Part of The Revenue Model Sets the Strategy
The Engagement Gate
No stakeholder sees value until the end user engages.
01
Why nothing else in the framework matters first
Every revenue model in digital health, however well structured, rests on one prerequisite. The end user has to actually use the product.
That sounds obvious. It isn't, and the reason is structural.
In traditional B2B SaaS, the end user is an employee who has been told to use the software. A manager bought it, IT deployed it, and there's an implicit expectation that they'll log in and do their job. Activation is a training problem.
In digital health, the end user is usually a patient or a health plan member who never asked for the product. A doctor mentioned it, or a health plan sent an email, or an employer added it to a benefits portal nobody visits. No manager is asking. No workflow requires it. The decision to engage is voluntary, and the default is to ignore it.
That threshold is the Engagement Gate, and it has to be crossed before any stakeholder anywhere in the chain sees value.
The clinic administrator can't see CPT reimbursement if patients aren't using the device. The health plan executive can't see cost savings if members aren't enrolling. The employer can't see reduced absenteeism if employees never download the app. The pharma partner can't see adherence data if patients aren't using the companion tool. Every value driver mapped in this framework, for every stakeholder, traces back to whether someone at the end of the chain opened the app, used the device, or completed the enrollment.
02
Activation and adherence are different problems
Crossing the gate isn't a single event. It takes two distinct phases, and collapsing them into one number is the most common way CS teams misread their own data.
Activation is getting the user to engage at all. It's typically a day 0 to 30 challenge, and it's a marketing and behavioral design problem: personas, channel strategy, message testing, campaign development.
Adherence is keeping them engaged over time. It's a day 30 to 365+ challenge, and it's a clinical and product design problem: habit formation, value reinforcement, friction reduction.
The plays aren't interchangeable. An employee who registers, completes onboarding, and stops opening the app by week six crossed the activation gate and failed the adherence gate. A single blended engagement number gives that employee partial credit, and across a book of business that is how a stage that collapsed reads as a program that is working. The revenue number shows nothing at all, because under a per-employee-per-month contract the fee is owed on the eligible population whether or not that employee ever returns. The failure surfaces at renewal, when the employer asks what a year of fees bought.
Most teams reporting a blended number aren't choosing it. It's what the analytics stack returns by default, and producing stage-level rates means someone has to define the stages and instrument them before anyone can read them.
In models where sustained usage drives revenue or determines renewal, that distinction separates short-term reporting success from long-term contract failure.
03
The Gate is a sequence, not a door
Activation and adherence name the two phases. Underneath them sit six positions a person can occupy, and knowing which one they are stuck at is what turns a bad engagement number into a diagnosis.
- eligible,
- reached,
- activated,
- adherent,
- outcomes visible,
- and value realized.
Eligible. The person sits inside the contracted population. Eligibility is opportunity, not value.
Reached. An invitation or referral actually arrived. Outreach volume is not reach. Undeliverable contact data and dead channels drop people here silently.
Activated. The first meaningful action happened: an account created, a device paired, a first session completed.
Adherent. The behavior continued long enough and often enough for the program to work.
Outcomes visible. Participation, duration, and data quality are sufficient to show clinical, financial, or operational change.
Value realized. Those results answer the question the buyer was actually asking. Outcomes that cannot be attributed or translated into the buyer's economic terms can still fail to support a renewal.
The diagnostic value is in locating the drop. If reach is strong and activation is weak, the problem is the offer, the onboarding friction, or the positioning. If activation holds and adherence falls away, look instead at product fit, burden, and clinical support. And when engagement is strong but no outcome is visible, the constraint is usually measurement design, data completeness, or attribution, which means the activation work has been landing and getting blamed for a gap somewhere else.
04
Measuring the Gate honestly
Any activation figure is meaningless without a named denominator, and any retention figure is meaningless without a named numerator. A rate is two claims, and both of them are choices.
The denominator half. The same program can report engagement rates that differ several times over, and in payer contracts where the eligible book dwarfs the engaged population, by more than an order of magnitude, depending on which of those six stages sits under the line. Divide activated users by the eligible population and the answer is a penetration number. Divide by the people actually reached and it becomes a conversion number. Divide by the people who began enrollment and it becomes an onboarding number. All three can be accurate at once, and only one of them answers the question the buyer is funding.
This matters more in digital health than in SaaS for two reasons. The eligible population is usually enormous relative to the engaged population, so denominator choice swings the result far more than it would in non-healthcare SaaS. And the buyer and the vendor frequently hold different denominators without either party realizing it. A health plan reporting against its full eligible book and a vendor reporting against its outreach-contacted cohort can both be honest and still be describing different programs.
