Black And White Boy SEO, Newcastle upon Tyne, September 2026
Newcastle firms are known by name. Ask who to hire and most of them vanish.
The 2026 Newcastle AI Search Visibility Benchmark. 36 accountants, solicitors and financial advisers, 51 buyer questions, three AI engines, seven days, 1,071 answers. Every number below is recomputable from the raw answers.
Engines
ChatGPT, Google AI Mode, Perplexity
Window
16 to 22 September 2026
Runs
7 per question per engine
Author
Paul Clarke, Black And White Boy SEO
Ask for your firm's numbers
The headline numbers
Across all three categories, here is how often each engine named a firm when a buyer asked who to hire, against how often it cited the firm's own website when asked about it by name.
ChatGPT
19%
of buying answers name a firm
Per firm, all three categories, 1,260 firm-answers
75% own site cited when asked by name
6% first pick on a buying question
Google AI Mode
21%
of buying answers name a firm
Per firm, all three categories, 1,092 firm-answers
89% own site cited when asked by name
7% first pick on a buying question
Perplexity
16%
of buying answers name a firm
Per firm, all three categories, 1,260 firm-answers
94% own site cited when asked by name
6% first pick on a buying question
It started with my own score
So why did I run this? Because I ran the same test on my own business first and scored 8 out of 90 on the buying questions. The engines knew who I was. They never picked me. I wanted to know if that was just me. It isn't. The first thing I found was my own Cloudflare settings telling the AI crawlers to keep out. That was a quick fix. Getting picked is not.
Recognised when named. Rarely chosen when it counts.
For each firm I asked the engines two kinds of question. A brand question names the firm and asks whether it is any good. A buying question asks who to hire, with no names. The table shows how often the engines cite the firm's own website on its brand question, and how often the firm turns up at all when a buyer asks who to hire. Engines are never added together. Google's surfaces are never blended.
| Engine | Category | Own site cited on brand question | Positive verdict on brand question | Named on buying questions | Own site cited on buying questions | First pick on buying questions |
|---|---|---|---|---|---|---|
| Google AI Mode | Accountants | 100%95% to 100%, n=74 | 68%56% to 77%, n=74 | 18%14% to 22%, n=360 | 15%12% to 19%, n=360 | 7%4% to 10%, n=360 |
| Google AI Mode | Financial advisers | 82%72% to 89%, n=72 | 81%70% to 88%, n=72 | 13%10% to 17%, n=372 | 9%6% to 12%, n=372 | 7%5% to 10%, n=372 |
| Google AI Mode | Solicitors | 85%75% to 91%, n=72 | 47%36% to 59%, n=72 | 31%26% to 36%, n=360 | 20%16% to 25%, n=360 | 8%6% to 12%, n=360 |
| ChatGPT | Accountants | 81%71% to 88%, n=84 | 40%31% to 51%, n=84 | 17%14% to 21%, n=420 | 7%5% to 10%, n=420 | 6%4% to 9%, n=420 |
| ChatGPT | Financial advisers | 73%62% to 81%, n=84 | 57%46% to 67%, n=84 | 6%4% to 8%, n=420 | 3%2% to 6%, n=420 | 4%2% to 6%, n=420 |
| ChatGPT | Solicitors | 70%60% to 79%, n=84 | 45%35% to 56%, n=84 | 35%31% to 40%, n=420 | 10%8% to 14%, n=420 | 7%5% to 10%, n=420 |
| Perplexity | Accountants | 92%84% to 96%, n=84 | 63%52% to 73%, n=84 | 23%20% to 28%, n=420 | 44%40% to 49%, n=420 | 7%5% to 10%, n=420 |
| Perplexity | Financial advisers | 92%84% to 96%, n=84 | 54%43% to 64%, n=84 | 6%4% to 9%, n=420 | 15%12% to 19%, n=420 | 5%3% to 7%, n=420 |
| Perplexity | Solicitors | 100%96% to 100%, n=84 | 58%48% to 68%, n=84 | 20%16% to 24%, n=420 | 29%25% to 34%, n=420 | 6%4% to 8%, n=420 |
Rates carry a 95% Wilson interval and the number of firm-answers scored. A buying-question rate is per firm: 12 firms scored on every answer.
