UK banking employer brands are close to invisible in AI candidate research

A candidate weighing up a banking career in 2026 doesn’t open your careers site any more. On a Sunday evening she asks an AI assistant which UK banks are actually worth working for, and reads the short answer it gives back.

The assistant names three employers in a couple of sentences. If yours is a UK bank, the odds are good that it isn’t one of them.

No job board fee was undercut. No advert was outbid. The assistant simply couldn’t find enough credible, machine-readable evidence to recommend you with confidence, so it left you out. To see how often that happens across the sector, I scored the 9 UK banking employers in The SetpointHQ Index, our monthly study of how employer brands surface across ChatGPT, Claude, Perplexity and Gemini. The figures below are from the 1 September 2026 run, compared against the first banking reading on 13 July. Both runs use the same methodology, so the movements are real movements, not a change in how we measure. If the difference between ranking and being cited is new to you, start with SEO versus GEO.

One leader, and a long tail

The shape of the field is stark. The median banking employer scores 55.0 out of 100, up from 53.0 in July, and is named in 6 of the 20 prompts candidates use to research banking careers. Seven times out of ten, the average bank here does not come up at all.

Above that median sits one clear leader. Barclays is rank 2 in the entire 100-brand Index, at 74.3, behind only BCG in professional services. The next banking brand, Lloyds Banking Group, is all the way down at rank 42. After that a long tail runs to Nationwide Building Society at rank 98 of 100. An entire major employing sector, represented near the top of the Index by a single name.

Scores out of 100, The SetpointHQ Index, 1 September 2026 run. “Cited” is how many of 20 candidate research prompts named the employer. “Citation” is the weighted citation dimension score. Each employer name links to its full Index profile.

The gap isn’t recognition. It’s machine-readability.

Here is the part that should change how a talent team thinks about this. The AI assistants already know these banks exist. On the knowledge-graph measure, whether the model understands the company as a distinct entity, UK banks score between 77.3 and 100 across the board. Recognition is not the problem.

The problem sits one layer down, in the evidence a model needs before it will name you in an answer. The Index scores five of those signals:

• Citation: how often you actually surface when candidates ask the questions they use to pick an employer
• Schema: the structured data on your careers and job pages that tells a machine what you are and what roles you have open
• Knowledge graph: whether your organisation exists accurately in Wikidata and the graphs assistants lean on
• Crawlability: whether AI crawlers can actually reach your careers content
• Social: the wider signal trail that supports the story your careers site tells

Schema is where banking falls down hardest. Four of the nine score 25 or below on it, and TSB and Nationwide score zero. Their careers pages are live, and in TSB’s case the crawlers reach them without trouble. They simply don’t carry the markup that lets an assistant cite them with any confidence. That chain, from crawled to read to cited, is where the sector breaks.

The bank that proved it can be fixed

In July, Standard Chartered scored 5 on schema and sat 7th of the 9 banks, at 48.0. By September its schema score was 45, its composite was 56.4, and it had climbed to 3rd in the sector and from rank 66 to rank 44 in the whole Index. Nothing else on its scorecard moved by more than a couple of points.

That is the clearest live evidence in the Index that the schema gap is fixable, and that fixing it moves the score within weeks. Standard Chartered’s citation count is still low, at 5 of 20, so the job isn’t finished. But no other bank moved as far between the two runs, and it was the one that fixed the machine-readable layer.

The name everyone knows, and the AI barely mentions

The clearest example of the opposite is Nationwide. One of the most recognised names in British retail finance, a household brand for generations, sits at rank 98 of 100 for AI candidate-research visibility, last of the nine banks. It is also the only bank that fell between the two runs, from 38.0 to 35.5, as its crawlability dropped from 48.7 to 40.7. It scores zero on schema. Being known by people is not the same as being legible to machines, and a candidate asking an assistant about building societies to work for may never see it named.

Schema alone doesn’t buy you the answer

It would be neat to say the schema score explains everything. Our own data says otherwise, and it is worth being straight about it.

Metro Bank has the second-strongest schema score in the sector at 70, behind only Barclays. It is also among the three least cited of the nine, named in 3 of 20 prompts. Good markup, and the assistants still rarely name it. Across the nine banks, schema and citation correlate at roughly 0.4: a real relationship, but a long way from the whole story.

HSBC makes the same point from the other side. It is named in 11 of 20 prompts, level with Barclays, yet it sits at rank 50 because its schema scores 30 and its crawlability 60. The assistants already talk about HSBC. Its careers site gives them little to quote when they do.

The honest reading is that schema is necessary rather than sufficient. It is the thing that lets an assistant understand and quote your page. What earns you the mention on top of that is the wider evidence trail: third-party citations, review patterns, coverage, and content that actually answers what candidates ask. Fixing the markup gets you into the running. It doesn’t win the race on its own.

Even the leader has headroom

It would be easy to read Barclays near the top as a job done. It isn’t. Barclays is near perfect on the dimensions that prove an AI knows who you are, 100 on crawlability and 98.8 on knowledge graph, and yet it is named in 11 of 20 relevant prompts, down from 13 in July. The most AI-visible banking employer in Britain still misses close to half the moments where candidates are researching banking careers through AI. If the brand at the top of the field has that much room to grow, the field below it has far more.