What one funnel produces
The funnel below is the standard shape: total population, targeted subset, activated members, engaged members, with two rates hung off the side. Versions of it appear in vendor collateral across the market, almost always without the definitions attached. This one comes from a payer-facing white paper that prints them, which is the only reason it can be examined at all. The member counts are round enough to read as scaffolding rather than client data. The rates beside them are different: a footnote attributes those to partners who implemented the vendor's best practices.
The funnel runs: 20,000 total members in the population, 15,000 targeted with communications, 5,000 who took action, and 2,500 still engaged after 365 days.
- 20,000total members in the population
- 15,000targeted with communications, labeled reached in the same graphic
- 5,000took action
- 2,500still engaged after 365 days
of those who activated
of those who were targeted
of all members
Most of those terms are defined. Targeted members are the ones eligible for the program that the plan intends to reach. Activation rate is activated divided by targeted, and utilization rate is engaged divided by activated. Activated members are the exception. The only guidance on what counts is a parenthetical carrying two examples, "went to a landing page, created an account," with no operator between them. A page view and an account creation aren't the same commitment, and nothing says whether either one alone is enough to put a member in the 5,000. On those definitions the program reports roughly 30% activation and 50% utilization.
Now take the 2,500 people still engaged after a year and divide them three ways.
Against the 5,000 who activated, that's 50%. Against the 15,000 who were targeted, 16.7%. Against the full 20,000 members, 12.5%.
The same 2,500 people. A program described as 30% activation and 50% utilization delivers sustained engagement equal to one member in eight. Every one of those three rates is arithmetically correct, and the math can be reconstructed from the document. The buyer paying a per-member fee across the full population is asking the 12.5% question.
The 15,000 isn't one stage either. The funnel graphic labels that tier twice, once as targeted members and once as members reached with communications. Targeting is intent and reach is arrival. The gap between them is the reach stage above, and collapsing the two produces a denominator that assumes every message landed. Two of the six stages above, fused into one number, with the published activation rate dividing by it.
In enrollment-gated contracts this stops being a reporting question entirely. There, one stage of the funnel is the billing denominator, defined in the agreement as a specific set of completed actions, and the gap between registration and enrollment is measured in dollars rather than in percentage points. The PMPM and PEPM model page covers how those definitions get written and what to settle before signing.
The drift is easy to watch happen. The same white paper defines activation as a share of targeted members on one page, then several pages later describes activating as much as half of an employer's workforce. Workforce is the whole population, not the targeted subset. Two different denominators, one word, one document, no bad faith anywhere in it. A performance chart elsewhere in the same paper tops out at 38% activation at month 24, which is a third reading again.
This document isn't an outlier. It's more forthcoming than most, which is the only reason any of this is legible. The questions it raises apply to any activation figure a reader has in hand right now. Which population sits under the line, and what single behavior puts a person above it? Were the rate and the funnel printed beside it calculated from the same data? The rules below are how to have those answers before a buyer asks for them.
The numerator half. The same trap sits on the other side of the line. End-user retention can be measured as bounded or unbounded, and the gap between them is a definitional choice rather than a performance difference.
Bounded retention asks whether the user was active in a specific window. Day 30 means day 30. Unbounded retention asks whether the user was active on day 30 or any day after, so a member who goes quiet for four months and returns in month five counts as retained at day 30 in hindsight.
Neither is the honest one. The right measure depends on how the product is meant to be used. A daily glucose logging tool should be measured bounded, because a patient who stops logging has stopped doing the thing the program depends on. An on-demand mental health service should probably be measured unbounded, because a member who needed support in March, didn't need it over the summer, and came back in September was well served. Bounded retention would score that member as churned for the crime of doing better. Episodic products get punished by daily-habit measures, and daily-habit products get flattered by episodic ones.
Bounded at day 30
Active on day 30?
Not retainedThe later return leaves the day 30 result unchanged.
Unbounded at day 30
Active on day 30 or any day after?
RetainedThe month five return counts in hindsight.
The tell isn't which calculation a vendor uses. It's when they chose it.
Four rules keep this defensible:
- Name both halves every time a rate appears, in the QBR deck and in internal reporting. A bare "72% retained" gives the reader nothing they can check.
- Fix the definitions in the contract or the implementation plan, not at the first business review, because renegotiating them after a bad quarter looks like exactly what it looks like.
- Match the retention measure to the intended usage pattern, and pick it before the data arrives. A definition revisited after a soft quarter is a different kind of number entirely.
