The discipline

Honest about its own data.

A data product about honesty has to be honest about its own data. Every figure here is either counted or estimated, and we never let one masquerade as the other. This page is the whole discipline, and the reason you can trust the rest of the site.

Exact

Counted, not modelled

Dispute counts, provider win rates, market share, and each payer's benchmark position are counted directly from the public record. No model, no estimate.

Estimated

And it says so

A few figures — like a market's total disputed volume — are modelled estimates, not counts. Wherever one appears it is labeled estimated, with its caveat stated in plain sight.

Absent

Is not zero

A state the record never covered is drawn stippled, not colored pale. A quarter that does not exist is a hole in the timeline, and playback steps over it rather than through it.

What's counted, exactly

Dispute counts, provider win rates, market share, and each insurer's benchmark position are counted directly from the public record. These are not modelled and not smoothed — they are what the record says, tallied. Where a figure is exact, we show it as exact, and say so.

What's estimated, and labeled

Some views show a modelled figure — for example, a market's total disputed dollar volume. When they do, the number is explicitly labeled estimated and carries a stated caveat. An estimate is never dressed up as a count, and a count is never quietly turned into an estimate.

What we won't manufacture

Some numbers simply can't be recovered from public records with integrity. Where that's true, we say so and leave the number out rather than guess. Any product that reports a precise figure the record can't support is guessing — and we won't.

Absent is not zero. A value that was never published is drawn stippled, not colored pale — and a gap in the timeline is a gap playback steps over, never through.