Trust Status at a Glance: Seeing Which Numbers Are Verified, Internal, or Plan

Knowing that Rexfin ranks evidence is one thing. Seeing that ranking on the screen, at the moment you’re looking at a number, is another. A CFO scanning a dashboard shouldn’t have to remember which figures came from a filing and which came from a spreadsheet: the interface should just show it.

That’s what the trust status layer does. Every number carries a visible tier: cited, internal, or plan. Click it and a Trust Inspector opens, showing exactly why it landed in that tier.

What the inspector actually shows

The inspector isn’t a new drawer bolted on top of the product: it’s the same provenance drawer that already traces a number back through extraction backends, reconciliation, citation, and integrity checks, now carrying a plainer summary on top. Open it and you get a gates checklist: did independent extraction backends agree, did the figure reconcile to the filing’s own printed total, is it citable to a specific document and page, did the integrity check pass, and has it been through review sign-off. Each gate reads pass, fail, or not applicable, not a color alone, so the status is legible even if you’re not relying on color to read it.

Alongside the gates sits the review status and any caveats attached to the number. A figure can pass every mechanical gate and still carry a caveat: for instance, a filed number from a source grade below full audit (interim, draft, or an unrecognized grade) is treated as advisory, not verified, and the inspector says so rather than rounding it up.

Why cited has to be earned, not assumed

The tiering logic is deliberately strict about what counts as cited. A number only earns that tier when it traces to a filing grade Rexfin recognizes as filed and audited. Anything it can’t confirm (an unrecognized grade, a broken link in the chain, a gap in the trail) floors to plan rather than being waved through. A number from a filed source that hasn’t cleared verification reads as “filed source: unverified,” never as verified. That asymmetry is the point: the system is built to under-claim rather than over-claim, because a confident label on an unverified number is worse than an honest one.

The same tiering reaches numbers that don’t have a document behind them at all. A budget-vs-actual comparison has no source atom for the budget line (it’s plan by definition) while the actual carries whatever tier its own upload earned. A quick popover on the cell shows both sides without forcing you into the full inspector for something you can absorb in a glance.

Nothing about this is inferred in the interface. The tier, review status, and gate results are computed once, on the server, from the same record the inspector displays: the screen shows what was decided, it doesn’t recalculate its own opinion of a number.

Rolling it up to a workspace view

A single number’s tier only tells you about that number. The Trust Score is the workspace-level version of the same question: across everything Rexfin has ingested for a company, how much of it is actually cited and verified? It’s a deterministic read of what’s already in the cache (no model call, no scoring judgment) covering verified-citable coverage, how many extracted figures remain unverified, how many sources are still pending review, how many filings have gone stale or been superseded, and how much of the picture is plan rather than filed. It surfaces as a card on the dashboard and as a fuller detail page, with the metrics that are dragging the score down ranked so it’s clear what to fix first.

Who this is for

This is for anyone who has to trust a dashboard before they trust the person presenting it: a controller reviewing a statement grid before a board pack goes out, or an auditor scanning for what’s still unverified before signing off. The tier and the score exist so that question can be answered by looking, not by asking. See how it fits with the rest of the platform in the Rexfin product tour.