What Does ChatGPT Say About Working at Your Company?
What Does ChatGPT Say About Working at Your Company?
Go on. Ask it.
Open ChatGPT right now and type: "What's it like working at [your company]?"
If you haven't done this before, prepare to be surprised. The answer AI gives candidates about your company is probably not what you'd write yourself — and for 68% of UK employers, it's either wrong, outdated, or embarrassingly thin.
This matters because candidates are doing exactly this, right now, at scale. ChatGPT has 800 million weekly active users. Perplexity is the fastest-growing search product in history. Google's AI Overviews appear on most employment-related queries. The first impression your company makes is no longer your careers page — it's an AI-generated summary that you didn't write and probably haven't seen.
We audited what AI says about 500+ UK employers. This article explains what we found, why it matters, and exactly how to fix it.
The Experiment: What AI Actually Says
We asked four AI platforms the same question about hundreds of UK employers: "What's it like working at [company]? What do they pay, what's the culture, and would you recommend applying?"
The responses fell into three categories.
Category 1: The Lucky Few (12% of employers)
AI gives a rich, accurate, specific answer:
"Revolut offers software engineers between £85,000 and £120,000 base salary depending on seniority, plus equity. The culture is described as intense and fast-paced, with a focus on ownership and shipping quickly. They operate hybrid from their Canary Wharf office, typically 2–3 days per week. Interview process is 4 stages: recruiter screen, technical assessment, system design interview, and culture fit. Glassdoor rating is 3.6/5 with reviews noting high growth opportunities but demanding expectations."
These employers have strong AI visibility because they publish structured data, get cited in authoritative sources, and have enough public information for AI to synthesise a credible narrative.
Category 2: The Glassdoor Hostages (20% of employers)
AI gives an answer, but it's almost entirely sourced from old Glassdoor reviews:
"Reviews of [Company] suggest a mixed work environment. Some employees praise the collaborative culture and interesting projects, while others mention concerns about work-life balance and management. The company has a 3.2 rating on Glassdoor based on 47 reviews."
These employers have ceded narrative control to third-party review sites. AI treats 3-year-old Glassdoor reviews as current facts. The company may have transformed its culture since then — but AI doesn't know that.
Category 3: The Invisible Majority (68% of employers)
AI essentially guesses:
"I don't have detailed information about working at [Company]. Based on limited data, they appear to be a mid-sized technology company based in the UK. I'd recommend checking their careers page or Glassdoor for more specific information about culture and compensation."
For the majority of UK employers, AI simply has nothing to work with. No structured data, no authoritative citations, no content that answers common candidate questions. The result is a non-answer that sends candidates to your competitors.
Why This Matters More Than You Think
Candidates Trust AI More Than Your Careers Page
Your careers page says "We're a dynamic, innovative team that values collaboration." Every careers page says that. Candidates know it's marketing.
When AI gives the same vague description, candidates interpret it differently — they think AI is being objective and simply can't find anything more specific. The conclusion isn't "this company is dynamic" — it's "this company doesn't have enough of a reputation for AI to know about."
That's a much worse signal.
The 60-Second Decision Window
The old candidate research journey took 20–30 minutes across 5+ touchpoints. The new journey takes under 60 seconds. One question, one AI response, one decision.
If AI's answer is thin, inaccurate, or unflattering, the candidate moves on. There's no second touchpoint. They never visit your careers page, never enter your ATS, never appear in your analytics. You've lost them at a stage you can't even see.
It's Not Just ChatGPT
This isn't a single-platform problem. Candidates get AI-generated employer summaries from:
- ChatGPT — 800M weekly active users
- Google AI Overviews — appearing on most employment queries
- Perplexity — fastest-growing search platform, cites sources explicitly
- Claude — growing adoption, especially in professional contexts
- Meta AI — 1B monthly users across WhatsApp and Instagram
- Microsoft Copilot — integrated into Windows, Edge, and Office
Every AI platform synthesises employer information differently, weights sources differently, and fills gaps differently. If you're not visible across all of them, you're invisible to a growing share of candidates.
The Five Things AI Gets Wrong Most Often
Based on our audit of 500+ UK employers, these are the most common inaccuracies:
1. Salary Data (Wrong for 73% of employers)
AI estimates salaries using a blend of job board data, Glassdoor reports, and general market benchmarks. The result is often £12,000–£25,000 off from what companies actually pay.
For employers who don't publish salary bands, AI defaults to conservative estimates. That means your actual £95,000 senior engineering role gets described as "likely around £70,000–£80,000" — and candidates filter you out before applying.
2. Company Size and Stage (Wrong for 41% of employers)
AI frequently cites outdated headcount or confuses companies with similar names. A 500-person scale-up gets described as "a small startup" because AI's training data is from when they were 30 people. An enterprise company gets confused with a different entity entirely.
3. Remote Work Policy (Missing for 64% of employers)
When AI can't confirm your remote policy, it hedges: "The company's remote work arrangements aren't publicly detailed." In 2026, that ambiguity is a deal-breaker for most candidates.
4. Culture Description (Sourced from outdated reviews for 58% of employers)
AI builds culture descriptions primarily from Glassdoor, Indeed, and Blind reviews. If your most recent reviews are from 2023 and mention problems you've since fixed, AI is still telling candidates about those problems today.
