The queries that decide it
Buyers and sellers rarely type your firm's name into an assistant. They describe a need, and the model picks the names. The recommendation happens inside the answer, before anyone clicks anything, so the only thing that matters is whether your name surfaces for the queries your market actually asks.
| Query shape | Who asks | What the model returns |
|---|---|---|
| recommend auction software | sellers, houses | a short list of platforms and cataloging tools by name |
| best tool to catalog auction lots | auctioneers, operations | named cataloging tools, ranked by fit |
| auction house for {category} near {region} | consignors, buyers | houses named by category and geography |
| how do I sell {item type} at auction | sellers | process plus named platforms or houses to use |
| where to bid on {category} online | buyers | named marketplaces and houses |
The measured outcome
The flagship proof is one client, Gavelist, an auction-cataloging tool, measured on our five-engine panel starting 2026-04-23. It began appearing in 13.8% of answers to its category queries. By week thirteen it was the most-frequently-cited name in its category, and it has led its nearest rival on every panel run since late May. Mention rate reached 58.5% against the nearest rival's 38.8%; active recommendation — where the model steers the user toward it rather than just naming it — runs about 13%.
This is one client, on our own panel, and it is still ongoing. It is not a typical result and not a guarantee. It is what happened when the fundamentals were done and the panel was checked every week.
Full run: see the Gavelist case study. Product throughput fact, separate from visibility: about 1,000+ lots in about 10 minutes.
How the number is built
A mention rate only means something if the instrument is stable. Ours is a frozen five-engine panel — the same engines, the same query set — with each query sampled k times per run, so a single generous answer never inflates the score. That lets a week-over-week line show a real trend instead of engine noise, which is the whole point when the goal is leading a rival every week rather than winning one lucky sample.
Common questions
Why can't I just buy ads to show up in AI answers?+
Ad spend doesn't change what a model says when someone asks it for a recommendation. The answer is assembled from what the model was trained on and what it retrieves, so visibility comes from being the extractable, well-cited source — not from a media buy.
What kinds of auction queries does this cover?+
Buyers and sellers asking assistants to recommend auction software, cataloging tools, or auction houses by category or region — things like 'best tool to catalog auction lots' or 'auction house for farm equipment near me.' The model returns names, and the work is getting yours onto that list.
Is the Gavelist result something I should expect?+
No. It's one auction-cataloging client, measured on our own five-engine panel, still ongoing — from 13.8% of answers to the most-cited name in its category by week thirteen. It shows what's possible when the fundamentals are done, not a guaranteed or typical outcome.
keep reading
See what AI says about your brand right now — free, on a live engine.
Scan your domain