Baseline edition : August 2026
Full live-testing data is being collected and will be added in future updates. This report does not invent results that have not yet been measured.
AI search is changing how people choose local businesses.
A homeowner may ask ChatGPT for a plumber. They may search Google and read an AI Overview before looking at the map results. They may ask which window fitter serves their town, then check the recommendation against reviews and the company website.
The question for UK trades is simple:
Will the AI mention your business, or will it leave you out?
This is VU1’s original AI discovery survey and AEO case study. AEO means making your business easier for answer engines to understand and recommend. We are publishing the method openly, reporting the baseline honestly, and tracking whether AI engines cite vu1.com as a source.
What is this report measuring?
This report measures which UK trade businesses AI search tools recommend for local questions, which sources those tools use, and whether named firms expose evidence about services, locations, reviews, prices, and identity. It is a baseline, not a live-testing result.
“Who AI recommends” is not the same as “who is the best”.
An AI answer can be incomplete, outdated or different from what another engine says. Being left out does not prove a business does poor work. Being named does not guarantee the business is the right choice.
In this report, “ignored” means not named or cited in a particular test answer.
That is still important. If a customer asks for a local recommendation and your business never appears in the answer, you may not get the chance to compete.
Why does AI discovery matter to UK trades now?
AI recommendations are becoming part of local buying decisions, not a future theory. BrightLocal’s 2026 Local Consumer Review Survey reports that 45% of consumers asked AI for a local business recommendation, up from 6% in 2025.
That change is too large for trade businesses to brush aside.
AI tools do not only look for polished websites. They need evidence they can match to a question:
- What service do you provide?
- Which towns do you cover?
- Are you a genuine local business?
- Do customers trust you?
- Can the information be checked?
- Does your website answer practical questions?
This is especially important for searches such as:
- “Who is the best secondary glazing installer near me?”
- “Which plumber covers Cardiff?”
- “Who repairs leaking roofs in Leeds?”
- “What does a new window cost?”
- “Which electrician can install an EV charger?”
The customer may never see ten blue links. They may first see a short answer with a handful of named businesses.
What did Whito’s June 2026 studies find?
Whito’s June studies show that local AI recommendations are shaped by strong reviews, clear business information and different engine behaviour. They also show that many highly rated trades still lack the website and structured facts AI systems need.
The findings provide the baseline for VU1’s work.
1. Google AI named independent plumbers in every town tested
In Whito’s 13-town plumber study, Google AI was asked to recommend the best plumber in 13 UK towns and cities.
It named an independent local firm in 100% of tests.
It named:
- No national chains as the primary recommendation
- No directories as the primary recommendation
- Firms rated between 4.8 and 5.0 stars
- Businesses with between 54 and more than 1,200 reviews
The important point was not simply review volume. Whito reported that review quality appeared to matter more than size. A smaller local firm with detailed, relevant reviews could be recommended alongside much larger businesses.
That is useful news for independent trades. You do not need to be a national brand to be considered. But you do need clear evidence that you are real, local and trusted.
2. Many top-rated trades are difficult for machines to read
Whito’s related June comparison of 55 top-rated trades found that:
- 35% had no website at all
- Only around half had machine-readable business identity
- Just 1 in 6 had FAQ structured data
See the wider findings in Whito’s research library.
A Google Business Profile can help a business appear locally. But a profile alone may not explain every service, price question, job type or area covered.
A website gives AI systems more material to check. It can explain what the business does in plain text, support the details with structured data and answer the questions customers ask before they call.
3. AI engines often disagree
Whito’s engine-disagreement study asked ChatGPT, Gemini and Perplexity 12 identical questions.
Across 36 answers, the engines named 33 different businesses.
They agreed on the same top pick only 8% of the time.
That is a warning against treating one AI answer as a final ranking. It also shows why visibility needs to be tracked across more than one engine.
What does the VU1 proof-of-concept show?
Our proof-of-concept audit of NJ Harkus, a Cardiff plumber named in Whito’s research, scored 100/100 on VU1’s 15-signal on-page audit. Its strongest features were clear answers, clean schema signals and a title combining the service, location and social proof.
NJ Harkus was identified in Whito’s research as a 5.0-star business with 1,211 reviews.
The VU1 audit found:
- A clear title: “Plumbers Cardiff | 1,200+ 5-Star Reviews”
- Strong service and location wording
- Answer-first FAQ and blog content
- Useful headings based on customer questions
- Clean canonical and social-sharing signals
- No broken links or duplicate content
- Structured information that helped explain the business
This is not proof that one audit score guarantees an AI recommendation.