Who the engines actually name
Buying questions only. Gold is how often the engine named the firm when a buyer asked who to hire. The navy marker is how often the same engine gave the firm a positive verdict when asked about it by name. The space between them is the work.
Named on buying questionsPositive verdict when asked by nameFirst pick = the engine led with this firm
| # | Firm | ChatGPT named | First pick | Google AI Mode named | First pick | Perplexity named | First pick |
|---|---|---|---|---|---|---|---|
| 1 | Armstrong Watson | 31%11 of 35 answers | 4 | 37%11 of 30 answers | 0 | 74%26 of 35 answers | 15 |
| 2 | UNW | 37%13 of 35 answers | 9 | 30%9 of 30 answers | 3 | 57%20 of 35 answers | 10 |
| 3 | RMT Accountants & Business Advisors | 23%8 of 35 answers | 5 | 23%7 of 30 answers | 6 | 49%17 of 35 answers | 0 |
| 4 | Ryecroft Glenton | 37%13 of 35 answers | 5 | 47%14 of 30 answers | 7 | 6%2 of 35 answers | 0 |
| 5 | Robson Laidler Accountants | 20%7 of 35 answers | 1 | 33%10 of 30 answers | 6 | 29%10 of 35 answers | 1 |
| 6 | Johnston Carmichael | 6%2 of 35 answers | 0 | 33%10 of 30 answers | 1 | 34%12 of 35 answers | 2 |
| 7 | Stephenson Coates | 31%11 of 35 answers | 1 | 3%1 of 30 answers | 0 | 6%2 of 35 answers | 0 |
| 8 | Azets | 9%3 of 35 answers | 0 | 7%2 of 30 answers | 1 | 20%7 of 35 answers | 0 |
| 9 | Acumen Accounting | 9%3 of 35 answers | 0 | 3%1 of 30 answers | 0 | 0%0 of 35 answers | 0 |
| 10 | Tennick Accountants | 0%0 of 35 answers | 0 | 0%0 of 30 answers | 0 | 6%2 of 35 answers | 0 |
| 11 | Jaccountancy | 0%0 of 35 answers | 0 | 0%0 of 30 answers | 0 | 0%0 of 35 answers | 0 |
| 12 | Vertax Accountants | 0%0 of 35 answers | 0 | 0%0 of 30 answers | 0 | 0%0 of 35 answers | 0 |
| # | Firm | ChatGPT named | First pick | Google AI Mode named | First pick | Perplexity named | First pick |
|---|---|---|---|---|---|---|---|
| 1 | Muckle LLP | 77%27 of 35 answers | 19 | 87%26 of 30 answers | 18 | 40%14 of 35 answers | 13 |
| 2 | Ward Hadaway | 60%21 of 35 answers | 0 | 67%20 of 30 answers | 5 | 20%7 of 35 answers | 0 |
| 3 | Sintons | 46%16 of 35 answers | 1 | 63%19 of 30 answers | 1 | 23%8 of 35 answers | 0 |
| 4 | Womble Bond Dickinson | 66%23 of 35 answers | 4 | 47%14 of 30 answers | 3 | 11%4 of 35 answers | 0 |
| 5 | Mincoffs Solicitors | 57%20 of 35 answers | 2 | 37%11 of 30 answers | 2 | 26%9 of 35 answers | 2 |
| 6 | Hay & Kilner | 34%12 of 35 answers | 4 | 40%12 of 30 answers | 1 | 40%14 of 35 answers | 3 |
| 7 | Clarke Mairs | 31%11 of 35 answers | 1 | 13%4 of 30 answers | 0 | 31%11 of 35 answers | 5 |
