AI tools like ChatGPT recommend businesses based on information they've absorbed from across the web — primarily review sites, directories, news articles, press releases, and well structured website content. 68% of AI citations come from third party sources rather than the business's own website, so what other people say about you matters far more than what you say about yourself.
Source 1: Review platforms
Google Reviews, Trustpilot, Facebook reviews, Yelp, and trade specific platforms like Checkatrade and TrustATrader are all ingested by AI models during training. The AI looks at volume (how many reviews you have), sentiment (whether they're positive or negative), and recency (how recent they are).
A business with 150 recent Google reviews averaging 4.8 stars sends a much stronger signal than one with 5 reviews from three years ago. AI models pick up on these patterns and use them to decide who to recommend with confidence.
The words used in reviews matter too. If multiple customers mention you're "quick to respond" or "great with boiler installations", the AI associates those qualities with your business. This is why asking for detailed reviews — not just star ratings — helps your AI visibility.
Source 2: Business directories
Directory listings on Yell.com, Bark, FreeIndex, Thomson Local, and niche directories act as confirmation signals. When AI sees your business listed consistently across multiple directories with the same name, address, and services, it builds confidence that you're a real, active business.
Inconsistency hurts. If you're "Smith Plumbing" on one directory and "J Smith Plumbing Services Ltd" on another, AI might treat those as two different businesses. Keep your name identical everywhere.
Source 3: News and press coverage
News articles, press releases, and media mentions are some of the strongest signals AI models use. A business mentioned in a press release distributed to 100+ news sites creates a dense web of third party mentions that AI training data picks up.
This is why press release distribution is the single fastest way to boost your AI visibility. One release through a service like EIN Presswire creates mentions on Google News, Bing News, Apple News, and hundreds of industry outlets — all the kinds of sources that feed directly into AI training.
Source 4: Your website (with conditions)
Your own website does matter, but less than you'd think. AI tools read your site looking for clear, factual information: what services you offer, where you're located, how long you've been trading, and what makes you different.
Two things determine whether your website content actually reaches AI tools. First, you need to not be blocking AI crawlers through your robots.txt file — many sites do this accidentally. Second, you need structured data (schema markup) that tells AI exactly what your business is in a machine readable format.
Vague marketing copy doesn't help. "Premium quality services" means nothing to AI. "Emergency plumber available 24/7 in Leeds, LS1–LS29, established 2008" gives AI everything it needs.
Source 5: Social media and forums
AI models also train on social media posts, Reddit threads, forum discussions, and community groups. If people are mentioning your business in local Facebook groups or recommending you in Reddit threads, that feeds into AI training data.
This is harder to control directly, but it's another reason why delivering good work and encouraging word of mouth matters. Organic social mentions are some of the most authentic signals AI can find.
Source 6: Wikipedia and knowledge bases
For larger businesses, Wikipedia articles and Wikidata entries carry enormous weight with AI models. These are treated as highly authoritative sources. Most small businesses won't have a Wikipedia page, but this is worth knowing: if your business does have a Wikipedia entry, make sure it's accurate and up to date.
The hierarchy of influence
If you want to prioritise, here's roughly how much each source type influences AI recommendations for a typical small business:
Highest impact: press coverage and news mentions across multiple sites. High impact: review volume and sentiment across multiple platforms. Medium impact: consistent directory listings, well structured website content. Lower impact: social media mentions, forum discussions.
The key insight is that no single source is enough on its own. AI looks for consensus across multiple independent sources. The more places that say the same positive things about your business, the more confident the AI becomes about recommending you.
Your action plan
Start with the highest impact actions. Distribute a press release. Build up your review profiles. Make sure your directory listings are consistent. Fix your website's technical access and content quality. Each one adds a layer of evidence that AI tools use to decide whether to recommend you.
Want to know exactly which sources are working for you and which are missing? Our GEO Starter audit checks all of them and gives you a prioritised plan. Or get in touch and we'll walk you through it.