Research
Which banks AI names in the UK
People ask ChatGPT, Gemini and Perplexity which banks to use. We ask the same 25 questions every day, plain and as 3 types of buyer, and count who gets named.
Live4 days of answers, 29 Oct to 1 Nov 2026
- 1Santander49.0%
- 2HSBC44.5%
- 3Nationwide39.0%
- 4Barclays26.9%
- 5Halifax25.3%
- Questions asked
- 25the same ones, every day
- Answers read
- 975across 3 AI tools
- Source links
- 8,809pages the answers pointed to
- Sites linked to
- 4481,916 separate pages
The main players
Pick one to filter everything below to that brand.
Scoreboard
Who wins each topic
A topic is a group of related questions. Here is how the banks compare on each one.
Santander leads 4 of 7 topics on visibility, more than anyone else. The rest are split between 3 others.
savings account
4 prompts
5 more companies named on this topic.
mortgage
4 prompts
5 more companies named on this topic.
personal loans
4 prompts
6 more companies named on this topic.
debt consolidation loan
4 prompts
6 more companies named on this topic.
best bank in uk
3 prompts
5 more companies named on this topic.
credit cards
3 prompts
5 more companies named on this topic.
fixed rate bonds
3 prompts
3 more companies named on this topic.
How often each company is named in that topic’s answers. A dash means AI never named that company on that topic, which is not the same as a zero.
The numbers
How often AI names each brand
Counted from every answer this report has collected.
A brand counts once for every answer that names it, however many times that answer repeats the name. Visibility is that as a percentage, mentions is the plain count, and citations are the links AI gave as its sources.
Every brand is on the charts. Pick one above, or click a bar, to follow it through all three.
Visibility over time
Share of AI answers that mention each brand (average).
Mentions
Number of AI answers that name each brand.
Total Mentions, all 9 brands
2,574
Citations
Sources AI answers cite in support of each brand, from any website, not only the brand's own.
Total Citations, 9 brands with citations
1,600
Sentiment
What AI says about each brand
The same answers, read for tone. Is a brand talked about well, neutrally or badly, and on which topics.
How this is measured
- 1
Find the mentions
From every answer, we keep only the sentences that name the brand.
- 2
Score the tone
Each of those sentences is scored positive, neutral or negative.
- 3
Average them
All the mentions average into one score out of 10.
ScalePositive7-10Neutral4-7Negative0-4
Most mentions are neutral, so most brands sit near the middle. A small gap between two scores is not a big difference.
All banks, weighted by mentions
6.7 / 10
Score zonesPositive7-10Neutral4-7Negative0-4
Averaged across 4 days of answers.
debt consolidation loan7.1 /10
27.5% of mentionsmortgage6.3 /10
20.9% of mentionspersonal loans6.4 /10
20.3% of mentionsbest bank in uk7.2 /10
16.3% of mentionssavings account6.6 /10
8.8% of mentionscredit cards6.6 /10
4.4% of mentionsfixed rate bonds6.4 /10
1.7% of mentions
Positive29%Neutral71%Negative0%7 topicsA bigger tile means more of mentions across all banks. A deeper colour means better sentiment.
Personas
Who’s asking changes the answer
We ask the same questions as 3 different types of person. The banks that come back are not the same ones.
A persona is the person we ask as, written into the question itself, so AI answers someone in their twenties in a city differently from someone in their fifties in a village. Every question is also asked plain, with nobody attached: that is the Base answer each persona is compared against.
Every question is asked 4 ways
- PlainNobody attached. This is the Base answer.
- Settled urban family35-44 years oldTownCouple with childrenWorks full time
- Pragmatic rural mature45-54 years oldRuralMulti-generational homeSelf-employed
- Young urban digital25-34 years oldCityCouple, no childrenWorks full time
Same question, same day, same three AI tools.
Who each persona hears about
The banks named most often in each persona’s answers, and how that compares to Base.
