Why does AI search keep ignoring small businesses?

It’s not a size problem, it’s a trust problem

There’s a comforting myth going round, that AI search only recommends big brands because the machines are biased toward name recognition. It isn’t really true, or at least it isn’t the main reason you’re being skipped. Ask an AI assistant to recommend a family-run bakery in Chorlton and it isn’t reaching for a supermarket chain, it’s looking for a bakery it can describe accurately without getting caught out. What actually determines whether you get recommended is whether the assistant can find consistent, verifiable facts about you quickly: your address, your hours, what you actually sell, and whether other people online back that up with reviews or mentions. A one-person joinery business in Levenshulme with a tidy, accurate website and a page of genuine Google reviews will out-rank a chain store with contradictory information spread across a dozen directory listings. Size isn’t the filter. Confidence is. AI models are trained to avoid recommending something they might get wrong, because getting it wrong, wrong hours, a shop that’s closed down, makes the assistant look unreliable. So being small isn’t your problem. Looking uncertain is.

The contradictions that quietly rule you out

Most small businesses have a handful of small, boring contradictions scattered across the internet, and each one chips away at how confident an AI system feels. Your website says you close at 6pm, your Google Business Profile says 5:30. Your old address on a directory from four years ago is still live, even though you moved across town last year. You trade as “Bolton Boiler Care” on Facebook and “BBC Heating Solutions Ltd” on Companies House, and nothing on either page tells anyone they’re the same business. None of that is dishonest, it’s just the normal mess that builds up over years of running a business without anyone tidying the digital paperwork. But an AI model reading all of that doesn’t see “slightly out of date,” it sees conflicting facts it can’t resolve, so the safest move is to leave you out of the answer entirely and recommend a competitor whose details line up. Fixing this isn’t glamorous work. It’s going through Google, Facebook, directories and your own site with a fine comb and making every fact match, everywhere.

No structured data means the machine has to guess

Structured data is the bit most business owners have never heard of, and it matters more than almost anything else here. It’s a small block of code on your website that spells out, in a format machines can read directly, exactly what you are: a business, at this address, open these hours, offering these services, charging around this much. Without it, an AI crawler has to guess your opening hours from a sentence buried in an “About” page, or work out what you actually do from marketing copy that never quite says it plainly. Machines are bad at guessing and they know it, so they either get it wrong or skip you altogether. Adding structured data, also called schema markup, doesn’t change how your site looks to a human visitor at all, it’s invisible unless you go looking for it in the page code. But it’s the difference between an AI system reading your business details like a form, and reading them like a riddle. If you’ve never checked whether your site has it, there’s a decent chance it doesn’t.

What a “thin” web presence looks like in practice

Thin doesn’t mean your website is ugly, it means there isn’t enough verifiable substance for an AI system to work with. A single-page site with a phone number and a paragraph of copy, no service pages, no location detail, no reviews linked anywhere, gives an assistant almost nothing to check you against. Compare that to a competitor with individual pages for each service, a Google Business Profile with forty reviews and recent photos, and a mention or two on local directories. The second business isn’t necessarily better at the actual job, plenty of brilliant sole traders have terrible websites, but it’s vastly easier for an AI model to verify, which means it gets recommended more often. The fix usually isn’t “write more words,” it’s “make what’s true about your business easier to find and confirm.” A clear services page, an active review profile and consistent listings across the handful of directories that matter will do more than a thousand-word blog post nobody asked for.