Usually not you. Every other tool promises to make you visible, then quotes a lift it cannot control. We do the opposite. We measure who AI actually names across the six AI tools your buyers use, then hand your team the evidence and the punch list. Not a promise. The receipts.
A Google search gives you ten links. An AI answer gives you three names, and nothing tells you when yours is not one of them. So here is a real market, measured. Six AI tools plus a Bing organic control, one Waikiki boutique hotel, and four questions its buyers actually ask.
An interactive map of a real measurement. Six AI tools and a Bing organic control on the left, five Hawaii hotels on the right, anonymized. For the question "best hotel in Honolulu", every tool that answered named other hotels and none named the boutique hotel. For "best hotel in Waikiki", one tool of six named it. For "top boutique hotels in Hawaii" and "boutique hotel in Honolulu walking distance to the beach", four tools named it. Measured June 21, 2026.
Real measurement, not a mockup. Hawaii hotels, measured June 21, 2026: 18 locked questions, 6 AI tools plus a Bing organic control, 9 sampled answers per surface per question across three runs. Four questions and five of the measured hotels are shown. A line is drawn when a tool names a hotel in at least half of its samples. Google AI Overviews is scored on the answers it shows. Hotels are anonymized here, the way every non-customer is on our public pages. A customer’s map ships fully named.
This hotel does not know any of this. Neither do its competitors. AI builds these answers from everyone else’s account of a business, not from its own site. See where AI sources its answers → or read the full Hawaii hotels teardown →
There are two jobs in AI search, and one vendor cannot honestly hold both. Optimization tools change your pages, then report how their own changes performed. We never touch your site, so the number we hand you answers to nothing but the measurement.
You can hire either. Only one can referee the other.
The question set is committed to a public timestamp before we measure. The method is documented in full at /methodology/, and every published number lives on a public claims record, dated and tied to the run that produced it. Your team executes, and we never grade our own work. The structural comparison, tool by tool →
Most of the people who call us already have someone shipping the work: an in-house team, an agency, a platform running on autopilot. Good. We are not trying to replace them. We exist for the question that arrives ninety days later.
A vendor who ships the change and then reports the lift is not lying to you. It is simply the only party in the room with a reason for the number to move. That is not an accusation, it is arithmetic, and it is why auditors do not keep the books, and why a hotel trusts a comp-set index published by a company that does not operate hotels.
We are structurally incapable of returning you a flattering number, and you can check every piece of that yourself:
The last one is the only one that counts. The rest is what any vendor would say. The retraction is the receipt.
If the work is already underway and the number has not moved, that is worth knowing early. If it has moved, you should be able to prove it to a board with something other than the invoice of the company that did the work. Read the retraction first, then decide whether to believe our numbers →
We measure whether the AI engines name a business when buyers ask the category question. Eleven measurements, three runs each, six AI tools. On the engines that search the live web, where most buyers already are, AI names three to five businesses per question. Someone in your category is getting named. The first job is finding out whether it is you.
Then there is the result the AEO industry would rather you did not see. The engines that answer from training data instead of searching the live web barely cite local businesses at all. Claude, on the same questions where web-searching engines reach 38 to 66 percent, collapses to near zero. You can look strong on one surface and be invisible on another, which is exactly why a single-engine tool cannot tell you where you actually stand.
| Category | Top web engine | Claude |
|---|---|---|
| Honolulu HVAC | ChatGPT search, 66% | 2% |
| Honolulu med spas | Gemini, 64% | 2% |
| Hawaii CPA firms | OpenAI, 60% | 1% |
| Austin CPA firms (cross-geo control) | holds outside Hawaii | 0% |
The HVAC line is the one we pre-registered. Before a single question ran, we committed the prediction to a public file with a timestamp. We measured two percent. The same story repeats in med spas and accounting, across three unrelated industries in two states. A business can look fine on one surface and be invisible on another. Most vendors cannot publish a table like this, because to publish it you have to actually measure, name the date, and lock the questions before you look. A vibe cannot be put in a table.
