A telehealth conversion funnel should distinguish interest, intake completion, payment, clinical review, and fulfillment. Those events answer different questions. A paid order is not a completed consultation, and a clinical decision is not a shipment. Define each event before comparing campaigns or evaluating a platform.
For a US cash-pay operator, the objective is a shared measurement language across marketing, finance, support, and clinical operations. The funnel should expose where work stops without turning clinical decisions into sales targets.
Define the unit you are counting
Choose whether each measure counts people, sessions, cases, orders, or transactions. Do not divide a transaction count by a patient count and label the result as if both represented the same population.
A person can start several sessions, submit one intake, receive more than one order, and generate multiple payment attempts. Keep stable internal references that connect those objects within authorized systems. Apply a documented deduplication rule before calculating conversion rates.
Record whether the analysis includes only new patients or also repeat activity. A refill should not create a second “new patient” just because it generates another payment.
Every metric definition should include:
- The event or status being counted.
- The unit, cohort, and observation period.
- Deduplication, refund, and cancellation treatment.
Build an event dictionary
The following event names are suggested internal definitions, not standard Meta, TikTok, or analytics-provider event names. Map them to external systems only after reviewing the destination’s requirements and permitted data.
| Internal event | Suggested trigger | What it does not mean |
|---|---|---|
| intake_started | First meaningful form interaction | A qualified patient |
| intake_submitted | Required intake successfully received | A completed clinical review |
| payment_succeeded | Confirmed successful collection | An authorized prescription |
| clinical_review_completed | Responsible clinician records an outcome | Every case was approved |
| order_shipped | Confirmed dispatch event | Delivery to the patient |
| refund_completed | Confirmed refund outcome | Future billing is canceled |
If your business offers services without shipped prescriptions, define the appropriate service-completion event instead. Do not force every program into a medication-commerce funnel.
Account for different event sequences
Some programs collect payment before clinical review; others collect later. Your reporting should reflect the actual workflow rather than assume that every patient follows one rigid sequence.
Represent clinical outcomes separately: additional information requested, service not appropriate, treatment plan completed, or another program-defined result. The clinical team should establish those categories. Marketing should not optimize toward increasing prescription approval as an isolated performance goal.
Use the subscription billing guide to define how payment, refunds, and pauses relate to the service. Use the pharmacy-routing guide to distinguish transmission from shipment.
Calculate stage rates from a consistent cohort
Suppose a hypothetical weekly cohort contains 1,000 distinct intake starters. Within the selected observation window, 600 submit, 300 make a successful payment, and 240 reach confirmed shipment. These are illustrative figures, not benchmarks.
The start-to-submit rate is 600 ÷ 1,000, or 60%. The submit-to-paid rate is 300 ÷ 600, or 50%. The start-to-shipped rate is 240 ÷ 1,000, or 24%.
Those calculations only work as sequential funnel measures if the counted outcomes belong to the same starting cohort and follow the defined path. Mixing this week’s starts with this week’s shipments from older patients breaks that relationship.
Choose a maturation window and label incomplete cohorts. A recently started patient may still be awaiting a normal review or shipment. Comparing that cohort with a fully matured one can make performance appear to decline when it has simply not finished yet.
Separate attribution from operational truth
Your order and payment systems describe recorded business events. Advertising platforms apply their own attribution rules to claim credit for outcomes. Those are different questions, so the totals do not have to match exactly.
Maintain a reconciled operational report for paid orders, refunds, new patients, and fulfillment. Evaluate channel reporting with documented attribution windows and models. Do not add together platform-attributed conversions as if they were necessarily unique people.
Use the patient LTV guide to connect acquired cohorts with subsequent contribution. A campaign that generates inexpensive intake starts may still attract poorly matched demand or costly support work.
Design measurement with privacy in mind
List every destination receiving events and the exact fields sent. Review identifiers, page paths, query strings, and free-text fields. A neutral-looking event can still carry sensitive context through its payload or URL.
HHS guidance on online tracking technologies addresses protected information in regulated contexts. Obtain a review appropriate to your organization and data flows. Moving collection to a server does not by itself make a disclosure permissible.
Keep diagnostic detail in appropriately controlled operational systems. Send only approved information to each analytics destination, and verify the actual network payload using synthetic test data rather than relying on a configuration label.
Use the funnel to prioritize work
Review unexplained changes by form version, program, device, and acquisition cohort where those dimensions are appropriate and available. Start with discrepancies that have a plausible operational explanation: broken uploads, delayed reviews, payment failures, or missing pharmacy updates.
The intake optimization guide helps investigate early-stage friction. Assign an owner to each suspected issue and confirm the fix against downstream results, not only the event count closest to the change.
Frequently asked questions
Is a purchase event enough to measure performance?
No. It can describe a payment milestone, but it does not capture refunds, clinical outcomes, fulfillment, or whether the patient is new.
Should we use one universal conversion rate?
Keep a headline measure if useful, but retain stage definitions and program-specific context. A single number can hide very different operational problems.
Why do ad-platform totals differ from our database?
Attribution rules, reporting windows, identity matching, event delivery, and refunds can all affect the comparison. Investigate the definitions before assuming either total represents the same thing.
What should every dashboard include?
Show the unit counted, cohort dates, observation window, exclusions, and event definitions. Without those labels, a precise-looking percentage can be easy to misinterpret.