| 8 | Burnetts Solicitors | 34%12 of 35 answers | 0 | 0%0 of 30 answers | 0 | 20%7 of 35 answers | 1 |
| 9 | Square One Law | 0%0 of 35 answers | 0 | 10%3 of 30 answers | 0 | 26%9 of 35 answers | 0 |
| 10 | JLF Law Firm | 6%2 of 35 answers | 0 | 10%3 of 30 answers | 0 | 0%0 of 35 answers | 0 |
| 11 | Gibson & Co | 11%4 of 35 answers | 0 | 0%0 of 30 answers | 0 | 0%0 of 35 answers | 0 |
| 12 | David Gray Solicitors | 0%0 of 35 answers | 0 | 0%0 of 30 answers | 0 | 0%0 of 35 answers | 0 |
| # | Firm | ChatGPT named | First pick | Google AI Mode named | First pick | Perplexity named | First pick |
|---|---|---|---|---|---|---|---|
| 1 | Robson Laidler Wealth | 0%0 of 35 answers | 0 | 39%12 of 31 answers | 11 | 40%14 of 35 answers | 14 |
| 2 | Newcastle Financial Advisers | 40%14 of 35 answers | 11 | 19%6 of 31 answers | 5 | 14%5 of 35 answers | 5 |
| 3 | St. James's Place Newcastle | 0%0 of 35 answers | 0 | 26%8 of 31 answers | 1 | 3%1 of 35 answers | 0 |
| 4 | Broad Chare Partnership | 0%0 of 35 answers | 0 | 23%7 of 31 answers | 1 | 6%2 of 35 answers | 1 |
| 5 | Grey Street Wealth | 14%5 of 35 answers | 3 | 6%2 of 31 answers | 1 | 0%0 of 35 answers | 0 |
| 6 | Rathbones | 0%0 of 35 answers | 0 | 19%6 of 31 answers | 2 | 3%1 of 35 answers | 0 |
| 7 | Rutherford Hughes | 9%3 of 35 answers | 3 | 10%3 of 31 answers | 2 | 0%0 of 35 answers | 0 |
| 8 | Saltus | 3%1 of 35 answers | 0 | 3%1 of 31 answers | 0 | 6%2 of 35 answers | 0 |
| 9 | Shackleton | 0%0 of 35 answers | 0 | 10%3 of 31 answers | 2 | 0%0 of 35 answers | 0 |
| 10 | Newcastle Financial Planning | 0%0 of 35 answers | 0 | 0%0 of 31 answers | 0 | 3%1 of 35 answers | 0 |
| 11 | Co-Navigate | 0%0 of 35 answers | 0 | 0%0 of 31 answers | 0 | 0%0 of 35 answers | 0 |
| 12 | Raymond James Monument | 0%0 of 35 answers | 0 | 0%0 of 31 answers | 0 | 0%0 of 35 answers | 0 |
The sources behind the answers
Firms the engines named that are not in this benchmark
The sample was frozen before the first run and not changed. These names came up on buying questions anyway. They go into the next quarter's sample.
- HJ Wealth Management: named in 29 answers
- Nicola Tempest-Hall: named in 17 answers
- S&W Group: named in 14 answers
- Jobling Wealth Management: named in 11 answers
- Weightmans: named in 11 answers
- Straker Paul Financial Planning: named in 10 answers
- Michael Westwood Wealth Management: named in 10 answers
- Sweeney Miller Law: named in 9 answers
- TT Law: named in 9 answers
- Mission Financial Planning: named in 9 answers
Domains cited most on buying questions
Count of answers citing each domain, all engines, all runs. Directories and rankings sit alongside the firms' own sites.