One finding per brand

• Barclays (74.3, rank 2). The only bank in the top 40. Strongest schema in the sector at 75, yet named in just over half of candidate prompts.
• Lloyds Banking Group (56.6, rank 42). Perfect crawlability and a 96.4 knowledge graph score, held back by schema at 25.
• Standard Chartered (56.4, rank 44). Schema up from 5 to 45 since July, lifting it from 7th to 3rd in the sector.
• NatWest (55.6, rank 47). Named in 7 of 20 prompts, up from 5 in July. Crawlability at 68 is its weakest technical score.
• HSBC (55.0, rank 50). Named as often as Barclays, 11 of 20, but schema at 30 and crawlability at 60 keep it mid-table.
• Metro Bank (54.6, rank 52). Second-best schema in the sector at 70, and among the least cited at 3 of 20.
• Santander UK (50.4, rank 62). The lowest knowledge graph score of the nine at 77.3. Citation up from 21.3 to 27.5 since July.
• TSB Bank (43.2, rank 83). Perfect crawlability and zero schema: the crawlers reach every page and find nothing to quote.
• Nationwide Building Society (35.5, rank 98). The only bank to fall since July, as crawlability dropped to 40.7. Zero schema.

The fixable part

The encouraging news is that the biggest banking gap is also the cheapest to close. Schema is a technical layer, not a brand rebuild. An employer that adds proper Organisation and JobPosting structured data moves from an assistant being unsure what the page is, to an assistant being able to name it. Standard Chartered has just shown it can be done inside two months. The recognition is already there. The machine-readable layer underneath is what most of these banks are missing, and it is fixable in weeks, not quarters. Generative engine optimisation for recruitment covers what sits on top of it.

How the Index measures this

The scores come from The SetpointHQ Index, 100 UK employer brands scored across the five dimensions above by probing four AI assistants with the questions candidates actually ask. The Index runs monthly on a published, versioned methodology. Banking is one sector within it. We have published the same analysis for social care and retail and consumer goods.

I’m Andy Pirie, a recruitment marketer. I’ve spent 20 years on employer brand and candidate acquisition. SetpointHQ measures how AI assistants see employer brands, so talent teams can see the gap before it costs them applications.

If you want a clear read before your next hiring campaign, that is what the £999 Recruitment Marketing Audit is for. It scores your whole hiring system, including how visible and credible you are to AI, and hands you a prioritised 90-day plan within 7 working days. If you need the scoreboard running continuously, talk to us about Pro, our longitudinal AI-visibility tracking for talent and employer-brand teams.

Frequently asked questions

How visible are UK banks to AI candidate research?

Across the 9 UK banking employers in The SetpointHQ Index, the median composite score is 55.0 out of 100 as of the 1 September 2026 run, and the median bank is named in 6 of the 20 prompts candidates use to research banking careers. One brand, Barclays, sits at rank 2 in the 100-brand Index. The rest sit at rank 42 or below.

Why do well-known banks score low for AI visibility?

Recognition and machine-readability are different things. On the knowledge-graph measure, which reflects whether an AI understands a company exists, UK banks score between 77.3 and 100. The gap is structured data. Four of the nine score 25 or below on schema, the markup that tells an AI what a company is and what roles it has open, and two score zero. The pages are live, but they are not legible to a machine.

Is fixing schema enough to get cited by AI assistants?

No. Schema is necessary rather than sufficient. In the banking data, schema and citation correlate at roughly 0.4. Metro Bank has the second-strongest schema score in the sector at 70 and is still among the least cited banks, named in 3 of 20 prompts. Structured data lets an assistant understand and quote your page; the wider evidence trail of citations, reviews and content that answers real candidate questions is what earns the mention.

Does fixing schema actually move an employer’s AI visibility score?

Yes. Standard Chartered raised its schema score from 5 to 45 between the July and September 2026 Index runs. Its composite rose from 48.0 to 56.4, lifting it from 7th to 3rd of the 9 UK banks and from rank 66 to rank 44 in the 100-brand Index.

What is the most fixable AI-visibility gap for an employer?

Schema is the cheapest gap to close on the whole scorecard. It is a technical layer, not a brand rebuild. An employer that adds proper Organisation and JobPosting structured data moves from an AI being unsure what the page is, to an AI being able to cite it by name.

How does The SetpointHQ Index measure employer brands?

The SetpointHQ Index scores 100 UK employer brands across five dimensions, citation, schema, knowledge graph, crawlability and social, by probing four AI assistants with the questions candidates actually ask. It runs monthly on a published, versioned methodology. Banking is one sector within that Index.

How do I find out where my employer brand ranks?

The £999 Recruitment Marketing Audit scores your whole hiring system, including how visible and credible you are to AI, and returns a prioritised 90-day plan within 7 working days. If you want the scoreboard running continuously, you can talk to us about Pro, our longitudinal AI-visibility tracking.

The answer about banking is being written now

The candidates have already moved. The assistants are already answering. Every month the Index runs, the answer about which banks are worth working for gets a little more settled, and a little harder to rewrite. The only question is whether it names you. If you want the answer for your own brand rather than the sector, that is what the AI Visibility Audit measures: what each of the four assistants says about you, scored and evidenced.