Key takeaways
- Around 84 to 85% of the sources behind an AI's recommendation are third-party sites you don't own, so your website alone cannot get you named.
- Reviews and social proof drive the biggest slice of brand-evaluation citations (about 57%), with directories adding roughly another 17%.
- A short list of trusted UK sources (Google Business Profile, Companies House, Trustpilot, your trade register) beats a hundred spammy directory listings.
- Identical name, address and phone on every listing is what lets an AI confirm you are one real, trustworthy business.
- The Second Floor scored 0/48 in its own AI-visibility audit largely because it was missing from these sources. We are fixing it the same way we would fix yours.
If you want AI assistants to recommend your business, most of the work does not happen on your own website. It happens on everyone else’s. When ChatGPT, Google’s AI Overviews, Perplexity or Copilot suggest a company, the large majority of the sources they lean on are third-party sites the business does not own: directories, review platforms, trade registers and reference sites. Recent analysis puts that figure at roughly 84 to 85% of citations. Your website confirms who you are. These independent sources are what convince a machine you are worth naming.
That is uncomfortable for most UK businesses, because it is the part they have never controlled. It was uncomfortable for us, too. When we ran our own AI-visibility audit on The Second Floor, we scored 0 out of 48: not recommended by a single assistant for the queries that matter to us. A big reason was simple. We were missing from the directories and citation sources these systems read. We have been fixing it, and this guide is the map we used, written for UK businesses.
This is one piece of the wider answer engine optimisation picture. Here we focus on one high-impact part of it: the third-party directories and citations that tip an AI towards recommending you, which ones are worth your time in the UK, and how to get listed without doing more harm than good.
Why do third-party sources matter so much to AI?
Because AI systems trust independent corroboration far more than self-promotion. An AI does not take your word for it. It cross-references what you say about yourself against what everyone else says, and it treats the places you cannot easily edit as the more honest account. Your homepage is marketing. A trade register, a review platform or a directory listing is evidence.
The numbers back this up. As above, around 84 to 85% of the sources behind an AI’s brand answer are third-party sites, not owned domains. Dig into what kind of third-party sources, and the picture sharpens further. One 2026 breakdown of branded AI citations found reviews and social proof accounted for about 57%, directories another 17%, and a brand’s own About, FAQ and home pages barely 5% between them. In other words, the pages you spend the most time polishing are the ones AI weighs the least.
This is not the machine being awkward. It is doing what a sensible person does when asked to recommend a plumber: it asks around rather than reading the plumber’s own leaflet. (If the whole idea of optimising for AI answers is new to you, start with what GEO means and come back.) The practical takeaway is blunt. You can have the best website in your sector and still be invisible in AI answers, because the sources that decide those answers do not include your website.
Which types of source actually influence AI recommendations?
Five categories do most of the work: business listings, review platforms, sector and trade directories, encyclopaedic or entity sources, and local press. Get a foothold in the right ones for your sector and you give an AI enough corroboration to name you with confidence.
- Business listings. Google Business Profile, Bing Places, Apple Business Connect and general directories like Yell. These are the backbone. They confirm your name, location, category and contact details in the places search engines and assistants read first.
- Review platforms. Trustpilot, Feefo, Yelp and Google reviews themselves. This is the category AI weighs most heavily when deciding between two similar businesses, and the one most UK firms under-invest in.
- Sector and trade directories. Checkatrade and Which? Trusted Traders for trades, professional registers for regulated work, industry lists for B2B. AI leans on these for “best [type of business]” questions because they are pre-filtered by relevance.
- Encyclopaedic and entity sources. Companies House, Wikidata and, for notable organisations, Wikipedia. These feed the underlying knowledge graph that AI models use to understand you as a real, distinct entity. Wikipedia is the single most-cited source across both ChatGPT and Google’s AI Overviews.
- Local press and civic listings. Regional papers, chambers of commerce and business improvement district pages. For “near me” and location questions, this local, independent corroboration carries real weight.
