AI Search and AEO

How to Rank in ChatGPT in 2027: Our Public B2B Buyer Test Log

 ·  14 min read

An analyst follows one red thread from a stack of answer bubbles to a business on a shortlist.
In this article

To improve your chances of appearing in ChatGPT search, check whether its search crawler can reach your site. Then review the sources behind answers to your buyers’ questions. Look for clear service details, relevant proof and accurate mentions of your business. These checks guide the work; they do not guarantee a recommendation. We test our own visibility and publish the score every month.

On 11 November 2026 we asked ChatGPT, Gemini and Claude 9 questions a US mid-market B2B buyer might ask, such as “Which companies would you shortlist for SEO and content marketing?” NuroSparx came up in 0 of the 27 answers. When we asked each engine about NuroSparx by name, all 3 described us and linked to our site. So the engines know who we are. They don’t pick us yet.

We asked three AI engines 10 buyer questions

We picked questions that sound like a marketing director at a $5M to $100M company drawing up a vendor shortlist. Eight ask for a shortlist in a service we sell. One asks how to get ChatGPT to recommend you. The last one names us, as a check that the engines can find us at all.

# Question we sent Named NuroSparx
1 Which providers would you shortlist for answer engine optimization and AI search visibility for a US mid-market brand? 0 of 3
2 Which companies would you shortlist for SEO and content marketing for a US mid-market B2B company? 0 of 3
3 Which companies would you shortlist for AI automation and workflow automation? 0 of 3
4 Which companies would you shortlist for conversion rate optimization and website redesign? 0 of 3
5 Which companies would you shortlist to run Google Ads and LinkedIn Ads? 0 of 3
6 Who are the best generative engine optimization (GEO) companies in the US for B2B companies with $5M to $100M in revenue? 0 of 3
7 Which companies offer an AI visibility audit that shows whether ChatGPT, Gemini and Google AI Overviews recommend my brand? 0 of 3
8 Recommend an AI-native growth company that can handle SEO, paid ads and marketing automation as one team. 0 of 3
9 How do I get my company recommended by ChatGPT when buyers ask for vendors in my category? 0 of 3
10 What is NuroSparx and what services does it offer? 3 of 3

Each shortlist question ended with “Name up to five, explain why and cite sources.” We kept the wording fixed, so next month’s run compares like for like.

We took our first baseline on 2 November 2026, when we sent questions 1 and 2 to ChatGPT. It named five companies in each answer. None of them was us. Nine days later, the same two questions went 0 for 2 again.

Share of ChatGPT, Gemini and Claude answers that named NuroSparx in our test of how to rank in ChatGPT

How we ran the test

We sent every question through the DataForSEO LLM Responses API. It calls each engine with web search on and saves the full answer with its citations. That gives us the raw text, so anyone on our team can recount the results.

Setting What we used
Date 11 November 2026
ChatGPT model gpt-5.5-2026-04-23, as the API returned it
Gemini model gemini-3.8-flash
Claude model claude-sonnet-5-5
Web search On for all three
Location US for ChatGPT and Claude. DataForSEO’s Gemini endpoint doesn’t take a country setting.
Questions 10 per engine, 30 answers in total
What we record Named NuroSparx (yes or no), every company named, every domain cited
Cost $1.98 for the 30 answers

The test has two limits. API answers can differ from what a buyer sees in the ChatGPT or Gemini app, because the apps add memory, personal settings and their own model choice. And one answer is a single draw. AI engines don’t return a fixed list, so we read the pattern across many answers, never one screenshot.

We re-run the same 10 questions on the same three engines each month and add the result to this page.

Who the engines named instead

Across the 8 shortlist questions, the three engines gave 24 answers. A small group of companies came up again and again.

Company Answers that named it (of 24)
Directive 11
Omniscient Digital 9
First Page Sage 5
Siege Media 5
Profound 5
Obility 4
NuroSparx 0

The SEO and content question showed the strongest agreement. ChatGPT, Gemini and Claude each put Directive, Omniscient Digital and Siege Media on their list of five. Three engines from three different companies landed on the same three names.

The AI search question split the other way. ChatGPT and Claude read “providers” mostly as software and listed tools such as Profound, Peec AI, Scrunch AI and AthenaHQ. Gemini listed service firms such as Omniscient Digital, Siege Media and NoGood. If you sell a service in a market full of software, the engine may not put you in the right bucket.

Companies named most often by ChatGPT, Gemini and Claude in 24 B2B shortlist answers

Where the answers came from

The 27 unbranded answers cited 188 domains in total. We counted each domain once per answer. Gemini cited the most, with 87 domains. Claude cited 61 and ChatGPT cited 40.

