AI Visibility

Win the Answer: The 2027 AEO Playbook for B2B, Built on 2026's Biggest Studies

Genmark AI Team14 min readPublished November 22, 2026
AI SearchAEOGEOB2B MarketingAI Visibility2027 PlanningBuyer Research

TL;DR: In 2026, half of B2B software buyers started their research in an AI chatbot, two in three Google searches ended without a click, and AI Overviews started answering most searches for brand names. The studies agree on what earns a place in those answers: a clear position, pages that answer buyers' real questions, consistent proof on sites you don't own, and access for the bots that read the web for AI. They also agree on what doesn't move the needle much: schema markup and llms.txt. Below are the 12 findings we'd plan 2027 around, and the seven moves they point to.

If you sell to businesses, your next customer may already have a shortlist before they visit your website. The question for 2027 isn't whether AI search matters. It's whether the AI's answer includes you, describes you correctly, and gives buyers a reason to pick you.

We read the most-cited 2026 research on AI search and B2B buying: surveys from G2, Forrester and Pew; traffic studies from SparkToro, Ahrefs and Demandbase; and announcements from Google, Microsoft, OpenAI and Cloudflare. Every number below links to the original source. Where we go beyond the data, we say so.

Part 1: What 2026's data says

1. Half of B2B software buyers now start in an AI chatbot

In G2's March 2026 survey of 1,076 B2B software buyers and decision-makers, 51% said they start software research with an AI chatbot more often than with Google, up from 29% in April 2025. 71% rely on AI chatbots at some point in their research, and 53% say research with a chatbot is more productive than traditional search (G2, April 15, 2026).

The answer also changes the outcome. 69% of buyers chose a different vendor than they first planned based on AI chatbot guidance, and 33% bought from a vendor they hadn't been familiar with (G2 press release, April 15, 2026).

What it means: the shortlist now forms inside the answer. For challenger brands, that 33% is the opening: AI can introduce you to buyers who have never heard of you.

2. Nearly every business buyer uses AI, often in tools you can't see

Forrester's 2026 State of Business Buying found that 94% of business buyers use AI in their buying process, and twice as many buyers named generative AI or conversational search as a more meaningful source of information than any other source, ahead of vendor websites, product experts and sales. 61% use private AI tools provided by their organization (Forrester, January 22, 2026).

What it means: a lot of your buyers' research happens in assistants that never send you a referral or show up in a dashboard. You can't measure all of it. You can control what those assistants find when they look you up.

3. Two in three Google searches end without a click

In the first four months of 2026, 68.01% of US Google searches ended without a click, up from 60.45% in 2024: the fastest increase in a decade, per SparkToro's analysis of Similarweb clickstream data (SparkToro, June 8, 2026). When an AI Overview appears, Ahrefs found the top-ranking page's average click-through rate is 58% lower (Ahrefs, February 4, 2026).

The open web as a whole is shrinking too. SparkToro's study published November 10, 2026 found that visits to the web's top sites in Q3 2026 were 12.2% below their 2022 high, while about 24% more people use the web. That's a 29.2% drop in visits per user (SparkToro, November 10, 2026).

What it means: attention didn't disappear. It stopped needing to click. Traffic is becoming a lagging, partial measure of whether buyers are hearing about you.

4. AI referrals are still small, but B2B referrals are growing fast

Keep this in proportion. Across 74,752 websites, AI chatbots sent 0.28% of total web traffic in March 2026; Google sent 28.12% (Ahrefs, May 15, 2026).

For B2B, the curve is steeper. Demandbase's platform data shows ChatGPT-referred visits to B2B websites rose from about 645,000 a month in June 2025 to 2.6 million in June 2026, a 303% increase (Demandbase, August 12, 2026).

What it means: don't judge AI search by its traffic share. Judge it by whether you're on the shortlist. The visits that do arrive come from buyers who have already done their research.

