When someone asks ChatGPT or Google for “a good plumber in Leeds” or “the best independent bike shop near York”, which businesses get named? This study measures how often UK small businesses appear in AI answers to everyday recommendation questions, which sectors are most and least visible, and what the businesses that appear have in common.
[PAUL TO ADD: headline findings in two or three sentences, once the data is collected.]
At a glance
| Question | How visible are UK small businesses in AI answers to local and product recommendation questions? |
| Platforms | Google AI Overviews, Google AI Mode, ChatGPT search, Perplexity |
| Sectors | 10 |
| Locations | 10 UK towns and cities |
| Prompts | 300 |
| Collection dates | [PAUL TO ADD] |
| Author | Paul Myers |
Why this study
Small business owners are being told AI search will either sink them or save them, mostly without evidence. Google says its AI features draw on its normal ranking systems, and the AI assistants document their own crawlers, but nobody publishes how often small, independent businesses are actually named in the answers people get. This study measures that directly, with a method anyone can repeat.
Method
What’s measured
For each prompt on each platform, the study records:
- every business named in the answer
- whether each named business is linked
- the position of each business in the answer (first named, second and so on)
- the sources cited or linked
- whether each named business is an independent small business, a regional chain, a national chain or a marketplace or directory
A small business is defined as an independent business with fewer than 50 employees, checked against its website and Companies House where possible.
Sectors
The proposed sectors reflect the small businesses most likely to be found through local and recommendation searches:
- Trades: plumbers and electricians
- Home improvement: builders and kitchen fitters
- Health: dentists and physiotherapists
- Beauty: hair and beauty salons
- Hospitality: independent restaurants and cafés
- Professional services: accountants and solicitors
- Retail: independent shops
- Fitness: gyms and personal trainers
- Automotive: garages and MOT centres
- Pets: groomers and vets
Locations
Ten UK towns and cities of different sizes across England, Scotland, Wales and Northern Ireland, so results aren’t dominated by London.
Prompts
Each sector has prompts in three styles, reflecting how people ask AI tools rather than how they type into a search box:
| Style | Example |
|---|---|
| Direct recommendation | “Recommend a good plumber in Leeds” |
| Criteria-led | “Which dentists in Cardiff are good with nervous patients?” |
| Comparison | “Who are the best independent gyms in Bristol and what makes them different?” |
The proposed set is 10 sectors, 10 locations and 3 prompt styles, giving 300 prompts per platform. The full prompt list will be published with the results.
Collection
- Prompts run in fresh sessions, signed out where the platform allows, with location set to the UK
- Each prompt runs three times on separate days to account for answers changing between runs
- Answers are saved in full with the date, time and platform
- Business classification is checked by hand, not inferred automatically
Exclusions
Prompts that return no businesses, refusals and answers made up only of directories are recorded and reported, but excluded from the share-of-mentions figures.
Findings
[PAUL TO ADD after collection. Planned sections:]
How often small businesses are named
[Share of named businesses that are independent small businesses, by platform.]
Sector differences
[Most and least visible sectors, with a chart explained in the text.]
Platform differences
[How Google’s AI features and AI assistants differ in who they name and link.]
What visible businesses have in common
[Observed characteristics such as reviews, a Google Business Profile, clear service pages or mentions in local press, reported as correlations, not causes.]
Limitations
AI answers change between sessions, days and locations, and platforms update their systems often, so the results describe a point in time. The study observes which businesses are named, not why, and can’t show that any characteristic causes visibility. Prompts written for the study may not match how every customer asks. Results for one town or sector shouldn’t be read as true for all.
Data
[PAUL TO ADD: download link for the anonymised dataset and full prompt list, once published.]
What it means for small businesses
[PAUL TO ADD after the findings. This section will set out practical steps drawn from the results.]
In the meantime, the AI search guide explains what the platforms themselves say about being shown, and the free SEO checker tests whether a page is ready for AI search.
Using this study
You’re welcome to quote and link to this research. Please credit “UK small business visibility in AI search” by Paul Myers and link to this page. How I approach research and corrections is set out in the methodology and editorial policy.
If you’d like to know how your own business appears, AI visibility tracking measures it over time, or you can get in touch.