Every number here is on the record, with the dated run and the documented method that produced it →
Hawaii Theatre Center, the 1922 landmark in downtown Honolulu, agreed to be named. We measured what the AI engines said about them against a 19-question buyer set. The interesting part was never a score. It was where the answers were coming from.
A Charity Navigator profile that had not been updated since 2023. A Better Business Bureau listing last touched in 1999. No presence at all on Bing Business Profile, which feeds Bing’s organic results. Authority backlinks pointing to the wrong places. We handed their team the memo and collaborated on the meta description rewrites, and their team shipped the work.
None of that is exotic. All of it is invisible to a standard SEO scan, because a standard SEO scan looks at your website, and none of those things are your website. That is the whole point. An AI answer is assembled from everyone else's account of you, and for a beloved 1922 theatre, part of that account was a quarter century out of date.
What we used to say here, and why we do not anymore. This page once presented a score lift and a Perplexity citation count from this engagement as evidence that our work caused them. Both were real measurements. The causal link was not, so we tested it against our own domain, failed our own pre-registered criteria, pulled the product inside a day, and retracted both claims. The diagnostic findings above are what survived, because they are what actually happened. The retracted figures themselves are quoted in exactly one place on this site now, on the accounting itself, so they can never be lifted from here and reused as proof.
We should also say the part that is easy to leave out. Our snippet was deployed on their site during that engagement, and it did move the score we were measuring. What it never moved was the citations. We were the vendor grading our own homework and we could not see it until we ran the test. That is the reason we do not touch your website now. The full accounting, including the test that killed our own thesis →
Your next guest asks AI before they ever see your rate.
Measured proof from one of the most contested hotel markets in the world, built for a property anywhere.
For hospitality →It is not your website. It is your category.
The most aggregator-captured market we have measured, and the cross-category data that proves the number is structural.
For real estate →The forecast we dated before the data. It held.
One of the few local categories where the citation is still up for grabs, measured to a pre-registered standard.
For home services →We measure. You execute. The client stays yours.
An independent measurement layer upstream of your team, resold into your accounts with your markup on top.
For agencies →Eleven measurements so far. Yours can be next.
Venues, tours and experiences, weddings, clubs, local services. If your buyers research before they buy, it can be measured.
Start the conversation →The engines re-read the web constantly. What they cite for your category changes week to week. A competitor ships a page, an engine re-crawls, a new source surfaces, and the answer moves without anyone telling you. A one-time audit is a photograph of a river.
Your standing changes every month. The month you stop measuring is the month you stop knowing.
So we measure the same questions three times a month, every month, and not once a day. Day-to-day movement is mostly noise, and reacting to it would have you chasing a number that means nothing. We are measuring the climate, not the weather.
Your position is contested share, and share is taken. The first read tells you where you stand. The second tells you whether you are moving. By the sixth you have a trend most of your category is not even watching.
NeverRanked measures what the AI answer engines cite for your category, split across two layers that fail in different ways.
The deliverable is a research memo and a prepped punch list, ordered by impact. Your team executes it. We do not. That separation is structural.
Atlas is the data-interpretation layer of your dashboard. It answers what the measurement shows. It refuses to tell you what to do. That separation is the engagement.
The boundary is structural. Prioritization lives in your monthly memo, written by the principal. Atlas holds the data. Crossing that line would damage the engagement.
This one is built to be. Four reasons the numbers can be trusted.
Every methodology claim is anchored in a hash-locked pre-registration before the test runs. The claim cannot move after the data lands.
The full measurement method is documented at /methodology/, against hash-locked question sets and dated runs on the public claims record. One of the six AI tools, Gemma, is open-weight, so the model itself is independently inspectable.
The same hash-locked questions, run after run across all six AI tools. An AI answer can change between two askings of the same question, so a single snapshot is not measurement.