The UK directories and citations that matter, by category
Focus on a shortlist of high-authority UK sources rather than chasing hundreds of listings. Below are the real, reputable platforms worth your time, grouped roughly by the categories above. You do not need all of them. You need the general listings, the review platforms, and the two or three sector directories your customers actually trust.
| Directory / platform | Who it is for | Why it matters for AI |
|---|---|---|
| Google Business Profile | Any UK business with a shopfront or service area | The single most important listing. Feeds Google’s entity graph, Maps and AI Overviews. Free. |
| Bing Places | Any UK business | Powers Bing, Copilot and some ChatGPT search results. Quick to claim, routinely ignored. |
| Apple Business Connect | Businesses wanting to appear in Apple Maps and Siri | Feeds the Apple ecosystem that millions of iPhone users ask for recommendations. |
| Yell | SMBs, trades and local services | Long-established UK directory that AI still reads for corroboration. |
| Trustpilot | Ecommerce, services, most B2C and B2B | The review platform AI cites most. An active profile alone sharply lifts visibility. |
| Feefo | Retailers and services wanting verified reviews | Invited, verified reviews carry a strong trust signal. |
| Yelp UK | Hospitality, food and local services | Feeds Google AI Mode and others, especially for “near me” questions. |
| Checkatrade, MyBuilder, Rated People, TrustATrader | Tradespeople (builders, plumbers, electricians) | Vertical authority. AI leans on niche directories for “best trade near me”. |
| Which? Trusted Traders | Vetted UK trades | A high-trust, human-vetted signal that stands out. |
| Trade bodies and registers (Gas Safe, FMB, RICS, Law Society, FSB) | Regulated and professional services | Official membership is strong proof of a real, credible entity. |
| Sector directories (Clutch, Crunchbase, industry lists) | B2B, tech, agencies and startups | Where AI looks for “best [type of firm]” shortlists. |
| Companies House | Every UK limited company | Free, official, maximum-authority proof you are a real registered entity. |
| Wikidata | Any verifiable business or brand | Structured entity data that AI models read directly to build their knowledge graph. |
| Wikipedia | Genuinely notable organisations only | The most-cited single source across ChatGPT and AI Overviews. Do not force it. |
| Local press and chambers (Leicester Mercury, Chamber of Commerce) | Local businesses with a news hook | Local, independent corroboration that AI trusts for regional questions. |
A word on prioritising. If you are a local trade or service business, your order is Google Business Profile, then reviews (Google and Trustpilot), then the one trade directory your customers rate, then Companies House and Bing Places. If you are a B2B or professional firm, lead with Companies House, your professional register, Trustpilot and the relevant sector directory. Everyone should claim Google Business Profile and Bing Places, because they are free, fast and feed the systems that matter most.
How do I get listed and keep my details consistent?
Claim each listing, fill it out completely, and use identical name, address and phone details (your NAP) on every one. Consistency is the whole game. When an AI or a search engine sees “The Second Floor, 12 High St, Leicester” in one place and “Second Floor Ltd, Unit 12, Leicester LE1” in another, it cannot be sure they are the same business, so it trusts you less. Here is the order of work.
- Fix your source of truth first. Decide the exact name, address and phone number you will use everywhere, matching your Companies House record and your Google Business Profile. Write it down. This is what every other listing must copy, character for character.
- Claim the free giants. Set up and verify Google Business Profile, Bing Places and Apple Business Connect. Complete every field: category, hours, services, description, photos. Half-finished profiles read as half-real.
- Add the review platforms. Create your Trustpilot profile (and Feefo or Yelp if they fit your sector) and, crucially, start actually collecting reviews. A dormant profile does little.
- Claim your sector and trade listings. Pick the two or three your customers genuinely use and complete them properly. Regulated professions should make sure their official register entry is public and correct.
