steep drop followed by a plateau. Most new subscribers try once and leave; those who stay, stay. Look at onboarding, expectation setting and the first delivery experience.
A staircase. Drops recur at regular intervals, for example at months three and six. This often tracks the product cycle: a three-month course ends, a doctor review happens, a discount expires. Look at what changes at those dates.
A slow constant slide. Customers leave at a steady rate without a clear cliff. This usually points to weak ongoing value, rising irritation with delivery or price, or an absent reminder and engagement system.
Do not diagnose before you segment. The same blended curve may be a plateau for one channel and a slide for another.
Cut the cohorts by what you control
Add one dimension at a time and compare the month-two retention of each slice. Useful slices for health brands:
- Acquisition channel: search, Meta, influencer, marketplace referral, doctor referral
- Entry offer: full price, first-order discount, free trial, bundled starter kit
- First product: which SKU the customer started with
- Delivery experience: on-time against late first delivery
- Prescription or plan type: where a clinician sets a regimen
- Payment method: card, UPI mandate, cash on delivery converted to prepaid
Large differences between slices are your hypotheses. If customers who arrived on a heavy discount leave sooner than those who started at full price, the offer is buying low-intent customers. If customers whose first delivery was late have much lower month-two retention, the fix is operational rather than marketing. Keep each slice large enough to be meaningful; a slice of twenty customers will mislead you.
Add refill-gap analysis
Cohort tables show whether a customer is active in a month. They do not show whether a customer ran out of product before the next order. For products with a fixed days supply, the refill gap shows the difference.
For each renewal, calculate:
refill gap = next order date minus (previous order date + days supply)
A negative or zero gap means the customer reordered before running out. A positive gap means the customer went without product. In a chronic or supplement context, those days without product often precede cancellation, because the habit is broken. Treat that as a hypothesis to test in your own data rather than a fact.
Plot the distribution of gaps for each renewal number. A hypothetical reading: for renewal two, most customers reorder within two days either side of the due date, but a visible tail reorders after ten or more days. The tail is the at-risk group. You can test the idea by splitting later retention by first-refill gap: if customers in the tail reach month four far less often than on-time customers, the gap is a usable early-warning signal.
Once you have the relationship, use it. The reminder system should aim to land the nudge before the supply ends, not after the missed shipment.
Refill timing in practice
Work out the median gap between delivery and the point where customers start to use up the product. For a thirty-day bottle delivered on day three, the customer is likely to run out around day thirty-three if they start immediately. A reminder or auto-ship date set at day thirty-five is already late.
Three timing rules to test:
- Ship early. Schedule the next shipment so delivery arrives around five to seven days before the expected run-out date, allowing for courier variability.
- Warn before charging. Send an order notice a few days before each charge, with a one-tap skip or pause. Surprise charges produce disputes and cancellations.
- Follow up after delivery. A short message after the first delivery that explains how to take the product and what to expect reduces confused early drop-off. Product usage guidance and any dose information must come from your clinical team.
Keep reminders compliant. Under India's Digital Personal Data Protection Act, 2023, you need a lawful basis and clear consent for messaging customers about their health products, and the notice must say what the data is used for. The practical points for marketing teams are in DPDP Act for healthcare marketing.
Find the month and fix one thing
After segmentation and refill analysis you should have one or two specific problems, for example:
- "Month-two loss is highest for discount-led Meta customers"
- "Customers whose first reorder is more than ten days late rarely reach month four"
- "Many cancellations happen within three days of the charge date"
Pick one. Design an experiment with a single change, a defined audience and a pre-agreed metric:
| Problem | Change to test | Primary metric | Minimum run |
|---|
| Surprise charges | Order notice three days before billing with a skip option | Month-two retention of the new cohort | Two full cohorts |
| Late first delivery | Promise a narrower window and track on-time rate | On-time first delivery rate and month-two retention | One month |
| Confusing onboarding | Add a usage guide message after delivery | Reorder within seven days of due date | One cohort |
| Discount-led churn | Smaller first-order discount, bundled starter kit instead | Month-three retention per customer acquired | Two cohorts |
Compare the experiment cohort with the previous cohort and with a holdout where practical. Seasonality, a price change or a courier problem can swamp the effect, so write down what else changed.
Cancellation flows and honesty
A cancellation flow can be improved without trapping people. Offer a pause, a smaller pack or a delivery delay as options, and always include a clear path to cancel. India's consumer protection authority has issued guidelines on dark patterns, and a hidden or obstructive cancel path is the kind of design they target. A clean flow that records the reason for leaving also feeds your analysis; keep the reasons as a fixed list so they can be counted.
Measure the economics
Retention is useful only if it changes revenue per customer acquired. For each cohort compute cumulative contribution after product, delivery, payment and support costs, month by month, and compare it with what it cost to acquire the cohort. The month where cumulative contribution passes acquisition cost is your payback month. If the payback month is later than your cash can bear, you either need to cut acquisition cost or lift early retention, and the cohort slices tell you which.
Setting this up in a CRM keeps it repeatable. The healthcare CRM guide and tracking patients in a healthcare CRM explain pipeline and ownership structures that carry over to subscription follow-ups, and patient retention strategies covers the clinic equivalent.
A one-week starting plan
- Export first-order date, renewal dates, days supply and status for every subscriber.
- Build the cohort table by start month, with pauses tracked separately.
- Mark the month with the steepest drop.
- Add two segment cuts: channel and entry offer.
- Calculate refill gaps for renewal one and two.
- Choose one experiment, write down the metric and the date, and leave everything else alone for two cohorts.
For help setting up the data model and follow-up system, see our health tech startup marketing page.
Filed underhealth subscription retentioncohort analysis subscriptionrefill timing analysissupplement subscription churnchronic care subscription retention India

About the author
Nishu Sharma
Founder & CEO · Gurugram, India
Nishu founded Branding Pioneers in 2016 with one rule that hasn't changed since: healthcare only. She'd run digital strategy at a top-10 Indian agency and watched generalist marketing underserve medical clients who needed something built for how patients actually search and decide. So she left to build the specialist instead. It's now an 80-person team working with healthcare brands worldwide.
Free guide · 15 pages
The Patient Acquisition Blueprint
A 90-day plan from first search to booked appointment.