Working Ideas

Analyst Rerun Test

A value model passes the Analyst Rerun Test when a payer's own analyst could rerun it from your documentation and land in your range.

Michael Alden · First published November 6, 2026

Question it answers

When a health system tells a payer what its access program is worth, will the payer believe the number?

Payers see many savings claims, and their analysts rebuild the ones that matter. In October I wrote that the metrics behind a differentiation claim need to hold up when a payer's analyst reruns them. The test makes that a standard a team can check in advance.

How to use it

  1. Document the method next to the range: data sources, time window, definitions, and each assumption with its value, source, and date.
  2. Choose conservative assumptions and show how each one moves the result.
  3. Price avoided care with unit costs the payer recognizes, ideally its own.
  4. Ask an analyst who did not build the model to rerun it from the documentation. If they land in your range, it is ready for a payer.

One method built to pass compares triage intent before and after. For each triage user, record where they planned to seek care before triage and afterward. Price the difference, then discount it for the factors that limit follow-through. The result is a range of potential cost avoidance with its assumptions in view.

In practice

A clinically integrated network uses that method for its virtual-first triage. Through August 2026, the model put potential cost avoidance at $0.6 to $1.7 million on conservative assumptions, and up to $4.4 million at the high end. These are modeled figures, and actual capture depends on patient behavior and the payment model behind each patient's care. In my experience, payers give real weight to a conservative number they can audit.

Where it breaks

Stated intent overstates behavior.
Paschal Sheeran's review of the research documents the gap between intentions and behavior ("Intention–Behavior Relations," European Review of Social Psychology, 2002). The follow-through discount is a judgment a payer's analyst may set differently. Linking triage records to claims narrows the gap.
A model can be rerun and still be wrong.
The test checks transparency. Accuracy comes from conservative assumptions and, later, claims. Most Medicare disease management and care coordination demonstrations did not reduce Medicare spending (Congressional Budget Office, 2012).
The payer has to see the data.
A model built on records the payer cannot access fails by design. Share the data under the contract, or build on the payer's claims.
Avoided cost needs someone to capture it.
Cost avoidance becomes savings only under a payment model that shares it, the subject of the Attribute-Steer-Pay test.

What it borrows