- Shore up your entity sources. Check your Companies House details are current. Consider a Wikidata entry, which any verifiable business can have and which AI reads directly. Leave Wikipedia alone unless you are genuinely notable, because a page that gets deleted helps no one.
- Keep it fresh. AI systems favour recent information. ChatGPT and Perplexity lean towards content updated in the last month, and stale listings quietly lose ground. Revisit your key profiles each quarter and update anything that has changed.
There is a UK-specific bonus here. AI systems use a hierarchy of trust, and official government records sit at the top. Companies House, VAT registration and council business-rates records are legally required to be accurate, so they carry the most authority when a model is confirming you are a real entity. Get those right and everything else corroborates more easily.
Quality over quantity: which directories should you avoid?
Skip low-quality, pay-to-list, link-farm directories. A handful of accurate listings on trusted sites beats a hundred on spammy ones. The consensus among local SEOs is stark: roughly 25 listings on high-authority, relevant directories will do more for you than 150 on low-quality ones, and the low-quality pile can actively work against you.
Two things go wrong when you chase volume. First, mass automated submission tools spray slightly different versions of your details across hundreds of sites, creating exactly the NAP inconsistencies that make AI trust you less. Second, sites that exist only to host links and sell listings carry no authority, so a model reading them learns nothing except that you appear on sites nobody trusts. Google has spent years discounting citation spam, and AI models inherit that scepticism.
Red flags that a directory is not worth your time:
- It asks for payment purely to appear, with no editorial or verification step.
- It is stuffed with ads and unrelated businesses, with no clear sector or region focus.
- It offers to submit you to “500+ directories” automatically.
- You have never heard of it and neither, plausibly, has your customer.
The test is simple. Would a real customer in your sector recognise and trust this site? If not, an AI has little reason to either. Curated beats crawled, every time.
How does this connect to reviews and local AEO?
Directories get you listed. Reviews get you recommended. Local AEO ties both to the “near me” questions your customers actually ask. These are not three separate jobs. They are the same footprint seen from three angles.
Reviews are the strongest single lever in the mix. In a 2026 study of more than 800,000 AI answers, simply having an active Trustpilot profile lifted a brand’s citation rate from about 1% to over 53%, rising past 75% for businesses with 80-plus reviews that reply regularly. The same pattern holds for Google reviews, which do double duty by feeding both your local rankings and the AI answers that draw on them. We go deep on this in do Google reviews affect local rankings, and the short version is that a steady stream of recent, specific reviews is what turns “listed” into “recommended”.
Layer on the location question and you are into local AEO: showing up when someone asks an assistant to name a business near them. That depends on the same consistent listings and reviews, plus a complete, well-categorised Google Business Profile. And if your goal is the broad money question, “which [type of business] should I use”, the ground you are preparing here is exactly what feeds a mention when someone asks ChatGPT to recommend a business. Directories, reviews and local signals reinforce each other. Neglect one and the others work harder for less.
Where should you start?
Start with the free, high-authority listings you can fix this week, then build reviews steadily from there. You do not need a huge budget or a hundred submissions. You need the right handful, done properly and kept consistent.
A sensible first fortnight looks like this:
- Lock down your exact name, address and phone, and check it against Companies House.
- Claim and fully complete Google Business Profile and Bing Places.
- Set up Trustpilot and start asking happy customers for honest reviews.
- Claim the one or two sector or trade directories your customers actually use.
- Confirm your Companies House record is current and consider a Wikidata entry.
Do that and you have covered the sources doing most of the deciding, without touching a single spammy directory.
If you would rather not pick through all of this yourself, this is the work we do. We run the audit that shows exactly where AI cannot see you, then build the directory, review and entity footprint that gets you back into the answers. It is scoped monthly, with weekly updates and no lock-in, so you can see it working before you commit further. Book a free growth review and we will show you where you stand, and what it takes to get named. We know the drill: we scored ourselves 0/48 first. All killer, no filler.