We never touch your website, your code, or your CRM. That is not only a security posture, it is what makes the number trustworthy. The moment the measurer is also the one being measured, the score stops being a measurement and becomes a sales document. We keep our hands off the property on purpose, so the only thing we can do is report what the engines actually cite.
Five stages. Plain words. No SaaS dashboard between you and the work.
Lock the category, the cohort, and the 18 buyer questions we will measure. One call, no homework. It is the only call the engagement requires.
See an example question set →Repeated runs across 6 AI tools, and the starting numbers frozen with the cohort and the question set. Nothing later can be measured until something is fixed in place first.
PDF or markdown to your team. Named competitors, observed gaps, the clear list of what to fix first.
You ship the work. We measure whether it lands. That separation is the whole position.
What moved, what did not. Updated punch list. Drift alerts when a competitor moves in your category.
The first research memo arrives at the end of the baseline month.
Nine categories deep in one proving ground, one of the most contested visitor markets in the world, plus a cross-geo control in Austin that shows the patterns travel. The method is geography-blind: lock the questions, run the engines, count the names.
Every teardown is built from a hash-locked question set, 3 measurement runs, and the same 6 AI tools. Anonymized at the firm level for non-customers, named in full inside paid engagements. Honolulu HVAC was the first finding we pre-registered: the prediction was committed to a public timestamp before the measurement ran. The Claude training-data collapse holds across three unrelated local-service industries, and the two newest teardowns, real estate and hotels, are where it inverts: AI hands the answer to the aggregators (Zillow, Booking), while the training-data engines cite the businesses more than the live-search engines do. In hotels, the boutiques AI names even out-cite the chains.
Before running a single query, we committed a prediction to a public timestamp: Claude would cite Honolulu AC companies under 5% of the time. It came in at 2%, while the web-searching engines reached 38% to 66% on the same questions. A forecast made before the data, not a case study written after it. This is the third unrelated industry where Claude collapses on local firms, after CPA and med spas. Read the pre-registered teardown →
13-hotel cohort. Only 17% of citations go to a hotel's own site. 74% go to Booking, Expedia, and travel editorial. But among hotels AI names, boutiques out-cite chains 493 to 457, and the most-cited hotel in Hawaii is a boutique. The contest is hotel versus OTA.
11-firm cohort. Local agents get just 15% of citations while 77% go to portals like Zillow and Realtor.com. The training-data engines cite local firms more than the live-search engines do, the first category where Claude is not the blind spot.
23-domain cohort. The widest cross-engine gap of any category measured. One bank owns the head queries on training-data tools. The long tail sits open in the Bing organic control.
42-firm cohort. AI defers to lead-gen aggregators (SmartAsset, Unbiased, Plannersearch) more than to any individual firm. The structural ground for firms sits outside that middleman tier.
46-practice cohort. Bing organic returns zero practice websites for the entire cohort. Whether ranking there changes what any AI answer engine cites is not something this measurement tested.
33-firm cohort. The dominant firm gets roughly three times the citations of the second-tier firms. For any firm outside the top 5, the closable ground is the long tail and Bing search (control).
41-firm cohort. Claude and Gemma cite Hawaii CPA firms less than 2% of the time. Competitive game plays inside OpenAI, Gemini, Perplexity, and Google AI Overviews.
37-firm Austin cohort. First non-Hawaii measurement. Claude's collapse holds across geographies, Gemma's does not. A category pattern from one geo turned out to be two once measured in two.
15-firm cohort. The Claude training-data collapse, found again in a second unrelated industry. Strong on the web-searching engines, near-zero on Claude. The competitive game is on the live-web tools.
12-firm AC-company cohort. The prediction was committed to a public timestamp before any data existed. Claude landed at 2%, the web-searching engines at 38-66%. The first finding we called in advance, not a case study written after.
The cross-category teardown reads every measurement against each other →
Which questions in your category get answered, who gets cited when you do not, and which kinds of sources the engines actually pull from when they decide.