| Domain | Answers citing it |
|---|---|
| robson-laidler.co.uk | 61 |
| armstrongwatson.co.uk | 56 |
| legal500.com | 54 |
| hay-kilner.co.uk | 53 |
| unbiased.co.uk | 51 |
| r-m-t.co.uk | 43 |
| unw.co.uk | 38 |
| google.com | 37 |
| linkedin.com | 37 |
| clarkemairs.com | 35 |
| newcastle.co.uk | 33 |
| johnstoncarmichael.com | 32 |
| sintons.co.uk | 32 |
| muckle-llp.com | 31 |
| vouchedfor.co.uk | 31 |
Read before quoting a number
Perplexity reads your site and does not say your name
In 203 firm-answers on buying questions, Perplexity cited a firm's own website in its sources but never wrote the firm's name in the answer. The site was retrieved. The brand was not surfaced. Mention rate and citation rate are therefore reported separately.
Google AI Mode went quiet for a day
On 17 September, 48 of Google AI Mode's 51 calls returned no AI answer at all. The other six days returned answers for every question. Those 48 calls are recorded as no-answer, not as zero, and Google AI Mode's rates use the answers it actually gave.
How this was measured
Sample
12 firms in each of three categories: accountants, solicitors, financial advisers. Selected from the Google Maps results for the category in Newcastle upon Tyne, ranked by review count, plus any firm the engines named unprompted in a calibration run. Frozen as version 1 before the first paid run.
Questions
Five buying questions per category (shortlist, decision, price, problem, recommendation) and one brand question per firm. 51 questions, frozen as version 1, never edited inside the window.
Engines and conditions
ChatGPT (gpt-5.5, web search on, location Newcastle upon Tyne, GB), Perplexity (sonar-pro, GB) and Google AI Mode (Newcastle upon Tyne, English), all through the DataForSEO API. API calls carry no account, no memory and no history, which is the logged-out stranger by construction. Seven runs, one a day at 15:20 UK time, 16 to 22 September 2026.
Scoring
A script scored each answer against the frozen firm list. Named: the firm's name or a known variant appears in the answer. Cited: a URL on the firm's domain appears in the sources or the answer. First pick: the firm is the first of the sample named on a buying question. Positive verdict: on a brand question, the answer names the firm with a positive verdict phrase and no negative one.
Statistics
Think of the interval as the wobble. Ask the same engine the same question tomorrow and the number moves. Seven runs shrink the wobble enough to trust the shape, not the decimal. All the rates carry a 95% Wilson interval and their denominator. Engines are reported separately and never averaged into one score.
Cost and reproducibility
The seven-day collection cost $40.70 in API fees (ChatGPT $35.11, Perplexity $4.16, Google AI Mode $1.43). Raw answers, the frozen question list, the firm list and the scoring script are kept so that every figure on this page can be regenerated.
What this study can and cannot claim
It measures answers from the engines' APIs at fixed settings. A person in the ChatGPT or Perplexity app may see a different model or search depth. A 15-answer hand-verification sample was run in the logged-out consumer apps on 30 September 2026. The app's first pick (or no pick) appeared in the API's seven runs for the same prompt in 11 of 15 answers: 73% (95% interval 48% to 89%). Against a single API run the match was 8 of 15: brand questions agreed 5 of 6 times, buying questions 3 of 9, because the first pick moves between runs in the apps and the API alike.
It is a seven-day snapshot. Named is not recommended, and cited is not named. Citation volumes on these platforms have swung by more than 80% around model updates this year, so a re-run in a different week can move every firm at once. The three measures are kept apart because each one needs a different fix.
Firms are named exactly as they trade. Any firm listed can request the raw answers for its own name and a free re-check.
At the end of the day, being cited is not being chosen. That gap is the whole finding, and I would not trust any score that hides it.
Run the same benchmark on your own firm
Same 51 questions, same three engines, same scoring. Black And White Boy SEO, the owner-operated SEO and AI-search consultancy founded by Paul Clarke in 2017, serving established businesses in Newcastle upon Tyne and the North East, ran this study and will run it for you.
Ask for your firm's numbers