5. Your brand search is now answered by Google, not by your homepage

As of September 29, 2026, AI Overviews appeared on 82.91% of US desktop searches for branded keywords, up from 61.3% in July. When those Overviews cited outside sources, Wikipedia appeared in 71.9% of them, YouTube in 38.6% and LinkedIn in 22.8% (Ahrefs, October 2, 2026).

What it means: when a buyer searches your name, the first thing they read is a summary you didn't write, built partly from pages you don't control. Your Wikipedia entry, YouTube presence and LinkedIn page are now part of your homepage. (We covered what to do when that summary is wrong in Your Buyers May Be Evaluating a Company That No Longer Exists.)

6. Ranking well is no longer enough to get cited

Only 37.9% of URLs cited in AI Overviews also ranked in Google's top 10 for the same query. 31.2% ranked between positions 11 and 100, and 31.0% didn't rank in the top 100 at all. Among the cited pages that didn't rank, 18.2% were YouTube videos (Ahrefs, March 2, 2026). In July 2025, about 76% of cited pages came from the top 10.

What it means: SEO still feeds AI answers, but AI search pulls from a wider set of pages. A page that answers a buyer's exact question can be cited without ranking first, and video counts.

7. Buyers trust answers more when they cite third-party proof

45% of B2B software buyers said citations from software review sites are the most confidence-inspiring signal in an AI-generated answer, and 85% think more highly of a vendor when AI includes it in an answer (G2, April 15, 2026). G2 runs a review site, so weigh the first number accordingly. But it matches what the citation data above shows: AI answers lean on sources other than the vendor.

What it means: what others say about you is evidence, and AI answers are built from evidence. Your own claims are only the starting point.

8. Schema markup and llms.txt don't move citations much

Two of 2026's most useful studies tested popular tactics directly.

  • Schema: Ahrefs tracked 1,885 pages that added JSON-LD schema between August 2025 and March 2026, compared them with 4,000 control pages, and found no major uplift in AI citations on any platform (AI Overviews −4.6%, AI Mode +2.4%, ChatGPT +2.2%, after removing platform-wide trends) (Ahrefs, May 2026).
  • llms.txt: of 137,210 domains, 28% publish an llms.txt file, and 97% of those files received zero traffic in May 2026 (Ahrefs, June 15, 2026).

What it means: keep schema accurate because it supports search features, but don't expect it to win AI citations. Put the hours into visible content and third-party proof.

9. AI agents are arriving, and most sites aren't ready

When Cloudflare scanned sites for its Agent Readiness score, 78% had a robots.txt file, but most were written for traditional search crawlers, not AI agents. Only 4% declared AI usage preferences, and 3.9% served markdown when an agent asked for it. Serving markdown cut token use by up to 80% in Cloudflare's tests (Cloudflare, April 17, 2026).

What it means: in 2027 more "visitors" will be software reading your pages for a person or acting for one. Pages they can read quickly and completely have the advantage.

10. The scale is no longer a niche

ChatGPT has more than 900 million weekly active users and more than 9 million paying business users (OpenAI, February 27, 2026). In the US, about half of adults now use AI chatbots, up from a third in 2024, and 44% say they have used ChatGPT; 38% of employed adults use chatbots for tasks at work (Pew Research Center, June 17, 2026).

11. Evaluation is now the longest stage of B2B buying

40% of buyers say evaluation is the longest stage of their buying journey, up from 36% last year, and IT security review is the biggest single source of delay (39%) between choosing a vendor and completing the purchase (G2, 2026 Buyer Behavior Report). More than 60% of business buyers now use a trial to evaluate solutions (Forrester, January 21, 2026).

What it means: AI questions don't stop at "who should I consider". Buyers also ask about security, integrations and pricing. If those answers live in a PDF behind a sales call, the AI answers without you.

12. AI search finally has first-party measurement

2026 is the year the platforms started showing site owners their AI numbers:

  • Google Search Console launched generative AI performance reports on June 3, 2026, and rolled them out to all websites worldwide by August 31 (Google Search Central).
  • Bing Webmaster Tools added an AI Performance report in February, then intents, topics, citation share and competitor comparison in June (Bing, June 16, 2026).
  • ChatGPT's Business Hub went live on November 10: businesses can claim and verify a profile and see an Insights beta with mentions, views, website clicks and discovery topics. A "mention" counts when ChatGPT references the business's website, not just its name (Digiday, November 10, 2026).

