Healthcare Payor Identity Cost Per Case | Moring
Cost per case gives a healthcare payor the true unit economics of agent work: what a prior authorization packet, an adjudicated claim, or a member interaction actually costs in tool calls, tokens, and retries. Agent identity is what makes the number computable, because only a per-agent identity ties every input to the case it served.
Short answer
Cost per case gives a healthcare payor the true unit economics of agent work: what a prior authorization packet, an adjudicated claim, or a member interaction actually costs in tool calls, tokens, and retries. Agent identity is what makes the number computable, because only a per-agent identity ties every input to the case it served.
How it works
Cost per case is the number a payor's finance organization actually manages by, and agent identity is what makes it computable. An agent supporting prior authorization touches eligibility systems, clinical criteria, and claim history within one packet, invoking multiple backend tools; the cost of that packet is the sum of every call and token attributed to the case. Without per-agent identity, those inputs pool under shared accounts and platform totals, and cost per case remains an estimate nobody can defend.
The case itself has to be the attribution key, layered on the agent identity. Each run should carry the case or claim identifier alongside the agent credential, so records show not just which agent acted but which packet it served. This pairing is what lets a payor answer the operational questions that matter: how the cost of an agent-adjudicated claim compares with a manually handled one, how cost per packet varies by service line, and where retries inflate the number. Attribution without the case key produces agent totals; the pairing produces unit economics.
Retry behavior is where cost per case breaks or holds. A timeout to the eligibility service halfway through a multi-step adjudication restarts the whole run, so the same case consumes double the tool calls and data access. Health plans running agents report exactly this pattern, and a payor without per-case attribution absorbs the duplicate spend invisibly inside a monthly total. With attribution, the retry is a flagged event on a specific case, and the platform team can address the timeout rather than the symptom.
Cost per case also exposes the quality side of the ledger. A low unit cost means little if the case comes back: rework, appeals, and corrections are cases too, and they belong in the same attribution scheme. A payor that attributes every agent run, including reruns, sees the true cost of an agent whose first-pass accuracy is drifting, and can compare that trajectory against the cost of adding review capacity. The measurement is only honest when failure paths are counted as work.
The comparison set matters as much as the measurement. Cost per case gains meaning against benchmarks the payor already holds: manual handling times, vendor processing fees, and the caseload each workflow carries. Those comparisons are what turn observability into budgeting, and they depend on attribution holding steady across volume. A claims season that doubles packet volume should double attributed cost proportionally; if it does not, the identity layer is leaking, and the leak is worth finding before finance loses confidence in the numbers.
For payors making the automation case, cost per case is also the credibility number. Programs live or die on whether leadership believes the agent's unit economics, and belief follows from records: which agent, which case, which tools, which tokens. A payor that can produce that trail for a sample of cases can defend its automation investment; a payor producing averages cannot, and averages are what shared service accounts leave behind.
Frequently asked questions
Why does cost per case need agent identity behind it?
Because a case's cost is the sum of tool calls and tokens, and only a per-agent identity ties each input to the case it served. A prior authorization packet touches eligibility, clinical criteria, and claim history in one run. Without attribution, a healthcare payor has platform totals and estimates; with it, the unit economics of each workflow become real numbers.
How should retries be treated in payor cost per case?
As attributed work on the same case. A timeout against the eligibility service midway through adjudication restarts the run, doubling tool calls and data access, and health plans report this pattern. Per-case attribution turns the retry into a visible event on a specific packet, letting the payor fix the timeout instead of absorbing duplicate spend inside a monthly total.
Should failed agent runs count toward cost per case?
Yes, including reruns, rework, and appeals, because they are cases the agent consumed budget on. Attribution that counts only first-pass successes hides the true cost of an agent whose accuracy is drifting. A healthcare payor counting failure paths sees when review capacity would be cheaper than the agent's error rate, which is exactly the comparison the number exists to support.
What benchmarks make cost per case meaningful for a payor?
Manual handling effort, vendor processing fees, and caseload for the same workflow, all tracked alongside the agent's attributed spend. A volume spike should move attributed cost proportionally; if it does not, the identity layer is leaking attribution. Steady, defensible unit economics are also what make the automation case credible to a healthcare payor's finance leadership.