Running a telehealth program at scale, Part 1: unit economics at 1,000 encounters a month

Start from the encounter, not the patient

Most telehealth models are built around the patient: acquisition cost, subscription price, lifetime value. That is the right frame for deciding whether to acquire. It is the wrong frame for deciding whether the program is efficient, because a patient is a bundle of very different events — one initial encounter and a string of renewals — that cost different amounts to deliver. At volume, the encounter is the unit.

Two properties of the encounter drive everything else. The first is type: an initial encounter carries a full intake, a longer clinician review, identity and consent capture, and often a live-visit requirement in some states; a renewal carries a check-in and a shorter review. The second is outcome: an encounter that produces a prescription adds pharmacy, medication, and shipping; a decline or a labs-only encounter does not. Build the model on those two axes and the rest falls into place.

The model

The lines below are the structure. The figures are illustrative inputs to show the arithmetic, not benchmarks — replace every one of them with your own.

Two lines deserve a note. Medication is a pass-through in most cash-pay programs: the patient pays for it and the program earns on care, so keep it visible but separate from margin on the encounter. And acquisition is deliberately absent — it belongs to the patient model, amortized over expected renewals, not to the encounter.

Fixed costs are what volume fixes

Take the fixed lines together — licensing and compliance, engineering, operations staff, the fixed part of platform fees — and call the total F. At 300 encounters a month each encounter carries F/300 of overhead; at 1,000 it carries F/1,000. If F is $30,000 a month, that is $100 per encounter at 300 and $30 at 1,000. Nothing about the encounter changed. That arithmetic is why programs that look hopeless at 300 become healthy at 1,000, and why the strategic question for a team at 500 is how to reach 1,000 without letting F grow in step.

F grows in step when scale means more vendors, more integrations to maintain, more states to keep licensed, and more clinicians to manage directly. It stays flat when those are someone else’s fixed costs, priced to you per encounter. That is the actual argument for infrastructure at the mid-market stage: not that it is cheaper per encounter today, but that it converts your fixed costs into variable ones and lets volume do its work. One national platform we work with described it as seven vendor contracts becoming one API and one invoice.

Where the money goes at scale

Once fixed costs are amortized, the variable lines dominate, and they are surprisingly concentrated.

• Clinician minutes. The largest controllable variable cost. A clinician reconstructing a history from free text spends several times longer than one reading a structured, schema-validated intake with a drafted summary. Minutes per case is the metric.
• Renewal share. Renewals are cheaper to deliver and carry no acquisition cost. Every point of renewal share is margin. The enemy is the lapse — a patient who runs out before the next shipment — covered in Part 4.
• Support tickets. In most programs the plurality of tickets are “where is my medication.” Each one that reaches a human costs real money; each one answered from live order status costs nothing.
• Shipping exceptions. Reships after failed cold-chain deliveries cost the medication and the shipping twice. Routing and carrier choice by state decide the rate.
• Declines and needs-more-information. Encounters that do not complete still consume clinician time. A high rate means intake is letting ineligible or incomplete cases through.

The five levers

1. Raise renewal share by measuring and attacking lapse: a cadence engine that triggers the check-in before supply runs out, and a clinician queue that authorizes in hours.
2. Cut clinician minutes with structured intake, protocol pre-screening, and drafted summaries for renewals. This is where agentic workflows earn their keep.
3. Flatten fixed costs by consolidating vendors and buying the regulated layers per encounter.
4. Route pharmacy by state and medication to cut reships, and hold a second pharmacy for redundancy.
5. Make status event-driven so support answers from data, and route clinical questions to the care team rather than support.

Pricing models and how they scale

How your infrastructure charges you decides which of these lines are fixed. Per-member-per-month pricing scales with patients, not encounters, and penalizes high-renewal programs. Pure per-visit pricing scales cleanly but can hide minimums. Platform-plus-per-visit — the model most infrastructure uses, including ours — puts a small fixed line in and keeps the rest variable, which is the right shape for a program growing from 500 to 2,000. Our pricing guide walks through the four models and the questions to ask about minimums, tiers, and exit terms.

Building the model

A spreadsheet with one row per month and these columns is enough: encounters by type (initial, renewal, labs-only), prescriptions issued, each variable line as a total and per encounter, each fixed line as a total, fixed cost per encounter, fully loaded cost per encounter by type, and revenue per encounter by type. Add three operational metrics alongside — clinician minutes per case, tickets per 100 encounters, lapse rate — because they predict next month’s cost lines. Review it monthly. When a line moves, you will know which lever to pull.

Frequently asked questions