Skip to content
Marketing, IT & AI automation · Surrey and Metro Vancouver
Digital Allo Marketing Services Ltd. Contact Us

Digital Allo Blog · November 19, 2026

AI Assistants Never Named 74% of the Businesses They Were Asked About — Would They Name Yours?

By Digital Allo Editorial Team Published

The test: 2,688 answers, and most businesses weren’t in any of them

A study published today, 18 November 2026, by Citefold asked a simple question: when a buyer asks an AI assistant for a recommendation, does the assistant actually name your business?

The method was straightforward. Citefold took 224 Y Combinator companies from the 2024–2026 batches, read each homepage, and wrote four questions a buyer in that company’s category would ask — questions like “what is the best X for Y”, written without the company name in them. Each question went to three engines: ChatGPT (GPT-5.6) and Gemini (Gemini 3.1) through their APIs with no web search, and Perplexity (Sonar), which always searches the web. Twelve answers per company. 2,688 answers in total.

The headline result: 74% of the companies — 166 of 224 — were not named in a single one of their 12 answers. Another 36 were named once. Only 7 were named three times or more, and the best result in the whole study was 5 of 12. The median AI visibility score Citefold calculated was 32 out of 100, and no company scored above 79.

That is not a ranking problem. It is an existence problem. For three out of four of these companies, the AI assistant’s answer to a buyer in their own category did not contain them at all.

Why: an assistant that answers from memory can’t name what it never learned

The gap between the engines explains most of it — and it is good news for a local business, if you read it right.

Answering from what the model already knows, with no web search, ChatGPT named the company in 0.9% of answers and Gemini in 0.1%. These are 2024–2026 companies, mostly newer than the models’ training data, so near-zero is partly a training-cutoff effect — Citefold says so itself. A Surrey business that has been trading for years is not in the same position as a 2025 startup. But the direction still matters, because the same split shows up when search is switched on.

Perplexity searches the web for every answer, and it named the companies in 9.2% of answers — roughly ten times ChatGPT’s memory-only rate. Citefold reran 39 of the companies on ChatGPT with web search switched on: ChatGPT named them in 9.0% of answers, and 23% of those companies at least once, against 0.6% without search.

So the practical rule is simple: an assistant that searches the live web can find you; an assistant answering from memory can only name you if it already knows you. And most buyers’ questions are now going to assistants that search — ChatGPT with search, Perplexity, Google AI Mode. Your job is to be findable by the search, not to hope the model remembers you.

What the named companies have in common

Citefold also ran 12 automated checks on every company’s homepage, and the differences between the named and the unnamed are worth knowing — with the study’s own warning attached: these are differences in a small sample, not proof of cause.

  • Comparison or alternatives pages — the biggest gap. 45% of the 51 companies with a “versus” or alternatives page were named at least once, against 20% of the 173 without one. And 77% of all companies had no such page at all. A page that says “us vs the alternatives” is exactly the kind of page an AI quotes when a buyer asks it to compare.
  • Question-and-answer content. 33% of companies with Q&A content on their site were named, against 21% without. But 44% had no Q&A content at all.
  • An llms.txt file. 48% of the companies already publish one — far more than Citefold expected — and 30% of those were named, against 22% without. The gap is small, and it is not the lever the comparison page is. Note also that Google has said it doesn’t read llms.txt files, so treat this one as a small, cheap extra, not a strategy.
  • Structured data on the homepage. 28% named with it, 23% without. Only 34% of homepages passed the structured-data check at all.
  • The basics still fail often. Only 21% of homepages passed the “clear company identity” check — one page in five doesn’t clearly say what the company is. 8% block AI crawlers in robots.txt outright.

None of this is magic. It is the same pattern every earlier study in this series has shown: AI answers are built from pages that answer buyer questions in plain language, and from other people’s pages that vouch for you.