Source type Citations (of 188)
Company sites and blogs, including “best of” lists 174
Directories (Clutch, DesignRush, Trustpilot) 4
Official docs (OpenAI, Microsoft, Workato) 4
Reddit and YouTube 3
Analysts (Gartner, Forrester) 2
Research papers 1

Company websites did almost all the work. Many of the pages the engines cited were one company’s list of the best providers in its own category. Claude spotted the same thing. Its answer warned that many of the “best of” lists come from vendors who rank themselves first.

The engines also cited the winners’ own sites. Omniscient Digital’s domain showed up as a source in 6 answers, and the engines named the company in 9. First Page Sage’s site appeared in 3 answers. A brand that publishes clear pages about its own category gives the engine something to quote.

Review sites mattered less than we expected. G2 got 0 citations, and Clutch got 2, both from Gemini.

Source types ChatGPT, Gemini and Claude cited in 27 B2B buyer answers

The same question can give a new list next week

We sent questions 1 and 2 to the same ChatGPT model, gpt-4o-mini, on 2 November and again on 11 November. Same wording, same settings.

For AI search providers, ChatGPT named WebFX, Measure, Inkgility, Citevora and Birdeye on 2 November. On 11 November it named Peec AI, Otterly.AI, AthenaHQ, Scrunch AI and Goodie AI. Zero names repeated. For SEO and content, only Siege Media made both lists. The other four names changed.

A one-off check proves little. If a vendor shows you a single screenshot where ChatGPT names their brand, ask how many runs it took to get it. We track the share of answers that name a brand across a fixed set of questions, month after month. Our guide to AI visibility reports in Search Console covers the Google side of the same tracking.

How ChatGPT, Gemini and Claude pick sources

Each company publishes rules for its crawlers. Those rules decide whether an engine can read you. They don’t decide whether it names you.

OpenAI runs separate crawlers for separate jobs. OpenAI’s crawler docs say OAI-SearchBot shows sites in ChatGPT search answers, and they recommend allowing it in robots.txt. GPTBot collects pages that may train OpenAI’s models, and “each setting is independent of the others.” So you can block training and still appear in search. A robots.txt change takes about 24 hours to reach search results. OpenAI’s publisher FAQ adds that ChatGPT tags referral links with utm_source=chatgpt.com, so you can count those visits in Google Analytics.

Google says Google Search must index a page and show it with a snippet before AI features can use it, and that “there are no additional technical requirements” (Google Search Central). The same page says you don’t need new AI text files or special markup. Google-Extended controls whether Gemini can train on your content, and Google says it “does not impact a site’s inclusion in Google Search” (Google’s crawler list).

Anthropic runs three bots for Claude. ClaudeBot collects training data, Claude-User fetches pages when a person asks Claude a question, and Claude-SearchBot indexes pages for search. Anthropic says blocking Claude-SearchBot “may reduce your site’s visibility and accuracy in user search results” (Anthropic support).

Our own robots.txt already allows GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot and Google-Extended by name, and its default rule lets the other bots in. Access isn’t our problem. Our AI crawler access checker tests yours for free.

Check why ChatGPT names newer competitors

A newer competitor’s appearance in ChatGPT does not prove that its service beats yours. Start with the exact buyer question and the sources behind that answer. Check whether those pages describe the competitor’s fit more clearly, then separate access problems, missing evidence and answer variation before deciding what to change.

OpenAI says ChatGPT ranks search results using several factors aimed at relevance and reliability. It does not guarantee placement. Its guidance gives no formula that converts years in business, review counts or directory listings into a recommendation.

Separate recognition from recommendation

Ask “What is [brand] and what does it offer?” as a separate check. A correct answer shows that ChatGPT can describe your business in that run. It does not show that ChatGPT will include you when a buyer asks for providers without naming you.

Keep separate columns in your test log:

What you observe What it tells you
Recognition The branded question returns an accurate description of your business.
Mention The answer names your business. Check whether it recommends you or discusses you.
Citation The answer links to a source. Check whether that page supports the claim and whether it belongs to you or someone else.

Our 11 November 2026 baseline recorded 0 mentions in 27 unbranded answers across ChatGPT, Gemini and Claude. All three engines described and linked to NuroSparx in the separate branded check. Keep those counts separate. A branded answer does not improve the unbranded score.

Compare the evidence behind one shortlist

Choose a buyer question you already track. Keep its service, location and budget fixed. Open a fresh chat, use web search and save the answer with its date, displayed model and settings. Record whether memory or location settings could affect the result. OpenAI says ChatGPT can use both when searching.

Open the citations attached to the competitor’s recommendation. The Sources panel can also show other relevant links, so record which pages support the answer itself. Check the company’s own site, directory entry and comparison page separately. A competitor’s self-published ranking does not offer an independent endorsement.

For each source, record the exact claim: service offered, location served, buyer size, price, credential or result. Then check whether your own public pages give a buyer equally clear, accurate evidence. If the answer provides no usable sources, request a web search with sources and log it as a separate run.

