How Does AI Decide Which Businesses to Recommend?

AI recommends businesses it can verify, understand, and trust. It draws from your website, directory listings, reviews, and mentions across the web — cross-referencing these sources for consistency. It favours businesses whose information is clear, specific, and corroborated by independent sources. The quality of your online presence determines whether you're the recommendation or the business that never gets mentioned.

That matters because the shift is already well underway. Over 800 million people now use ChatGPT weekly (Sam Altman, DevDay, October 2025), and 58.5% of Google searches now end without a click to any website (SparkToro/Datos, 2024). When a customer asks AI "who makes the best bespoke furniture near me?" — the answer they get is the only answer that matters. If you're not in it, they don't scroll past you. You simply don't exist.

What is AI actually doing when it recommends a business?

Think of it like asking a knowledgeable friend. A good friend doesn't just repeat what one person told them — they check multiple sources, weigh the evidence, and give you a recommendation they're confident about.

AI does something similar. The Consilience Project — a research initiative studying how societies make sense of complex information — calls this process "sensemaking": building understanding from multiple independent sources rather than relying on any single one. The Project's focus is democratic governance, not AI search — but the parallel is useful. When ChatGPT recommends a business, it's performing a basic version of this same process. It pulls information from your website, directory listings, review platforms, and mentions across the web, then cross-references them to build a picture.

Research from Onely found that authoritative list mentions account for roughly 41% of what influences ChatGPT's recommendations, followed by awards and accreditations (18%) and online reviews (16%). The common thread: independent verification from multiple sources.

That mechanical process raises a harder question: should an algorithm be deciding which craftspeople customers find? Trades that have thrived on word-of-mouth for generations are now subject to a system that rewards online verifiability over workshop reputation. We don't have answers to that yet — we're working through it alongside everyone else. But understanding how the system works is the first step, whether you choose to optimise for it or not.

What signals does AI look for in your content?

Many business owners assume a well-designed website is enough. But AI can't appreciate your photography or your branding. It reads code, text, and data — and it's looking for specific things.

Structured data tells AI what your business does in machine-readable terms. Most small business websites don't have it, which means AI is guessing from page text alone — and often guessing wrong.

Direct answers to questions — the ones your customers actually ask — give AI something it can extract and cite. "How long does a bespoke kitchen take?" answered clearly on your site is more valuable to AI than a gallery of finished kitchens.

Consistency across platforms matters because AI cross-references your details across directories. When your name, services, and location match everywhere, AI gains confidence. When they conflict, it moves on.

Freshness counts too. Onely's research found that 71% of ChatGPT citations come from content published or updated between 2023 and 2025. Outdated information gets overlooked.

Why does the quality of your information matter more than the quantity?

Here's where it gets interesting. More content doesn't automatically mean more visibility. Based on current research, AI tends to discount vague, inconsistent, or unverifiable claims.

The Consilience Project describes a useful distinction between what they call "epistemic hubris" — overclaiming with false certainty — and "epistemic humility" — being honest about what you know and don't know. AI, in its own mechanical way, seems to favour the latter. Content that says "we specialise in oak kitchen cabinetry and have completed 40 projects in North London since 2015" gives AI something concrete to work with. Content that says "we're the leading kitchen designers in London" gives AI a claim it can't verify — and Onely's data shows that what drives AI recommendations is verifiable evidence (list mentions, reviews, accreditations), not unverifiable assertions.

The principle is the same one that makes shared public information reliable: claims backed by evidence from multiple independent sources are more trustworthy than assertions backed by nothing. AI operationalises this principle at scale.

How does this apply to a small craft business specifically?

If you're a furniture maker or kitchen designer reading marketing advice aimed at enterprise brands, most of it feels irrelevant. But here's the unexpected part: craftspeople often have a structural advantage in AI search.

AI rewards specificity and depth over scale. A furniture maker with 20 years of experience and detailed project descriptions — materials used, design decisions, the brief they were given — can outperform a national chain with generic service pages. As we found when testing AI recommendations across London trades, the businesses AI recommends aren't always the biggest or the best-known. They're the ones whose information is specific, verifiable, and consistent.

Industry data supports this: research from Neil Patel suggests that 40% of AI citations come from pages ranking beyond position 10 on Google. You don't need to dominate traditional search to win in AI search. You need AI to understand what you do and trust that you actually do it.

When does content quality matter less than you'd think?

Here's an honest counter to this article's thesis: for most small businesses right now, content quality is a second-order factor.

Completing your Google Business Profile, claiming listings on relevant directories (Houzz, Checkatrade, Bark), and actively gathering reviews across multiple platforms will do more for your AI visibility than any blog post — including this one. Onely's data bears this out: authoritative list placements and reviews together account for over half of what drives AI recommendations, while content quality operates as a supporting signal.

A furniture maker with a complete GBP, 20 verified reviews, and consistent listings on three directories will likely outperform one who writes thoughtful blog posts but hasn't claimed their GBP. Get the foundations right first. Content quality amplifies a strong base — it doesn't replace one.

We'd be overstating our case if we suggested otherwise. The businesses we see struggle most with AI visibility aren't the ones with poor content. They're the ones with no directory presence at all.

A note on our perspective

Full transparency: Findcraft helps craftspeople optimise their online presence for AI search. We have a direct business interest in this topic. That's exactly why we're sharing our methodology openly — and including perspectives that might suggest you don't need our services.

Everything in this article works whether you hire us or not. The five-step guide in our What Is AEO? post costs nothing to follow. But you should know where we sit when reading our take on how important content quality is for AI visibility.

Further reading

These are independent sources — none affiliated with Findcraft:

• Consilience Project — "Democracy and the Epistemic Commons" — the research that frames how shared information quality affects collective decisions, including AI recommendations.
• Onely — "How ChatGPT Decides Which Brands to Recommend" — data-driven breakdown of what actually influences ChatGPT's business recommendations.
• Microsoft — "Optimizing Your Content for Inclusion in AI Search Answers" — guidance from Bing's team on what their AI looks for in content.

Frequently asked questions

Content produced through the M.A.R.C. methodology — our framework for evidence-based, ethically-governed content.