How vendor-research rankings actually work

Every ranking your agent produces is a weighted scorecard, not a black box.

Here's exactly how scoring, re-ranking, head-to-head comparison, refreshing, and correcting a result all work.

Weighted scorecards you control

Every vendor is scored on the same rubric. You decide what matters most.

Every vendor is scored, criterion by criterion, against the same rubric — pricing, integrations, support, whatever matters for this purchase — and those per-criterion scores combine into one comparable overall score. That's a weighted scorecard, not a gut call.

You set how much each criterion matters, from 1 to 5. Change a weight and every vendor's score recomputes and the ranking re-orders immediately — no new research run, no waiting on a fresh crawl.

Example
"You said price mattered more than integrations. I reweighted the field — here's who moves to the top now."
A rubric, not a gut call.

A rubric, not a gut call

Every vendor scored against the same criteria.

Weights you set

Decide how much each criterion matters to your business.

Instant re-ranking

Change a weight and watch the field re-order.

No new run required

Reweighing recomputes existing scores — it doesn't re-research anything.

A vendor research's Strength Matrix: a stacked score chart over five vendors, then a table with one row per criterion showing its weight badge, each vendor's score with a trophy for the leader, and a Leaders column
The Strength Matrix. One row per criterion, its weight, every vendor's score, and who leads it. Totals at the bottom.
The Edit criterion weights dialog listing each criterion with a description and a 1 to 5 weight field
Edit weights. Set 1 to 5 per criterion and save. Scores recompute and the ranking re-orders without a new run.
Head-to-head against named competitors

See exactly where you're ahead, and where you're behind.

Beyond ranking a field of vendors, you can see your own company scored directly against specific named competitors, criterion by criterion, side by side — a strength matrix. It shows where you're ahead, where you're behind, and by how much on each criterion, not just an overall score.

It's the same comparison whether you're sizing up vendors to buy from or checking how you stack up against competitors surfaced for your own company — one mechanism, two jobs.

Side by side, not just a score.

Named, not generic

Compared against the specific competitors you care about.

Criterion by criterion

See exactly where the gap is, not just a single number.

One mechanism, two jobs

The same head-to-head view works for vendors you're evaluating and competitors you're up against.

The Market module's Where you win section: your company scored against five named competitors, category by category, with trophies marking each row's leader
Where you win. Your company against named competitors, category by category. The same matrix, from the other side.

Available on Business plans and above — see pricing.

Refresh on demand

Rankings aren't static, and they aren't on a timer either.

You decide when to refresh — pull fresh data on a single vendor, or rebuild an entire evaluation from scratch — and your agent re-crawls the sources and re-scores on the spot. Nothing re-evaluates itself in the background without you asking.

You decide when it's worth re-checking.

One vendor or the whole field

Refresh a single vendor's data, or rebuild the entire evaluation.

Fresh sources, re-scored

A refresh re-crawls and re-scores, not just re-renders old numbers.

You decide when

No background re-evaluation running without you asking for it.

The Rebuild research dialog over a vendor research page, with the title and requirements brief, a note that an unchanged brief just re-scores the existing vendors, and a Use cache checkbox
Rebuild research. Keep the brief to re-score the same vendors, or change it to re-discover the field. Nothing runs until you click Rebuild.
Correct it, and revert if you change your mind

Your agent's work is checkable, and correctable.

If your agent gets a fact about a vendor wrong, you can correct it. The correction is applied and timestamped, and if you change your mind, you can revert to what was originally found — your correction never permanently overwrites the source data.

Fix it, and you can always go back.

Fix what's wrong

Correct a fact your agent got wrong about a vendor.

Timestamped

Every correction records when it was made.

Always reversible

Revert to the original source data whenever you want.

The Company page's organization card marked Corrected, with a pencil button to correct it and a revert button whose tooltip reads Revert to crawled data
Correct and revert. The pencil records a correction; the revert control puts the crawled data back.

Questions people ask

Can I change how much a criterion matters?

Yes. Set the weight on any criterion, from 1 to 5, and every vendor’s score recomputes and the ranking re-orders immediately — no new research run.

Does a ranking update automatically over time?

No — refreshing a vendor or rebuilding an evaluation is something you ask for, not something that happens on a timer in the background. You decide when it’s worth re-checking.

Can I see my company compared directly against one named competitor?

Yes, criterion by criterion, side by side — a strength matrix. It’s the same comparison whether you’re sizing up vendors to buy from or checking how you stack up against competitors surfaced for your own company. Available on Business plans and above.

What happens if the AI gets a fact about a vendor wrong?

Correct it. The correction is applied and timestamped, and you can revert to what was originally found if you change your mind — your correction never permanently overwrites the source data.

Is this the same as AI-visibility tracking?

No. This page is about how a ranking or comparison is scored. Tracking how often your own company is cited by AI as a vendor versus a competitor, over time, is a separate feature — see what your agent does.

See it on a real purchase.

Tell your agent what you're looking for and watch the scorecard build itself.