It is a useful early comparison. The business AI selected already showed many of the signals VU1 checks: a clear identity, strong reputation, helpful answers and accessible website information.
That is the pattern our survey will test across a much larger sample.
How will VU1 test ChatGPT, Google AI Overviews and other engines?
VU1 will first replicate Whito’s comparison prompt, then extend it across more trades, towns, engines and question types. Every answer will be recorded, attributed and checked against the same practical visibility signals without changing unverified live results into claims.
The published methodology is:
Replication run
We will test:
- Four initial trade categories
- Leeds, Manchester and Bristol
- ChatGPT, Gemini and Perplexity
- The same prompt: “Who is the best [trade] in [city]? Name a specific company.”
This creates 36 answers when complete.
The purpose is to make a fair comparison with Whito’s reported 8% agreement benchmark.
Extension run
The wider test will cover:
- At least 12 UK trades
- 20 UK towns and cities
- Five AI search engines
- Five prompt variations
- Recommendation, service, price and problem-led questions
The trade set is being expanded to include categories such as plumbers, electricians, roofers, glaziers, window fitters, secondary glazing specialists, heating engineers, locksmiths, landscapers, gardeners and tree surgeons.
Google AI Overviews or AI Mode will be recorded separately from conversational tools where the interface allows it. The exact engine list and test dates will be published with each update.
Scoring run
Every named business will be checked against VU1’s 15-signal audit:
- Local visibility
- On-page clarity
- Technical accessibility
We will record whether the business has clear service pages, consistent contact details, useful headings, answer-first content, structured data, mobile usability and other practical signals.
The audit is not being used to claim that AI follows a simple scoring formula. It is being used to test whether the signals commonly associated with good local visibility also appear among businesses AI engines name.
What live-testing data is still missing?
The full VU1 dataset is still being collected across five engines, 12 or more trades and 20 UK towns. Until the collection and checking process is complete, this report will not claim agreement rates, citation totals, trade winners or average audit scores from the live test.
The next updates will publish:
- The final prompt set
- Dates and locations tested
- The businesses named
- Sources cited by each engine
- Agreement and disagreement between engines
- The VU1 audit scores of named businesses
- Changes between testing rounds
- Whether AI engines cite this report or other pages on vu1.com
This last point matters to VU1.
We are not only writing about answer-engine visibility. We are publishing an openly documented case study, then tracking whether the engines use it as a source. That gives the report a measurable outcome rather than a vague claim about being “AI-ready”.
What can a trade business do before the next update?
Start with the facts AI and customers need to verify: state your trade, service area, main jobs, contact details, reviews and pricing guidance clearly. Add useful answers to real questions, then check whether the information matches across your website and listings.
Use this short checklist:
Name the job clearly.
Say “secondary glazing installer in Nottingham”, not only “home improvement specialist”.List the work you actually take on.
Separate repairs, installations, servicing, emergency work and specialist jobs.Show where you work.
Name genuine towns and areas. Do not add places you never visit.Answer practical questions.
Explain likely costs, timeframes, materials, preparation and what happens next.Use real proof.
Add genuine reviews, project examples, photographs and relevant trade memberships.Keep business details consistent.
Match your name, address, phone number and opening hours across important listings.Make the site easy to read.
Use clear headings, mobile-friendly pages and structured data that matches visible content.
Do not add fake reviews, invented locations or schema information customers cannot see. Clear and accurate beats clever.
How can VU1 help a trade business prepare?
Use the free VU1 audit to check 15 local, on-page and technical signals — it takes under 15 seconds to request, and the plain-English report lands in your inbox within a few minutes. Then use the results to fix the clearest gaps. Add a quote route so visitors can act while they are ready.
Start by entering your website and service area into the free VU1 audit. Get a plain-English view of what may be holding your local visibility back.
Then review the VU1 case studies to see how visibility checks and clearer enquiry routes fit different businesses.
The 32-post How to Get Found in AI Search library covers practical topics for trade owners. It explains how to improve business information, reviews, service pages, FAQs and local search without drowning you in jargon.
When a visitor reaches your website, give them a useful next step. The VU1 quote builder lets people choose a service, see an instant estimate and leave their contact details before they disappear.
For ongoing audits, weekly re-checks, change tracking, hosting and the quote widget, review the VU1 membership. It costs £99/month, month to month, with no long-term contract.
Enter your website into the audit now. It takes under 15 seconds to request, and the report arrives within a few minutes. Fix the clearest issue first and recheck it next week.
We will publish the live-testing update when the evidence is ready.