What it means: "we can't measure it" is no longer a reason to wait.

Part 2: The seven moves for 2027

These moves are our reading of the evidence above, not a forecast. Each one names the findings it rests on, so you can judge it yourself.

Move 1: Start from your buyers' questions, not your keywords (findings 1, 2, 6)

Buyers ask AI three kinds of questions: about the category ("best [category] for [team size]"), about competitors ("[competitor] alternatives", "[you] vs [competitor]") and about requirements ("works with [tool]", "SOC 2", "under $[budget]"). Write down 30–50 of them from sales calls, win/loss notes, support tickets and the questions prospects ask in demos. That list becomes your plan and your scorecard. (How to pick them: How to Choose the 40 Buyer Questions That Matter for AI Search Tracking.)

This week: ask four assistants five category questions each and note who gets named, how you're described, and which sources are cited.

Move 2: Get your position sharp enough to repeat (findings 1, 7)

An AI can only recommend you for something it can say in one sentence. "AI-powered revenue platform" gives it nothing to work with. "Lead routing for inbound sales teams under 50 reps" does. Pick the one job you're the best answer for, name who it's for in your buyers' own words, and use that same sentence on your homepage, LinkedIn, review profiles and directory listings.

This week: check whether your homepage, LinkedIn About section and top review profile describe you the same way. If they don't, fix that first.

Move 3: Publish the pages AI quotes when buyers compare (findings 6, 11)

Buyers compare, so AI compares. The pages that answer those questions are:

Page Answers
[You] vs [Competitor] "Which is better for us?"
[Competitor] alternatives "What else should I look at?"
Best [category] for [use case] "Who fits a team like ours?"
Pricing, or how pricing works "Can we afford it?"
Security, integrations, implementation "Will this pass review?"

Write them in plain text, with specific facts, and be honest about when another product is the better fit. That honesty makes a page credible enough to cite.

This week: publish your security, integrations and pricing facts as an HTML page anyone can read, not a gated PDF.

Move 4: Earn proof on the sites AI already trusts (findings 5, 6, 7)

AI answers come from consensus. If your site says one thing and your reviews, Wikipedia entry and community threads say nothing, the answer goes to whoever has the evidence. Work on:

  • Review sites: current profiles with recent reviews that mention the use cases you want to be known for.
  • Wikipedia, YouTube and LinkedIn: the three sources AI Overviews cited most often for brand searches (finding 5).
  • Roundups and comparisons: find the articles AI cites for your category questions in Move 1, then make sure their authors have accurate, current facts about you.
  • Communities: answer real questions in detail, under your own name, without a pitch.

This week: take the sources AI cited in your Move 1 test and sort them by how often they appear. That's your outreach list.

Move 5: Treat your brand search as a page you don't own (findings 5, 12)

Your brand's AI summary is now the first page most buyers see. Search your company name in Google, ChatGPT, Gemini, Copilot and Perplexity, and check every fact: what you do, who it's for, pricing, leadership, product names. Old facts live on in third-party pages long after you change them.

This week: claim and verify your profile in ChatGPT's Business Hub if you're eligible, make sure its details match your website, and save the first mention count as a baseline.

Move 6: Let AI search in, and make your pages easy for agents to read (findings 8, 9)

  • Allow the AI search crawlers in robots.txt and in your CDN or firewall bot rules: OAI-SearchBot (ChatGPT search), Claude-SearchBot, PerplexityBot, and Googlebot and Bingbot, which feed AI Overviews and Copilot. Blocking training crawlers like GPTBot is a separate decision.
  • Put key facts in the HTML, not only in scripts, images or PDFs.
  • Don't spend your 2027 budget on schema or llms.txt for citations (finding 8). Keep schema accurate; move the hours to content and proof.