- Report the absolute counts alongside every rate so the reader can reconstruct the math. A percentage with no N behind it formats identically whether it describes eighty thousand people or four, and a reporting template will render both the same way without flagging the difference.
05
What makes the Gate hard to open
The obstacles are consistent across digital health, whatever the product.
The user never chose it. A doctor mentioned it, or a benefits portal listed it. Nothing in the situation creates obligation.
Friction is fatal early. Someone with no stake in the tool won't push through a bad first experience.
Access varies. Device availability, connectivity, language, and literacy differ across any covered population, and one activation design won't reach all of it.
Trust comes before engagement. Health isn't a category where people extend the benefit of the doubt to an unfamiliar sender.
Outreach is constrained. Anything touching personal health information limits channel, timing, and message content in ways consumer marketing never faces.
These compound differently depending on the population being served. A commercially insured employee population and a Medicaid population using the same product face different barriers and need different activation designs. Treating them as one segment is how activation programs underperform without anyone being able to explain why.
06
The Gate shifts by model
The Engagement Gate is the unifying concept, but the specific behavior that has to change, and who has to change it, varies by revenue model.
| Model | Who must change behavior | What crossing actually looks like | Lag to the revenue signal |
|---|---|---|---|
| CPT / RPM Reimbursement | The patient using the device, and the provider documenting qualifying interactions | Sustained device days plus documented clinical time, not just enough usage to trigger a code | Near immediate, visible monthly |
| PMPM / PEPM | Eligible members or employees enrolling, then staying engaged | Enrollment followed by sustained use across the contract year | 24 to 36 months, surfacing at renewal |
| SaaS Platform Licensing | Clinicians and care managers adopting a new workflow | Weekly active use inside the clinical workflow, not license assignment | Roughly 12 months, surfacing at renewal |
| Value-Based / Outcomes-Based | Enough enrolled members for the evaluator to validate the result, then sustained engagement | Enrollment volume plus adherence sufficient for the agreed measurement methodology | 18 to 24 months, surfacing at the measurement period |
| Hybrid | Shifts as contract phases evolve | Depends on which stream is dominant in the current phase | Varies by stream; map each one separately |
07
The lag is the thing worth knowing
The gate governs value realization and renewal. It doesn't always govern this month's revenue, and the distance between those two things varies enormously.
- CPT / RPMNear immediate
- SaaS Platform LicensingRoughly 12 months
- Value-Based / Outcomes-Based18 to 24 months
- PMPM / PEPM24 to 36 months
In RPM, engagement converts to revenue almost instantly. Every month, every patient either generates a billable event or doesn't. The feedback loop is short enough to be self-correcting.
In PMPM, revenue can flow for 24 to 36 months with the gate barely open. The invoice arrives on schedule whether or not anyone engages. The signal comes later, in a renewal conversation that was decided before it started.
That lag is the single most useful thing to know about a given model, because it determines how much runway a team has to course correct and how early the warning systems need to fire. A short-lag model corrects itself. A long-lag model looks healthy on a revenue dashboard right up until it isn't, which is why the leading indicators in PMPM and value-based accounts have to be engagement metrics rather than financial ones. By the time the financial metric moves, the decision has already been made on the buyer's side.
08
What this demands of the CS function
The gate is why end-user activation isn't a customer success problem in the traditional sense. It's a marketing and behavioral design problem living inside the CS function. It needs a marketer's toolkit, meaning personas, channel strategy, testing, and multichannel campaign development, executed against contract performance rather than campaign performance.
It's also the clearest reason digital health customer success operates as a commercialization discipline rather than a post-sale one. That argument sits on the framework hub.
09
Where to go next
Each revenue model deep dive covers how the gate behaves inside that model and gives an example drawn from practice. Four of them also map the stakeholder value drivers and the shifts the buyer's payment environment produces. The hybrid page maps contract structures instead, since a hybrid inherits its stakeholders from the models it combines. Start with the model that carries the majority of revenue.
The three buyer lens pages come at it from the other direction, covering how pharma, health plans, and employers each define value and what each one accepts as proof.
Take the working structures with you
The Field Kit pulls the framework's durable layer out into structures built for use rather than reading: the identify-your-model diagnostic, blank stakeholder and KPI worksheets, the tiered QBR builder, the hybrid trade-off script, the believable-proof checklist, and the seven-step guide to putting the framework to work on a live account. Version-stamped, designed to print, and free to subscribers, who get it first when it ships. Billing-code specifics stay on this site by design, so the kit stays accurate as the rules change.
Your email delivers the kit when it ships, plus occasional new insights from Digital Health Success. No spam. Unsubscribe anytime.