5. Interview Process (Missing for 82% of employers)
This is the biggest gap. The vast majority of employers publish nothing about their interview process in a format AI can use. The result is either vague generalities or, worse, AI fabricating a process based on industry norms that doesn't match your actual experience.
How to Check What AI Says About You (5-Minute Audit)
You can do a quick manual check right now:
Step 1: Ask the question
Open ChatGPT, Claude, Perplexity, and Google AI and ask:
- "What's it like working at [your company]?"
- "What does [your company] pay for [your most-hired role]?"
- "What's the interview process at [your company]?"
- "Would you recommend working at [your company]?"
Step 2: Score the responses
For each response, note:
- Accuracy: Is the information correct?
- Completeness: Does it cover salary, culture, benefits, interview process?
- Recency: Is it citing current information or data from years ago?
- Source quality: Is it citing your own content, or third-party reviews?
- Sentiment: Would a candidate reading this want to apply?
Step 3: Get the full picture
For a comprehensive, automated audit across all four major AI platforms — with a citation chain analysis showing exactly which sources AI is pulling from — run a free OpenRole audit. It takes 30 seconds and gives you a scored breakdown with specific fix recommendations.
How to Fix What AI Says About You
The good news: AI visibility is fixable, and the impact is fast. Companies that take action see measurable changes in as little as two weeks.
Fix 1: Publish the answers candidates actually want
AI can only cite information that exists publicly. The seven questions candidates ask most are:
- What do you pay for [role]?
- What benefits do you offer?
- What's the interview process?
- What's the culture like?
- Do you allow remote work?
- What career growth is available?
- What do employees actually say?
Answer all seven on your careers page. Be specific. Use real numbers. Structure them as FAQs — question-and-answer format is the most extractable format for AI models.
Fix 2: Implement structured data (schema markup)
AI models prioritise machine-readable data over prose. Implementing schema markup on your careers page is the single most effective technical change you can make:
- Organization schema on your homepage (company details, size, founding)
- JobPosting schema on every job listing (salary, location, remote policy)
- FAQPage schema on your careers FAQ (culture, process, benefits)
This gives AI authoritative, structured facts instead of forcing it to guess from unstructured text.
Fix 3: Update Glassdoor and review sites
AI heavily weights Glassdoor, Indeed, and Blind. If your reviews are outdated:
- Encourage current employees to leave honest, recent reviews
- Respond to negative reviews with specific changes you've made
- Update your Glassdoor employer profile with current benefits and salary data
You can't control what employees write, but you can ensure the most recent picture is accurate.
Fix 4: Unblock AI crawlers
Check your robots.txt file. Many companies inadvertently block AI crawlers:
# Check for these — they should NOT be blocked:
User-agent: GPTBot # ChatGPT
User-agent: ClaudeBot # Claude
User-agent: PerplexityBot # Perplexity
User-agent: Google-Extended # Google AI
If these are blocked (via Disallow: /), AI literally cannot access your careers content. Unblocking them is a 5-minute change with outsized impact.
Fix 5: Create linkable, citable content
AI models cite content that other sources cite. A single, well-written blog post about your engineering culture — linked from a few industry publications or shared by employees on LinkedIn — can shift your entire AI narrative.
This isn't about volume. It's about creating one definitive piece that becomes the authoritative source AI reaches for when candidates ask about you.
The Companies Getting This Right
Example: The Employer Who Owns the Narrative
When you ask ChatGPT about Monzo, the answer is rich, specific, and largely accurate:
Specific salary bands, detailed engineering blog posts about culture, published interview process, employee testimonials, and a well-maintained Glassdoor presence.
Monzo doesn't have better AI technology. They have better AI inputs. They publish extensively about what it's like to work there, in formats AI can extract, and they've done it consistently enough that AI treats them as an authoritative source.
Example: The Invisible FTSE 250
Meanwhile, companies 10x Monzo's size — household names in the UK — get vague, generic responses from AI. The problem isn't their brand awareness. It's that they publish employer content in formats designed for humans browsing careers pages, not AI models synthesising information.
The gap between these two approaches will define employer competitiveness in 2026 and beyond.
What Happens If You Do Nothing
If you don't take control of your AI employer narrative:
- AI will keep guessing — and guesses trend conservative (lower salaries, vaguer culture, generic descriptions)
- Competitors who act first will get cited — AI builds on existing citations, creating a compounding advantage
- The zero-click candidate population will grow — more candidates will form opinions without ever visiting your site
- Your cost-per-hire will increase — as top candidates self-select out based on AI's incomplete picture
- You won't know it's happening — this is an invisible leak, not a measurable one
The cost of AI misinformation isn't theoretical. Companies with inaccurate AI salary data are losing an estimated £240K per month in candidate attrition they can't see.
Your Next Step
Find out what AI says about your company right now. It takes 30 seconds.
Run your free AI employer brand audit →
You'll see exactly what ChatGPT, Claude, Perplexity, and Google AI tell candidates about your company — and get a scored breakdown of what to fix first.
The candidates are already asking. The only question is whether you've given AI the right answers.
Source: OpenRole audit data from 517 UK employers, March 2026. AI platform usage statistics from OpenAI, Meta, SparkToro, and Pew Research as cited. For full methodology, see the UK AI Employer Visibility Report 2026.