Work the claim that matters first.
A claims team carries hundreds of open files. AnantState tracks each claim’s condition and ranks the queue by money at stake and complexity, with the evidence for each decision kept on record.
In plain termsA claims queue ordered by what each file is worth.
How a claims queue gets ordered
An illustration of the two judgements that decide which file comes first.
The entities are the things a claim runs through: the policy, the claim, the claimant, the provider and the reserve held against it. Each has a condition that moves with the case mix and the season, and occasionally with something nobody scheduled.
Which claim first · an illustration of how a queue is ordered
- Claim AA lot Money at stake, Complex How complex
- Claim BA lot Money at stake, Complex How complex
- Claim CA lot Money at stake, Simple How complex
- Claim DLittle Money at stake, Complex How complex
- Claim ELittle Money at stake, Simple How complex
- Claim FLittle Money at stake, Simple How complex
- Positions show relative standing, not measured values.
What moves in a claim
| State | Baseline drivers | Unexpected drivers |
|---|---|---|
| Claim posture | Policy terms, claim type, season | An unusual pattern, organized activity |
| Reserve adequacy | Case mix, the usual development of a claim | A shift in severity |
| Leakage and fraud posture | Historical rates, review depth | A new scheme, a provider pattern |
| Backlog | Arrival rate, staffing | A catastrophe event, a surge |
Where it leaks
- Files are worked in arrival order. A large, complex claim waits behind small simple ones because nothing ranks them by what they are worth.
- Review is all or nothing. Without a price on a targeted review, teams either review too little or slow every file down.
- Decisions are hard to defend later. The evidence behind a settlement sits in several systems and in people’s memories.
A worked example
Eight thousand open claims, and a targeted review on one in ten.
Round, illustrative figures for open claims: they show the arithmetic, leave out the subscription, and are for you to change in the calculator.
| Input | Value | What it stands for |
|---|---|---|
| Open claims in scope | 8,000 | Everything the team is responsible for |
| Value at stake per one | $2,200 | Leakage on one claim if it is paid without review |
| Cost of the action per one | $90 | A targeted review or a request for evidence |
| Share acted on | 10% | The claims ranked highest by money at stake: 800 of 8,000 |
| Modeled, USD | Pessimistic | Expected | Optimistic |
|---|---|---|---|
| Share of value at stake avoided | 5% | 12% | 25% |
| Cost to act, in total | $72,000 | $72,000 | $72,000 |
| Avoided loss | $88,000 | $211,200 | $440,000 |
| Net position | $16,000 | $139,200 | $368,000 |
Break-even is 4.1%: each action costs $90 against $2,200 at stake, so it has to avoid that share of the loss to pay for itself.
What exists today
Insurance exists as a catalog template and as a starting point for an ontology workshop. It is not a live simulated world.
Rank a real queue
Bring a sample of open claims and what a review costs. We will show the order the money suggests.