Take one named reference. At Hawaii Theatre Center, the readout surfaced what a standard scan walks past: a Charity Navigator profile not updated since 2023, a Better Business Bureau profile last touched in 1999, a missing Bing Business Profile, authority backlinks pointed at the wrong places. The quiet, citation-shaping detail nobody is looking at.
No bundled tiers, no per-seat math.
Two ways in. Monitor tells you where you stand. Audit proves whether anything moved.
Monitor is the meter. Every month, which of the six AI tools name you and which do not, question by question, plus the answers built from your own site that never credit you. No setup, no call, cancel anytime. You will not find a precise-looking percentage in it, because one run a month cannot support one honestly.
Audit is the record. The baseline month freezes the competitor set, the question set, and the starting numbers, so later months have something real to be measured against. Then three full runs a month, thresholds agreed before the work starts, and the written research memo: where you stand, what moved and what did not, a punch list with the exact first click for each item, and what we are watching next.
Start with the free instant check. Paste your URL and see in seconds what the six AI tools can read from your site. The full engagement is the other half of the picture. It runs 18 real customer questions per category across all six AI tools, measures throughout the month, builds a cohort baseline that makes the numbers mean something, and hands you a clear list of what to fix.
AI is becoming the front door, fast.
Sources: Google, OpenAI, Pew Research.
NeverRanked is a research practice, not a software company. The measurement, the memos, and the punch lists are produced by Lance Roylo, in Honolulu. There is no account layer between you and the person doing the research.
How the practice operates: we measure, we do not execute. We report what the AI engines actually cite, never what we claim our work caused. We do not promise a citation lift in advance. A finding that cannot be substantiated does not ship. That discipline is the product.
Lifted from the inbound emails Lance answers most often.
SEO measures search engine ranking factors on your own site. We measure what AI tools actually cite when buyers ask category-shaped questions, across 7 measured surfaces. The deliverable is also different: SEO tools give you a dashboard to interpret yourself. We hand off an interpreted research memo plus a prepped punch list your team executes. See /vs/ for the structural comparison.
Two reads on this. First, AI search usage in B2B and high-consideration consumer decisions is already non-trivial and growing on the curve we have public visibility into. Second, even if your buyer is not asking ChatGPT today, your competitor showing up there first when they do is the move you can not undo. We measure that surface so you know whether the move has already started.
We do not promise a lift in advance. The only promise we make is the measurement itself: you will know what AI cites for your category, what gaps exist, and what conditions a buyer of your category typically closes to move the needle. Whether your team executes the punch list well is what determines lift, and we measure that monthly so the answer is observable, not asserted.
By day thirty you have the baseline and the punch list: where AI names you, where it names competitors instead, and the fixes ordered by impact. From there the work is your team’s, and we measure whether it moves every month, with drift alerts when a competitor shifts. We do not hand you a vanity timeline. We hand you observable movement. You will always know whether it is working, in dated numbers on the public record.
We are not asking you to drop local marketing. Reviews, Maps, and reputation still matter. AEO reaches a different moment: the buyer who asks AI "who is the best [category]" before they ever open a map, and takes the three names it gives them. For a considered purchase, that shortlist is the whole game, and today you cannot even see the answer that left you out. We measure the one surface no ad budget shows you. And if your buyers do not research before they buy, we are the wrong spend, and we will tell you so.
We measure. We do not execute. We do not write content, edit pages, deploy schema, update profiles, or change your site. Your in-house team or your agency executes against the punch list we deliver. That separation is structural and is the whole position. It also means we never compete with your agency for execution hours.
Yes, and it is free. The instant self-serve check at check.neverranked.com tells you in seconds what the six AI tools can read from your site, no signup. When you want to go further, book 15 minutes and we scope one category together.
More: the full FAQ covers cancellation, NDAs, agency channel, data handling, and what happens if a finding turns out to be wrong.
See where AI puts your business across the six AI tools your buyers use.
See what AI sees →