The answers are built from other people’s pages

When Perplexity searched, it cited github.com in the reports of 92 companies, dev.to in 46 and g2.com in 36. Those are the tech world’s equivalents of the directories and review sites that matter to a Surrey business. An assistant that searches reads the pages that already talk about a category — and a business absent from them is usually absent from the answer.

That is the same finding as the studies we covered in our earlier post on how AI decides which businesses to recommend: third-party signals — reviews, directory listings, media coverage, community discussion — carry the most weight, followed by clear, specific, well-sourced pages on your own site. The new study adds the other half of that picture: before reviews can decide whether AI recommends you, you have to be named at all. And being named starts with being findable in the places AI actually reads.

The IT-services warning: this gap is worst for firms like ours

There is a second study that makes this personal for Digital Allo — and for every IT and digital-services firm in Surrey. CodaAI’s AI Blind Test for B2B tech, with snapshots from 5 November 2026, put 100 software and IT companies through eight buyer questions each across ChatGPT, Claude, Gemini, Perplexity and Google AI Mode, all with web search on.

Software vendors were named in 32.6% of answers. System integrators and IT service providers were named in only 5.9% — and 16 of the 40 never appeared in a single answer. In 46.4% of answers the AI named a competitor instead. And the company’s own website was a source in only one answer in ten.

If you run an IT, managed-services or digital agency in Surrey, that is your category. Your buyers are already asking AI assistants to shortlist vendors, and the assistants are mostly naming someone else — usually a mid-sized competitor — because that competitor has the comparison pages, the directory presence and the third-party mentions that you don’t.

A third study, from Okara, explains why the fix has to be done engine by engine. Across 230 answers to B2B software buying questions, the four major assistants agreed on which brands to recommend about 41% of the time — but the overlap in the sources they cited was just 6% between any two engines. ChatGPT cited the vendor’s own product pages for more than half its sources; Claude, Gemini and Perplexity mostly cited third-party roundups. There is no single place to be visible. You have to be findable on each engine’s terms.

The 10-minute self-test

You don’t need a study to find out where you stand. Run this now, in a fresh chat on each assistant, with no company name in the question:

  1. Write down the question your best customer would ask — the way they would actually type it. “Best managed IT provider for a small business in Surrey” or “affordable website design company Surrey BC” or whatever your buyer asks.
  2. Ask it in ChatGPT, in Perplexity, and in Gemini (or Google AI Mode). Fresh chat each time, no follow-ups, no hints.
  3. Write down whether your business is named — and who is named instead.
  4. Repeat it once more on a different day. Answers change between runs: in a Mumbai local-business study published 15 November, about a third of the business names Perplexity returned came back on the second run. One check on one day tells you little; two checks tell you a pattern.

If you are not named, you now know the gap is real for you — and you know exactly which competitors the assistants are naming instead. That list is your to-do list.

The 5-step fix

  1. Step 1 — Run the self-test above, and write down the competitors named instead of you. Those are the businesses whose pages, listings and reviews the AI is reading. Look at what they have that you don’t: a comparison page, a directory listing, a review count. That is your benchmark, not a guess.
  2. Step 2 — Add one comparison page and one Q&A page. The Citefold data says the comparison page is the single biggest gap: 77% of companies don’t have one, and the ones that do get named more than twice as often. Write “us vs the alternatives” honestly — what you’re best at, what you’re not, who should choose the other option. Then add a page that answers the five questions your buyers actually ask before they call you, in plain language, one question at a time. AI assistants quote pages like these.
  3. Step 3 — Get listed where AI looks. Your Google Business Profile is the first listing to fix — ours has 6 reviews and our competitors have 70-plus, which is exactly the gap the earlier studies say decides recommendations. Then the directories and review sites in your trade: the ones a buyer would check, and the ones Perplexity-style search leans on. A business with no third-party footprint is invisible to an assistant that searches.
  4. Step 4 — Make your own site readable by the crawlers. Check three things: your robots.txt lets AI crawlers in (8% of the Citefold companies blocked them outright), your homepage has structured data and one clear H1 that says what you do and where, and your XML sitemap exists and is current. Your homepage should pass the “clear company identity” check that four out of five companies in the study failed.
  5. Step 5 — Check your own Search Console data, then re-test monthly. Under Performance, then Search results, then Generative AI, you can see how often your pages appear in AI Overviews and AI Mode. And re-run the self-test once a month — answers change, models change, and the November core update is still rolling out, so don’t judge anything until at least a week after it finishes (expected around 24 November). Keep a note of the date you tested and what changed.

