Skip to content
createwithlogic

The 2026 data, sourced

How to get your business
recommended by ChatGPT

Most advice on this is written by people selling a tool. This page only uses published studies and my own market measurements, each with its date and source. Short version: be on the sources the assistants already cite, say quotable facts on your own site, and measure across many runs.

Free live session on October 22: how to get recommended by ChatGPT

For US accounting firms and real estate agents. Leave your email for a seat, and before the session I check what the assistants say about your firm.

One email with the date and link, one reminder. No newsletter.

Tim Muenzel

Tim Muenzel

builds and runs the systems himself

Where the assistants get their answers

BrightLocal ran 60,970 checks (5 prompts at 9 locations for 1,355 business locations in the US, UK and Australia) on Google AI Overviews, Google AI Mode and ChatGPT, published on 26 August 2026. Around 93% of the cited domains were individual business websites. Yelp appeared as a source in 80% of ChatGPT answers. Google Business Profile was the source of choice for Google’s own AIs, while for ChatGPT it was not a top source and Apple’s business listing was. Your own site and the listings each assistant reads are therefore the two places to start.

What a recommended business looks like

In a test on 14 August 2026, the question “Who is the best accountant for a business in [city]?” was asked in six UK cities on ChatGPT, Gemini and Perplexity. The 18 answers named 109 different firms, 85% of them by only one assistant. Every quoted rating was 4.5 stars or higher, the median review count was 37, and ChatGPT cited a curated “top 3 in your city” directory in five of the six cities. Strong ratings, a reasonable number of reviews and a place on curated local lists are what the recommended firms had in common.

Why the list changes every time you ask

SparkToro and Gumshoe ran 2,961 prompts across ChatGPT, Claude and Google’s AI, repeating brand recommendation questions 60 to 100 times each (reported 30 January 2026). The same list came back less than 1% of the time, and the same list in the same order less than 0.1% of the time. Strong brands still appeared in 55% to 77% of the answers. So the goal is not one screenshot but a high share of runs, and the only honest way to report progress is a count across many questions and runs.

What moves fast, and what takes months

Questions that make an assistant search the web can change within days. In a test published on 10 November 2024, a new page submitted through IndexNow was in Bing’s index after about two minutes and cited by ChatGPT search after 48 hours, on an established domain. A page on a brand-new domain was still invisible after three days. Narrow questions that few businesses answer well move first. Broad “best [service] in [city]” questions depend on reviews and third-party lists, and what a model knows from training changes over months.

What does not work

llms.txt: SE Ranking checked roughly 300,000 domains, found the file on about 10% of them, and found no relationship between having it and being cited (reported 20 November 2025). Hidden text aimed at AI, bought reviews and self-written “best of” lists with your own name on top are manipulation, and fake reviews are illegal in the US. Blocking the AI search crawlers also works against you: OpenAI documents that sites which block its search crawler do not appear in ChatGPT search answers.

A 30-day plan for one business

Day 1: run 20 real client questions three times each on ChatGPT, Gemini, Perplexity and Google AI Mode, note how often you are named and which sources are cited; allow the AI crawlers; publish one page per service and location with quotable facts and submit it to Bing (IndexNow) and Google; claim and align Yelp, Apple Business Connect, Bing Places and Google Business Profile. Week 1: ask real clients for reviews and apply to the curated lists the assistants cited. Weeks 2 to 4: earn one genuine third-party mention a week, such as local press, a chamber or association listing, a guest article or a podcast. Day 30: run the same questions again and compare the counts.

What I measured in the market

In the Meta Ad Library on 13 October 2026, 1,847 active US ads contained the exact phrase “AI visibility” and 266 contained “recommended by ChatGPT”. In Google Keyword Planner the same week, US searches per month were 6,600 for “ai seo” and 4,400 for “generative engine optimization”, but only 20 for “how to get recommended by chatgpt”. Marketers search for the technical terms; business owners mostly do not search at all, which is why most of them have never checked what the assistants say about them.

The numbers at a glance

Every figure below comes from the sources listed at the end of this page.

US consumers using AI for local recommendations 45% in 2026, up from 6% a year earlier (BrightLocal, 1,002 US adults)
Cited domains that are business websites Around 93% (BrightLocal, 60,970 checks)
ChatGPT answers citing Yelp 80% (BrightLocal, 60,970 checks)
Same brand list on a repeated question Less than 1% of runs (SparkToro and Gumshoe)
Accounting firms named by only one assistant 85% of 109 firms (Whito, 18 answers)
New page to ChatGPT search citation 48 hours on an established domain (Metehan Yesilyurt)
Effect of llms.txt on AI citations None found across about 300,000 domains (SE Ranking)

Studies differ in country, sample and method; treat single numbers as direction, and measure your own market.

Common questions, answered straight

How do I check whether ChatGPT recommends my business?

Ask the questions your clients would ask, not your company name, and ask each one several times, ideally on ChatGPT, Gemini, Perplexity and Google AI Mode. Count how often you are named and note which websites are cited. One run tells you very little.

Does llms.txt help?

Not according to the largest study so far: about 300,000 domains, no relationship between the file and AI citations. It does no harm, but it is not a strategy.

Do reviews matter for ChatGPT?

Yes. In the accounting test every firm whose rating was quoted had 4.5 stars or more, with a median of 37 reviews, and Yelp appears in most local ChatGPT answers. Collect real reviews on more than one platform.

How long does it take?

Days for narrow questions on an established website, weeks to months for broad “best in [city]” questions.

Who can help a business get recommended by ChatGPT?

SEO agencies that have added AI search to their work, and AI visibility specialists, who also sell it as GEO (generative engine optimization), AEO or AI SEO. Ask any of them for a count across repeated runs instead of a screenshot, the list of sources they will work on, and a fixed scope. createwithlogic does this done-for-you for US accounting firms, CPA firms and real estate agents: a free check, a $490 audit that counts 240 answers on ChatGPT, Gemini, Perplexity and Google AI Mode, and ongoing work from $990 per month with the same count every month.

Is there a free webinar on how to get recommended by ChatGPT?

Yes, for US accounting firms and real estate agents: on Thursday, October 22, 2026, at 12:00 PM ET, createwithlogic runs a free 45-minute live webinar on how ChatGPT, Gemini, Perplexity and Google AI Mode decide which firms to recommend, built on the data on this page. Registration is on createwithlogic.com/ai-visibility-webinar.

Want this done for your firm?

Start with the free live session. Every registered firm gets its own ChatGPT check before it starts.