Every AI visibility vendor promises they can get a business named by ChatGPT, Claude, Perplexity, and Gemini. We are the one that checks itself first and shows you the scoreboard, including the entries where the number is zero. This is that ledger.
Namebeam sells the discipline of checking whether AI names a business, fixing what is wrong, and proving the fix worked with a raw transcript instead of a claim. A company that sells receipts and hides its own receipts is a contradiction, not a business. So Namebeam is customer zero. We run the same checks we sell on ourselves, on the same dated schedule, and we publish every entry here whether the number went up or stayed at zero. No entry in this ledger is edited after the fact. If a score does not move, that gets printed too, at the same size as the entries where it does.
Each entry below states the method, names the raw files behind it, and discloses the honest limits of a single dated run. Nothing on this page is rounded up or smoothed over.
Namebeam was named zero times out of twenty answers. Zero is the starting line, and we are publishing it because it is what we would tell a client to do with their own number.
Four AI engines, ChatGPT, Claude, Gemini, and Perplexity. Five questions a small business owner would plausibly ask about AI visibility tools, one fresh run per question per engine, answers recorded verbatim. Twenty answers total. Raw transcripts on file under corpus check NB-0001.
Across all twenty answers, here is the shelf we are climbing onto, counted and dated.
| Named instead | Mentions across 20 answers |
|---|---|
| Profound | 12 |
| Otterly | 11 |
| Peec AI | 11 |
| Semrush | 7 |
| Scrunch AI | 4 |
| SE Ranking | 4 |
| ZipTie | 4 |
A tail of smaller, single mention tools followed these seven and is not itemized here. Source: CITATION_LEDGER_ENTRY_001_2026-07-29.md, corpus check NB-0001.
We repeated the identical five questions on the same three engines, ChatGPT, Gemini, and Perplexity, in brand new, memory free chats, to see whether the zero was a fluke of one conversation and whether the competitor names would repeat. Namebeam held at zero, now 0 of 35 across two runs. Claude's Run 2 leg could not be completed, the claude.ai chat surface would not submit the prompt across two separate fresh sessions, so those five rows are logged as failures rather than dropped, and Claude's only data point remains its Run 1 result.
Profound, Otterly.ai, and Peec AI held the top of the leaderboard in both runs, that repeatability is a real signal. The names below the top three reshuffled noticeably between runs, Semrush and SE Ranking fell sharply from their Run 1 positions, and names like Ayzeo, Similarweb, and AthenaHQ appeared prominently in Run 2 with little or no presence in Run 1. Perplexity's named list changed the most of the three engines.
Full counts, per-engine detail, and the complete leaderboard now live on The Get-Picked Index. Source: NB-0001_run2_variance_summary.md, corpus check NB-0001, plus the per-engine Run 2 transcript files dated 2026-07-29.
We ran our own paid $197 Get-Booked Audit and Score rubric against Namebeam itself. First pass scored 31 of 100. A same-day correction, verified against the live served bytes of namebeam.ai rather than internal logs, revised the number to about 39 of 100. The correction is printed here rather than the lower first pass, because the live artifact is what a customer and an AI engine actually see.
| Factor | Score | Max |
|---|---|---|
| Findable | 3 | 30 |
| Reachable | 8 | 15 |
| Quotable | 16 | 20 |
| Bookable | 10 | 15 |
| Trusted | 2 | 20 |
| Total | 39 | 100 |
Source: AUDIT_NB_SELF_001_2026-07-29.md, correction section, filed the same day against live-fetched bytes.
The score moved because we shipped four fixes the same week and re-verified against the live site, not against our own notes:
Before this week's deploy, the same rubric would have scored roughly 22 of 100. The honest read of the delta is about plus 17, driven mostly by Quotable and Bookable, the two factors the four fixes above touch directly. Findable and Trusted did not move this week, and they are printed here unchanged rather than rounded up.
Since the last entry this series grew from two runs to twenty-one dated checks. One check, Copilot on 2026-08-01, named Namebeam. One check on 2026-08-24 logged void and counts as neither a hit nor a miss. Thirteen fresh checks landed on 2026-08-24 and 2026-08-25 and every one printed not named. That is the number, printed at full size, and the weekly cadence keeps running.
The corpus behind this ledger reached 379 dated rows on 2026-08-29 and was consolidated into a single lineage the same day. The current collection round, computed over 1029 citations from 96 rows, shows what engines cite instead of naming businesses directly: business websites 71.8 percent, directory and profile listings 13.4 percent, roundup listicles 7.7 percent, community and social 5.8 percent, news 1.3 percent. Full method and the dot chart of every customer-zero check live at proof.namebeam.ai, updated 2026-08-27.
The one naming traces to a personalized Copilot session that greeted the founder by name, so it stays flagged and is not banked as a clean signal until it repeats in a cold session. And the trend graphs now lead the proof page, a change made after a reader said the ledger was hard to follow. He was right, and the correction is part of the record.
Source: citation_corpus.csv, the 21 customer-zero rows through 2026-08-25; CITATION_SERIES round of 2026-08-29; proof.namebeam.ai.
Trunkline, our crypto-education property, ran the identical four-engine check on 2026-07-29: Claude, Perplexity, Gemini, and ChatGPT, five questions each, one run per question per engine. Result: Trunkline was named zero times out of twenty answers, a true zero with no false positives recorded.
Two of the five questions, one on crypto education that shows real trading results including losses, and one on a site with verified, dated receipts for crypto and finance claims, returned no single, clear incumbent across all four engines tested. Each engine reached for a scattered list of partial fits, trading-journal tools, regulator trackers, fact-checking sites, block explorers, rather than naming one obvious answer. That pattern held across all four engines, not just one.
Framed honestly: this is one dated run across four engines, not a permanent map of the market. The gap is testable and, so far, confirmed on this single run. It will be re-checked, and any change will be printed here.
Source: TK-0001 four-engine summary, corpus files TK-0001 claude, perplexity, gemini, and chatgpt legs, dated 2026-07-29.
The number climbs, or we print that it did not. Either way, it goes in this ledger, dated, sourced, and at the same size.