This week: check your robots.txt and firewall rules for AI search crawlers.

Move 7: Measure the answer, not just the click (findings 3, 4, 12)

Use a monthly scorecard with five lines:

  1. Presence rate: for your fixed set of buyer questions, how often you're named (for example, "named in 9 of 12 runs"), by assistant.
  2. Accuracy: how many answers describe you correctly.
  3. First-party AI reports: Search Console's generative AI reports, Bing's citation share, ChatGPT Business Hub mentions.
  4. AI referrals: a separate channel for AI assistants in your analytics, tracked against pipeline, not sessions.
  5. Self-reported attribution: a "How did you hear about us?" field on your forms, with "ChatGPT or another AI assistant" as an option.

This week: add the attribution field. It's the cheapest measurement you'll ever run, and it catches the private-AI research from finding 2.

What to stop doing in 2027

  • Judging AI search by traffic alone. AI chatbots sent 0.28% of web traffic in March 2026, yet 69% of buyers changed vendors because of an AI answer.
  • Gating the facts buyers need to evaluate you. If the answer is behind a form, the assistant can't read it.
  • Adding markup and hoping. The 2026 tests found schema and llms.txt barely move citations.
  • Describing yourself five different ways. Inconsistent descriptions make you harder to recommend.

What the data can't tell us yet

  • How much of B2B research happens in private AI tools. Forrester's 61% suggests a lot, and none of it shows up in referral reports.
  • How much of ChatGPT's Business Hub will cover B2B companies. At launch, the business agent feature was limited to US e-commerce and online services businesses.
  • How ads inside AI answers will change organic recommendations. ChatGPT now sells placements for business agents. It's too early for evidence on the effect.

We'll update this guide as new studies land.

Your first 30 days

Week Do this
1 Write 30–50 buyer questions. Run them in four assistants. Record presence, accuracy and cited sources.
2 Fix your one-sentence position everywhere. Check robots.txt and firewall rules. Add the attribution field.
3 Publish or update your top comparison, pricing and security pages in plain HTML.
4 Work through the top sources from your week 1 list. Claim your ChatGPT profile if eligible. Set the monthly scorecard.

Want the week 1 baseline done for you? The free AI Visibility Snapshot shows how AI assistants describe your company today, with no account needed.

Sources

Frequently asked questions

What is AEO for B2B companies?

Answer engine optimization (AEO) is the work of making sure AI assistants such as ChatGPT, Google AI Overviews and AI Mode, Gemini, Copilot, Claude and Perplexity can find, understand and recommend your company when buyers ask about your category, compare vendors or check requirements.

How many B2B buyers start their research in AI chatbots?

In G2's March 2026 survey of 1,076 B2B software buyers and decision-makers, 51% said they now start software research with an AI chatbot more often than with Google, up from 29% in April 2025. 71% rely on AI chatbots at some point in their research.

Does schema markup or llms.txt help you get cited by AI?

Not much, according to 2026 data. Ahrefs tracked 1,885 pages that added schema and found no major uplift in AI citations on any platform. Ahrefs also found that 97% of llms.txt files received zero traffic in May 2026. Schema is still useful for accuracy in search features; it is not a citation lever.

How do I measure AI search performance in 2027?

Combine the new first-party reports (Google Search Console's generative AI performance reports, Bing Webmaster Tools' AI Performance and citation share, and ChatGPT's Business Hub Insights beta) with a fixed set of buyer questions you test monthly across assistants, AI referral traffic in your analytics, and a self-reported attribution field on your forms.

Is AI search traffic big enough to matter for B2B?

As a share of visits, not yet: Ahrefs measured AI chatbots at 0.28% of total web traffic in March 2026. But the influence is far larger than the clicks. Buyers form shortlists inside the answer, and Demandbase saw ChatGPT referrals to B2B websites rise 303% between June 2025 and June 2026.

See How AI Describes Your Brand

Discover where you stand in ChatGPT, Gemini, and Perplexity — and how Genmark AI can help you get cited.

Check Your AI Presence