Settled urban family
35-44 years old · Town · Couple with children · Works full time
- 1Santander50.7%▼ -1
- 2HSBC48.4%▲ +4.1
- 3Nationwide45.8%▲ +9.8
Climbs the most here: Nationwide, up 9.8 points on Base, still 3rd.
Pragmatic rural mature
45-54 years old · Rural · Multi-generational home · Self-employed
- 1Santander48.0%▼ -3.7
- 2HSBC41.8%▼ -2.5
- 3Nationwide40.4%▲ +4.4
Climbs the most here: Lloyds Bank, up 4.7 points on Base, 7th to 4th.
Young urban digital
25-34 years old · City · Couple, no children · Works full time
- 1Santander44.9%▼ -6.8
- 2HSBC43.6%▼ -0.8
- 3Nationwide34.7%▼ -1.3
Climbs the most here: First Direct, up 2.7 points on Base, 6th to 4th.
Ranked by how often each brand is named. The number beside it is the change against the same brand’s Base answer, in percentage points.
Sources
Where AI gets its information about banks
AI answers link to the pages they used. Here is what those pages are, and who owns them.
What a source is
- A mention is not a source
- When AI answers, it names brands and it links to pages. The names are mentions. The pages are the sources it read in order to answer.
- Links, pages and sites
- One link inside one answer counts once, so a page linked in fifty answers counts fifty times. A page is one web address. A site is where that page lives.
- The part you can change
- You cannot edit the AI. You can edit your own pages, and you can work to get mentioned on the pages you do not own.
This report collected 8809 source links across 448 sites and 1916 separate pages.
- Brand and product sites
- 38.2%
- Comparison and review
- 27.1%
- Media and blogs
- 27.0%
- Authority and reference
- 4.8%
- Other
- 1.6%
- Marketplaces
- 1.3%
- Community
- <0.1%
Adds up to 100% of the links collected. Navy is the sites the banks own themselves. A marketplace counts as somebody else’s site, even though the listing on it is the brand’s own.
- moneysavingexpert.com14.5%
- which.co.uk9.1%
- moneyfactscompare.co.uk8.1%
- moneysupermarket.com4.6%
- hsbc.co.uk3.4%
- forbes.com2.2%
- finder.com2.1%
- lloydsbank.com2.0%
- nationwide.co.uk1.9%
- mpamag.com1.9%
Concentration
49.8%
of all 8,809 source links go to just 10 sites.
AI links to hundreds of sites here, but almost all of the reading happens on a handful of them.
- moneysavingexpert.com14.5%
- which.co.uk9.1%
- moneyfactscompare.co.uk8.1%
- moneysupermarket.com4.6%
- forbes.com2.2%
- finder.com2.1%
- mpamag.com1.9%
- money.co.uk1.3%
- hoa.org.uk1.1%
- moneyhelper.org.uk0.9%
Places that write about the banks rather than sell for them, so brand sites, shops and marketplaces are all left out. Shares are of every source link in the report, most of which go to the brands’ own sites, so these read small. Bars are scaled to the longest row, not to 100%, so read the numbers rather than the lengths.
Patterns
How AI answers about banks
| AI tool | Brands named | Sources shown |
|---|---|---|
| All AI tools | 7.2 | 7.7 |
| ChatGPT | 7.6 | 5.6 |
| Gemini | 7.0 | 2.7 |
| Perplexity | 6.8 | 14.7 |
- Mentions of these banks with a source
- 46.1%
Brands named counts every brand an answer names, not only the ones we track. Sources shown are the links each AI tool showed with its answer, as we collected them.
Answers
What AI actually said about banks
Every number on this page is counted from answers like these. Open one to read it in full, with the brands it named and the pages it linked to.
9 answers picked at random from the 975 we have collected, not a hand-picked selection. The pick changes daily, and each answer is shown exactly as AI returned it.
Other countries