The bottom line

Three out of four businesses in today’s study were never named in a single AI answer to a buyer in their own category. Not ranked low — never mentioned. The assistants that search the live web named them ten times more often than the ones answering from memory, and the businesses that did get named were the ones with comparison pages, Q&A content, and a footprint on other people’s pages.

For a Surrey business, that is actually good news. The fix is not a new platform or a special file. It is the unglamorous work that was already on the list: a page that compares you honestly with the alternatives, answers to the questions your buyers ask, a complete business profile with real reviews, and a site the crawlers can read. Do that, run the self-test, and you will know — in ten minutes — whether the assistants are naming you yet.

Want Digital Allo to run the test on your business and fix what it finds? Our AI search visibility work covers exactly this — the comparison and Q&A pages, the profile and review fixes, and the crawlable-site checks — alongside our SEO services and reputation management. Start with a free AI Opportunity Assessment, or contact us — (604) 401-0699.

FAQ

Would an AI assistant name my business if a buyer asked? Maybe not — and a study published 18 November 2026 found that 74% of the 224 businesses tested were never named in a single one of 2,688 AI answers to buyer questions. Run the test yourself: ask ChatGPT, Perplexity and Gemini the question your buyer would ask, with no company name in it, and see whether your business appears.

Why do AI assistants name some businesses and not others? Assistants that search the live web name far more businesses than ones answering from memory. In the Citefold study, Perplexity — which always searches — named companies in 9.2% of answers, against 0.9% for ChatGPT and 0.1% for Gemini answering from what they already knew. With web search switched on, ChatGPT named them in 9.0% of answers. Businesses with comparison pages, Q&A content and a footprint on third-party sites were named more often.

Does an llms.txt file make AI assistants name my business? Only a little, if at all. In the Citefold study, 30% of companies with an llms.txt were named at least once, against 22% without — a small gap, and the study warns the differences are correlations in a small sample, not proof of cause. Google has also said it doesn’t read llms.txt files. A comparison page showed a much bigger gap: 45% named with one, 20% without.

My business is an IT or digital services firm — is the gap worse for me? Yes, according to CodaAI’s blind test of 100 software and IT companies. System integrators and IT service providers were named in only 5.9% of AI answers, against 32.6% for software vendors, and 16 of the 40 IT service firms never appeared at all. In 46.4% of answers the AI named a competitor instead.

How do I check whether AI assistants name my business? Ask the question your buyer would ask — without your company name — in a fresh chat on ChatGPT, Perplexity and Gemini or Google AI Mode, and write down whether you are named and who is named instead. Repeat it on a different day, because answers change between runs. Then fix the gaps: a comparison page, Q&A content, a complete Google Business Profile with real reviews, and a site AI crawlers can read.

Service areas

Serving businesses across British Columbia

On site and remote across Metro Vancouver and the Fraser Valley. Remote-first on Vancouver Island and in the Okanagan, with visits by arrangement.

Surrey & South of the Fraser

On-site and remote

Vancouver, Richmond & North Shore

On-site and remote

Burnaby, Tri-Cities & Maple Ridge

On-site and remote

Fraser Valley

On-site and remote, with planned visits further east

Vancouver Island

Remote-first, visits by arrangement

Okanagan & Thompson

Remote-first, visits by arrangement

See all service areas