Shortlisted by Machines: The New Rules of Being Found for B2B

The shortlist is written before you're invited

A procurement lead in Dubai opens an AI assistant and types: "Which firms help mid-sized logistics companies in the Gulf with cyber resilience? Give me three." Thirty seconds later she has a confident shortlist, a line on each firm and a handful of sources. Before lunch it is in the inbox of a buying group of more than a dozen colleagues.

Two of those firms will get a call. If yours is not on the list, there is nothing for anyone to check, no proposal to compare and no meeting to win.

For two decades, being found in B2B meant two things: rank in Google and get mentioned in the press. Both still matter. But the shortlist is increasingly assembled by machines that read the whole web, weigh who is talked about and repeat what independent sources agree on. Being found now means being the brand that machines and people both vouch for.

This article sets out what has changed, why AI recommends some brands and not others, and ten rules B2B leaders can act on now. Every figure is sourced, and we flag where the evidence comes from vendors rather than independent research.

What actually changed

AI now shapes who gets considered, but people still decide who wins. That is the single most useful way to read the last two years of buyer research.

The cleanest independent signal comes from Forrester. Its 2025 Buyers' Journey Survey found 94% of business buyers now use AI somewhere in the buying process, and twice as many name generative AI as their most meaningful information source as name anything else. That puts AI ahead of vendor websites, product experts and sales conversations.

Vendor-run surveys point the same way, more dramatically. G2's April 2026 report says 51% of software buyers now start research in an AI chatbot more often than in Google, up from 29% a year earlier. Treat that as directional: G2 sells into the market it is describing.

Now the counterweight. Gartner's May 2026 survey of 645 buyers found 69% validate AI-generated insights with a sales rep. Forrester also found around one in five buyers felt less confident because AI gave them inaccurate or misleading information. Buyers use AI to build and stress-test the list, then turn to people and proof to make the call.

The traffic picture follows. Pew Research Center tracked real Google searches and found users clicked a traditional result 8% of the time when an AI Overview appeared, against 15% when one did not. Pew is clear this is observational, not causal. Seer Interactive adds the nuance that matters most: brands cited inside the AI Overview earned 35% more organic clicks and 91% more paid clicks than those left out.

So the unit of success has shifted. It is less about the click and more about the mention: being named, described accurately and trusted in an answer that may never send a visitor.

Why AI recommends some brands and not others

AI recommends brands that other people talk about. Think of an answer engine as a cautious researcher: it repeats what many independent sources agree on, and stays quiet about what it cannot corroborate.

It works on two layers. The first is memory: what the model absorbed about your brand during training, which reflects how often and how consistently the web discusses you. The second is retrieval: for live answers, ChatGPT Search, Perplexity, Gemini, Copilot and Google's AI features run their own searches, often several related ones at once, and synthesise what they find. Both layers reward the same thing, which is frequent, consistent, third-party corroboration.

The evidence for that third-party bias is now broad. A University of Toronto study (Chen et al., 2025) found AI search leans heavily towards earned media compared with Google; in one category, earned sources made up 81.9% of AI results against 45.1% for Google. A 2026 multilingual study found AI grounded brand answers in third-party sources 85.7% of the time. Both are preprints, not yet peer reviewed.

Large commercial datasets agree. Muck Rack's May 2026 analysis of more than 25 million cited links found 84% came from earned media. In Ahrefs' study of 75,000 brands, branded web mentions correlated with AI Overview visibility at 0.664, roughly three times more strongly than backlinks at 0.218. A follow-up found YouTube mentions correlated more strongly still.

A caution worth stating plainly: these are correlations from companies that sell SEO and PR tools. The honest reading is not "mentions cause recommendations" but "brands that are widely and credibly discussed get recommended". For B2B leaders, that is still a clear instruction: your reputation off-site now shapes your visibility on-screen.

The ten new rules

Rule 1. Win the shortlist before the search

Buyers rarely arrive at AI with a blank page. They bring a mental shortlist built from experience and reputation, then use AI to expand, compare and challenge it. In 6sense's 2025 Buyer Experience Report, the eventual winner was on the Day One shortlist 95% of the time. In Singapore, Green Hat and 6sense put it at 98%.

The implication for leaders is uncomfortable but simple. Brand awareness is no longer a separate budget line from demand generation. It is an input into the AI answer itself.

Rule 2. Be talked about, not just linked to

Backlinks built SEO careers. In AI visibility, mentions carry more weight: credible people and publications naming your brand, in context, for the problems you solve. Link-building for its own sake is now a weaker lever than becoming a brand your market discusses.

Practically, that means giving people something to talk about: original data, a clear point of view, executives who say something specific, and customers willing to be named.

Rule 3. Earn coverage where AI actually reads

Earned media dominates AI citations, but most PR is aimed at the wrong targets. Muck Rack found only 2% overlap between the journalists PR teams pitch and the journalists AI cites. AI also leans on journalism most for industry-trend questions, exactly where B2B leaders want to be quoted.

Audit which outlets the engines cite for your category, in your markets. Then feed those outlets what they need: research, commentary and stories with numbers in them.

Rule 4. Let customers make your case

Reviews are a floor, not a lever. Analysis of ChatGPT software recommendations found almost every recommended tool had G2 and Capterra profiles, yet review volume explains under 2% of citation variance. Being absent from the right review platforms can rule you out; piling up more reviews rarely moves you up.

