First reading: what six AI engines named in four UK industries

Do the engines agree on which companies to name?

Not often. For each industry we took every engine's five most-named companies (more when companies tied for fifth) and checked which companies made all six lists.

So a company can sit in one engine's top five and be missing from another's. In online accountants, only Crunch made every list, and the six lists held 11 different companies between them.

Each engine's most-named company

Where an engine's first choice was not the company ranked first overall:

• Car and van leasing: ChatGPT named Arval in 23 of 60 answers and Select Car Leasing, first overall, in 22.
• Home EV chargers: Claude named Wallbox in 41 of 60 answers and Ohme, first overall, in 36.
• Home EV chargers: Google AI Overviews named myenergi in 37 of 52 answers and Ohme, first overall, in 35.
• HR and payroll software: Gemini named Xero in 41 of 60 answers and Sage, first overall, in 35.
• HR and payroll software: Google AI Overviews named Xero in 31 of 56 answers and Sage, first overall, in 24.
• Online accountants: Gemini named Gorilla Accounting in 44 of 60 answers and Crunch, first overall, in 33.
• Online accountants: Perplexity named The Accountancy Partnership in 23 of 60 answers and Crunch, first overall, in 22.

Read Claude's line on chargers with care: 24 of its 60 answers used US products or terms, such as Level 2 or ChargePoint Home, and in the other 36 Ohme (27) comes ahead of Wallbox (22).

Three of these seven are two answers apart or fewer, so read them as a near tie rather than a different winner.

Where do the engines disagree most?

The widest gap between two engines for a single company, one per industry:

• Zenith (Car and van leasing): named in 30 of 60 Gemini answers (50.0%) and 8 of 60 ChatGPT answers (13.3%).
• Wallbox (Home EV chargers): named in 42 of 60 Gemini answers (70.0%) and 12 of 60 Google AI Mode answers (20.0%).
• PayFit (HR and payroll software): named in 28 of 60 Gemini answers (46.7%) and 3 of 60 ChatGPT answers (5.0%).
• Gorilla Accounting (Online accountants): named in 44 of 60 Gemini answers (73.3%) and 11 of 60 Claude answers (18.3%).

Gemini was the high engine in all four. It also names more of the companies we track per answer than any other engine, so a company it names often can look much stronger there than on the others.

For a company watching its own AI visibility, one engine's figure says little about the others.

Which engines name the most companies?

Gemini named 5.8 of the companies we track per answer and left 1.7% of its answers without any of them. Google AI Mode named the fewest, 3.1. ChatGPT was the most likely to name none of them: 19.2% of its answers.

These counts cover only the companies on our lists, 15 to 25 per industry. An answer that names other firms, or gives general advice without naming anyone, counts as naming none.

Which industry is hardest to be named in?

Online accountants: 16.3% of its answers named none of the 25 companies we track, the highest of the four, and the average company on the list was named in 11.8% of answers, against 17.5% to 26.6% elsewhere.

In online accountants, UK rules questions were the emptiest: 33.3% of those answers named none of them.

Lists differ in size (15 in home EV chargers, 25 in the other three), and a longer list lowers the average, so the share of answers naming none is the fairer comparison.

Which questions name the most companies?

In all four industries, questions asking an engine to compare options named more of the companies we track per answer than any other kind of question.

Figures are tracked companies named per answer, across every engine.

Which sites do the engines cite?

We counted how many answers cited each site, then checked which of each industry's 15 most-cited sites belong to a company we track.

In car and van leasing the engines leaned most on companies' own sites: 10 of the 15 most-cited sites belong to companies we track. The top two were the two most-named companies' own sites, selectcarleasing.co.uk (88 answers) and nationwidevehiclecontracts.co.uk (78 answers).

In home EV chargers they leaned least: 2 of 15. The most-cited sites there were google.com, reddit.com and bestchargers.co.uk.

The google.com links are almost all Google Shopping product listings inside Google's own AI answers.

Does the leader lead for every kind of buyer?

Each industry's questions are split by the kind of buyer asking. This is the most-named company in each segment, next to the overall leader's figure there.

Select Car Leasing leads car and van leasing overall, but Octopus EV led the EV and salary sacrifice segment (69.7%). Select Car Leasing was at 37.0% in that segment.

Crunch leads online accountants overall, but Provestor led the landlords segment (41.1%). Crunch was at 7.8% in that segment.

Segment figures rest on 86 to 203 answers each, fewer than the headline figures, so small differences between segments may be noise.

Which leads are clear?

• Car and van leasing: Select Car Leasing (51.0% ±8.6) leads Nationwide Vehicle Contracts (35.6% ±7.0) by 15.4 points, more than the two margins of error combined (±11.1).
• Home EV chargers: Ohme (67.0% ±6.2) and myenergi (62.8% ±7.3) are 4.2 points apart, within the two margins of error combined (±9.6): too close to call on one reading.
• HR and payroll software: Sage (60.7% ±8.0) and Xero (59.3% ±9.2) are 1.4 points apart, within the two margins of error combined (±12.2): too close to call on one reading.
• Online accountants: Crunch (46.6% ±8.3) leads Gorilla Accounting (33.7% ±6.0) by 12.9 points, more than the two margins of error combined (±10.2).

The margin of error is the range the figure would land in 95 times in 100 if we had asked a different mix of similar questions. A gap between two companies is less certain than either figure, so we call a lead clear only when it is wider than the two margins combined (the square root of the sum of their squares).

What one reading cannot tell you

• Which way anything is moving. Changes start with the next reading, and a change counts as a move only when it falls outside the margin of error.
• Fine differences. The leaders' margins of error run from ±6.2 to ±8.6 points, and a figure for a single engine rests on 52 to 60 answers.
• Companies we do not track. We count 90 companies across the four lists, and an engine can name others.
• A second asking. Each question went to each engine once this week.
• Claude's searches. In well over half of its 240 answers on 3 November 2026, Claude said its web search had stopped or failed, or that it was working partly from its own knowledge. Leaving those answers out keeps the same three companies on top in every industry and moves no company's overall figure by more than about 4 points, but read Claude's own column with more care than the others.

How we measured

We asked 240 questions, 60 per industry, once per engine, between 10:57 and 14:40 UTC on 3 November 2026. Of 1,440 attempts, 4 failed and 15 Google searches showed no AI answer, leaving 1,421 answers. The full method, with the question lists and how the margin of error is worked out, is on how we measure.

The rankings

• Car and van leasing
• Home EV chargers
• HR and payroll software
• Online accountants