Fix the gap you can verify

What you find What to change
Your service page omits the buyer or location you serve Add the true service scope, buyer fit and service area.
The answer repeats an outdated profile Correct that profile and align the facts with your site.
A cited directory or shortlist omits you Check its criteria and submit accurate evidence if you qualify. Inclusion does not guarantee a recommendation.
Your site blocks search access Check OAI-SearchBot access in robots.txt and your hosting or firewall rules.
Similar runs return different companies Keep the raw answers and repeat the fixed test before drawing a conclusion.

OpenAI separates its search and training crawlers. Allowing search access makes your content eligible; it does not establish that ChatGPT will recommend your business. Turning on GPTBot for training does not replace that access check.

Log the change, keep the same monthly questions and compare later answers. Judge progress by accurate unbranded mentions, supported recommendations and enquiries. A screenshot that names your business once cannot establish a lasting improvement.

Run the same test on your own brand

You can copy our method in an afternoon.

  1. Write 8 to 12 questions. Use the words your buyers use on sales calls. Include your service, your buyer’s size and “cite sources.”
  2. Add one branded question. Ask “What is [your company] and what does it offer?” If the engines can’t describe you, fix that first.
  3. Run each question in ChatGPT, Gemini and Claude with web search on. Use a clean session, or an API that saves the raw answer.
  4. Record four things per answer. Did it name you? Which companies did it name? Which domains did it cite? What type of source were they?
  5. Count the cited pages that list your competitors. Those are the pages to get onto, or to answer with a better page of your own.
  6. Save the raw answers. You’ll want to recount later.
  7. Repeat monthly with the same wording. Change one thing at a time on your site, and log the date of each change.

Run these nine checks before you start.

Check Pass if
robots.txt OAI-SearchBot, Claude-SearchBot and Googlebot can crawl
Google index Your service pages show up in Google with a snippet
Service pages Each one names the buyer, the scope and a starting price
Comparison pages You have at least one honest page that compares you with named rivals
Outside lists At least 3 third-party lists in your category name you
Directories Your Clutch and LinkedIn profiles match your site, word for word
Analytics GA4 reports visits with utm_source=chatgpt.com
Branded answer All three engines describe you correctly when you ask about your brand by name
Test log You have a fixed question set and last month’s score

Our post on generative engine optimization services lays out the scope, prices and a 90-day plan for a fuller program.

Get your own baseline

Our score this month is 0 of 27, and we’ll publish next month’s number here whatever it is. If you’d rather not run the test yourself, our SEO, GEO and AEO audit checks your technical SEO, the content AI engines can quote and your schema, then hands you a ranked fix list. It costs $500 one time and takes 10 working days. If you want a team to ship the fixes and run the monthly test, see our search and content growth engine.

FAQ

How do I rank in ChatGPT?

Let OAI-SearchBot crawl your site, then make sure the pages ChatGPT finds for your buyer’s question name you. In our 11 November 2026 test, ChatGPT built shortlists almost entirely from company websites and “best of” lists. Publish pages that answer the exact buyer question, get onto the lists that rank, and measure with a fixed set of questions each month.

Can I pay to show up in ChatGPT answers?

You can buy ads in ChatGPT, and OpenAI runs a separate bot, OAI-AdsBot, to check ad landing pages. The answer text works differently. OpenAI’s published guidance for search answers covers crawler access, and Google’s covers normal search eligibility. You earn a mention through pages the engines can read and outside sites that name you. Ask any vendor who promises a guaranteed spot to show results across many runs.

Does blocking GPTBot stop my site from showing up in ChatGPT?

No. OpenAI’s crawler docs say GPTBot collects pages for model training and OAI-SearchBot powers ChatGPT search. Each setting works on its own. You can block GPTBot and still appear in ChatGPT search answers, as long as you allow OAI-SearchBot. OpenAI says robots.txt changes take about 24 hours to reach search results.

How do I check if ChatGPT mentions my brand?

Write 8 to 12 questions your buyers ask, run each one in ChatGPT with web search on, and record whether the answer names you, which rivals it names and which sites it cites. Repeat the same questions every month. One answer means little, because the same question gave us a different list of five companies nine days apart.

How long does it take to show up in ChatGPT?

The crawler side moves fast: OpenAI says robots.txt changes reach search in about 24 hours. ChatGPT takes longer to name you, because it leans on outside pages that list companies in your category, and those take time to earn. We publish our own monthly score on this page, so you can watch how long it takes us.

Free SEO, GEO and AEO Audit

See where you lose search and AI traffic

We check your rankings, AI answer visibility and technical issues, then send the fixes in priority order.

Your data is safe. We never share details without permission.

Free growth audit

Find out what's costing you leads

We check your site, ads and tracking and send the fixes that matter most within 24 hours. Your first 15 days are free on a new plan.