So get present on the platforms that matter for your category: G2, Capterra and Gartner Peer Insights for software; Clutch and Google Business Profile for services. Then invest in named case studies and customer voices. They create independent mentions for the machines and arm the human validation step that, per Gartner, still decides most deals.

Rule 5. Show up where buyers and machines both listen

Semrush's study of 325,000 prompts found LinkedIn appeared in 11% of AI answers on average, and around 95% of the LinkedIn posts cited were original content rather than reposts. YouTube and community discussions also feature heavily.

The common thread is substance. Engines quote specific, original insight from real people, not corporate announcements. A partner's sharp post on regional procurement rules will do more than a polished brand video.

Rule 6. Treat SEO as the entry ticket, not the finish line

Google's May 2026 guidance is explicit: AI Overviews and AI Mode run on the same index as search, with no special optimisation required. Technical health, crawlability and genuinely helpful content are non-negotiable, because Google cannot cite what it has not indexed.

But ranking no longer guarantees being quoted. Ahrefs found only 12% of links cited by ChatGPT, Gemini and Copilot appear in Google's top ten for the same prompt. Answer engines fan out into many sub-questions, so cover the full set of questions your buyers ask, not just the head term.

Rule 7. Write to be quoted, and keep it current

Engines lift statements that are specific and verifiable: how pricing works, typical implementation timelines, honest comparisons, original benchmarks, a clear definition of who you serve and where. Generic copy gives them nothing to quote.

Keep key pages maintained. Ahrefs' analysis of 17 million citations found AI-cited pages were about 25.7% fresher than Google's organic results. And describe your company the same way everywhere: website, LinkedIn, directories, review profiles and press boilerplate. Consistency reduces the chance a model confuses or omits you.

Rule 8. Open the door to the right bots, and ignore the hacks

Make sure answer and search crawlers can reach your site: Googlebot, Bingbot, OAI-SearchBot, Claude-SearchBot and PerplexityBot. Decide separately whether to allow training crawlers. Check your CDN settings too, because some block AI bots by default.

Be sceptical of GEO hacks. SE Ranking's 300,000-domain study found no link between llms.txt and AI citations, and Google says no special files or markup are needed. Anyone selling a shortcut is selling a story.

Rule 9. Measure presence, not position

AI answers are not stable rankings. SparkToro and Gumshoe ran the same prompts nearly 3,000 times and got the same list of brands fewer than once in a hundred runs. A "rank in ChatGPT" is meaningless. How often you appear across many runs is not.

Track share of presence across a fixed set of buyer prompts, several engines and each market, plus how accurately you are described and which sources are cited. Pair that with AI referral traffic, branded search trends and a simple "How did you hear about us?" field, because much AI influence arrives as direct or branded traffic rather than a tidy referral.

Rule 10. Localise your proof for the market you sell into

The UAE (73.3%) and Singapore (64.3%) lead the world in generative AI use, against a global average of 18.8%, and Google's AI Mode now runs in Arabic. Research also shows the sources AI uses to describe a brand change by language, market and engine.

A single global English strategy will not carry you across the Gulf and Southeast Asia. Regional trade press, local customer stories, local review profiles and Arabic content where your buyers use it all feed a different answer set.

What not to do

The fastest way to waste budget in this space is to chase tricks instead of reputation. A few traps we see repeatedly:

• Publishing AI-generated content at volume. Engines look for specific, non-commodity insight, and generic content gives them nothing to quote.
• Buying mentions or paid placements dressed up as editorial. They are fragile, and they erode the trust that makes recommendations stick.
• Judging success on a single prompt's answer. One screenshot proves nothing when answers change on almost every run.
• Treating llms.txt or special markup as a strategy. The evidence does not support it.
• Blocking answer crawlers by accident through a default security or CDN setting, then wondering why you have vanished.

Reputation, made measurable

The brands AI recommends are the ones the market already vouches for. None of these rules is a trick. They are old marketing truths (reputation, proof, relevance) made urgent by a new gatekeeper and, for the first time, measurable.

For B2B leaders, the shift is organisational as much as tactical. SEO, PR, customer marketing, sales enablement and leadership visibility used to sit in separate plans with separate owners. In an AI-shaped buying journey they feed the same answer, so they need a shared view of how your brand is being described, where and by whom.

This is the work we do at ViMi Digital. We help B2B teams across Asia and the Gulf understand and improve how they show up when buyers ask AI for a shortlist. Our AI Visibility Health Check maps what ChatGPT, Gemini, Perplexity, Copilot and Google's AI features say about you across your priority buyer questions and markets, which sources they lean on, and where competitors are being recommended instead. Alongside it, our signal engineering work makes sure the demand that does reach you is tracked and attributed properly.

How we rated the evidence

We separate independent research from vendor-published data. Independent or official sources here include Forrester, Gartner, Pew Research Center, Google Search Central, Microsoft and academic preprints. Large-sample studies from Ahrefs, Semrush, SE Ranking, Seer Interactive, Muck Rack and SparkToro are robust but published by companies that sell related tools. Buyer surveys from G2, 6sense and Green Hat are useful for direction, and each publisher benefits from its conclusions. Most AI-citation data is correlational, so we say "linked to", not "causes". Figures are current as of October 2026.