Results-Based Compensation
Status: concept, vision — the deal structure PRISM will pilot with its first client.
PRISM charges nothing for access, suggestions, or membership. It is paid only when a suggestion leads to a documented early detection — and then as a pre-agreed percentage of the savings that detection produced.
The payment chain
Every payment traces a complete chain of evidence through ordinary claims data — the same data PRISM reads, with no new infrastructure on either end.
| step | event | where it is visible |
|---|---|---|
| 1 | The ensemble reaches consensus on a patient and the suggestion goes to the primary-care physician through existing insurer channels | PRISM's own batch records |
| 2 | The physician, exercising their own judgment, orders the suggested screening test | a claim for the test, carrying the PRISM-specific tracking code |
| 3 | The result leads to diagnosis and treatment of the condition | follow-on claims showing early-treatment codes for the same patient |
| 4 | The pre-agreed percentage of the documented saving is paid | the contract's settlement, computed from the claims trail above |
If the chain breaks at any point — the physician declines, the test comes back negative, no treatment follows — PRISM earns nothing and the client owes nothing. A suggestion is a flag, not a billable event. Every dollar PRISM does earn corresponds to a specific documented detection that both parties can point to in the claims record. And because the entire trail lives in the claims stream the insurer already processes, verifying an outcome is a database query, not an audit project.
A pre-agreed share of documented savings
Compensation percentages are fixed per condition before deployment, negotiated from actuarial data on the cost difference between treating that condition early and treating it late. Different conditions carry different economics — the magnitude of the complications prevented, the cost of the early treatment pathway — so each gets its own percentage rather than a blanket rate. The insurer retains the majority of every documented saving, so its net cost falls with each successful detection; PRISM's revenue exists only inside value that has already been created and measured.
Fixing the percentages up front makes the arrangement modelable on both sides: the insurer can project net savings against its own population and prevalence figures before committing anything, and can verify afterward that it kept the larger share of every saving it paid on.
The honest limit: "documented savings" are necessarily counterfactual — the estimated cost of the late course the patient did not experience. The estimation methodology is agreed in the contract, per condition, before the first suggestion is ever sent, and is not re-litigated case by case. Projecting treatment costs from claims data is standard actuarial work for an insurer; accepting that the projection stands in for an unobservable alternative is the price of a compensation model with no other revenue source, and PRISM considers it a price worth paying.
One code, two jobs
The linchpin is deliberately small: a PRISM-specific tracking code — a billing modifier — appended to the test order when a physician acts on a PRISM suggestion. Modifier codes are routine; billing systems already carry, store, and report them. This one does two jobs at once.
| job | what it does |
|---|---|
| outcome audit trail | marks the test as PRISM-prompted, so that follow-on early-treatment claims for the same patient complete the payment chain; the same trail supports ongoing performance monitoring and, in the production vision, outcome feedback into retraining |
| patient cost-coverage flag | tells the claims system that this screening carries no out-of-pocket cost for the patient — no copay, no deductible, no balance |
The coverage half is not a courtesy; it is load-bearing, for one ethical reason and one practical one. Ethically: the financial benefit of a PRISM-suggested screening accrues first to the insurer, so asking the patient to pay toward it would mean charging them for someone else's savings. Practically: even modest cost-sharing measurably suppresses screening completion, and an unperformed test breaks the chain for everyone — the patient loses the early detection, the insurer loses the saving, and PRISM loses the payment. Removing the cost barrier is the cheapest intervention in the entire system, and the tracking code delivers it through billing machinery that already exists.
Why not fee-for-service
Every conventional pricing model for healthcare technology pays for volume, and volume is exactly the wrong thing to reward here.
| model | revenue scales with | what PRISM would be paid to do |
|---|---|---|
| per-suggestion fees | suggestion volume | lower consensus thresholds and flood physicians with marginal flags |
| subscription | contract renewals | optimize engagement and retention rather than detection |
| per-member per-month | covered lives | nothing in particular — revenue is identical whether suggestions work or not |
| results-based (chosen) | documented early detections | catch real conditions early, and only that |
The failure is the same in each rejected row: revenue detaches from whether patients were actually helped. Per-suggestion pricing is the starkest case — it would pay PRISM to erode its own quality controls until physicians learned to ignore the noise, destroying the one communication channel the system depends on. Results-based pricing has its own obvious temptation, over-suggesting in the hope that something sticks — but over-suggestion doesn't pay either: a test that finds nothing triggers no early treatment and completes no chain, while still incurring a covered screening cost. And any suggestion must first pass a physician who keeps full authority over the order, so the only volume that ever reaches a claim is volume a clinician judged worth acting on. Restraint is priced in.
The deeper alignment is architectural rather than contractual. PRISM is a recall instrument: it earns only by catching conditions early, and it structurally cannot earn by gatekeeping care, because no pathway exists for it to recommend against a test or deny anything. There is no cost-avoidance-through-restriction product to sell, even if someone someday wanted to sell one.
The zero-outlay pilot
The first-client engagement is shaped so the client risks operating trust, not budget. The client commits two things: access to its own claims history and a place to run — power, network, and a secured space for PRISM-managed on-premises hardware. No license fee, no integration project, no per-member charge. Payment begins only when the chain above completes on real patients: suggested test, tracking code, early treatment, documented saving.
PRISM carrying the financial risk is deliberate at this stage. The evidence behind the method is the 2026 synthetic proof of concept, which proved the machinery works given a clean, manufactured precursor pattern — it did not prove that real conditions leave patterns as learnable. A vendor asking a client to trust that leap should not also be asking for money up front. The pilot's job is to replace the leap with documented outcomes; until it does, nothing is owed.