What HR Teams Need to Know About AI Search in 2026

Right now, a candidate is asking ChatGPT what it's like to work at your company. They're not visiting your careers page. They're not reading your carefully crafted EVP statement. They're typing a question into an AI chatbot and making a decision based on whatever comes back.

This isn't a future scenario. It's happening today, at scale, and most HR teams have no idea.

The Shift HR Teams Missed

AI search is the use of artificial intelligence models — ChatGPT, Claude, Perplexity, Google AI Overviews, and Gemini — to research and evaluate information, replacing or supplementing traditional search engines. For recruitment, this means candidates are now using AI as their primary research tool when evaluating potential employers.

The shift happened faster than anyone predicted. While HR teams were optimising job boards and refining careers pages, candidates quietly changed how they research employers entirely.

According to OpenRole's analysis of 500 UK employers, 80% of candidates under 30 now use AI to research employers before applying. That number rises to 87% for tech roles and 73% for professional services. These are not early adopters — this is mainstream behaviour.

What Candidates Actually Ask AI

Understanding what candidates ask is the first step to managing your AI employer brand. Based on OpenRole's analysis of common AI queries about UK employers, the most frequent questions fall into five categories:

1. Culture and Work Environment

• "What's it like to work at [Company]?"
• "Is [Company] a good place to work?"
• "What's the culture like at [Company]?"

These are the broadest and most common queries. AI responses typically synthesise information from Glassdoor reviews, news articles, and whatever structured data exists on your website. If you haven't published culture information in an AI-readable format, the response will rely almost entirely on third-party data — much of which may be outdated.

2. Salary and Compensation

• "What does [Company] pay for a [role]?"
• "What's the salary range at [Company] for senior engineers?"

Our salary accuracy research found that AI salary estimates are accurate to within 10% for only 38% of UK employers. Companies that publish salary ranges on their careers pages see accuracy jump to 91%.

3. Career Progression

• "Can you progress quickly at [Company]?"
• "What are promotion prospects like at [Company]?"

This is where AI responses are most likely to contain hallucinated or fabricated information, because very few companies publish structured career progression data.

4. Comparison Queries

• "Should I work at [Company A] or [Company B]?"
• "Compare [Company] to [Competitor] for graduate roles"

Comparison queries are particularly dangerous because AI directly ranks employers against each other. If your competitor has better AI visibility, they'll be recommended over you — regardless of whether they're actually a better employer.

5. Application and Interview Process

• "How do I apply to [Company]?"
• "What's the interview process like at [Company]?"

Our research into the AI interview preparation problem found that candidates are using AI to prepare for interviews, and the information AI provides is often wrong.

Why Glassdoor and Indeed Data Gets Recycled

Here's the uncomfortable truth about how AI constructs your employer brand: AI models are trained on web-crawled data, and the richest source of employer information on the web is review platforms.

Glassdoor, Indeed, and similar sites have hundreds of thousands of employer reviews. This data is text-rich, opinion-dense, and exactly the kind of content that language models learn from. The problem is threefold:

1. Recency doesn't exist in training data. A scathing Glassdoor review from 2021 carries the same weight as a glowing one from last month. AI models don't timestamp their training data in a way that prioritises recent information. A company that had a difficult period three years ago may still be described negatively by AI, even if things have dramatically improved.

2. Review platforms over-represent negative experiences. People are more likely to leave reviews when they're unhappy. This negativity bias in the training data translates directly into AI's characterisation of employers. OpenRole's analysis found that outdated reviews create persistent AI perception problems for employers.

3. Your own content is harder for AI to access. Many careers pages are built with JavaScript frameworks that AI crawlers can't parse. PDFs are invisible. Videos don't get indexed. Meanwhile, Glassdoor reviews are plain HTML text that's trivially easy for AI to ingest.

The result? AI's version of your employer brand is built primarily from third-party review data, not from the content you've carefully created and published.

What AI Actually Shows Candidates (And What's Missing)

When a candidate asks ChatGPT about your company, the response typically includes:

The pattern is clear: AI has plenty to say about your company, but much of it is incomplete, outdated, or sourced from places you don't control.

The Zero-Click Candidate Problem

Zero-click candidates are job seekers who research, evaluate, and make decisions about employers entirely within AI interfaces, without ever visiting the company's careers page or website.

This is not a minor trend. OpenRole's data suggests that up to 40% of initial employer research now happens without a single click to the employer's website. For HR teams, this means:

• Your careers page analytics undercount interest. Candidates are evaluating you, but your analytics show nothing.
• Your employer branding content may never be seen. The EVP video, the employee testimonials, the benefits breakdown — none of it matters if AI doesn't cite it.
• You can't track or influence the candidate journey. Traditional marketing funnels don't work when the funnel starts and ends inside ChatGPT.

Understanding the death of the careers page as the primary candidate touchpoint is essential for any HR team planning their 2026 strategy.

What HR Teams Can Do Today

You don't need to become a technical expert. You need to take five practical steps, starting this week:

Step 1: Run an AI Employer Brand Audit

Before you can fix anything, you need to know what AI currently says about you. Run a free audit at openrole.co.uk to see exactly how ChatGPT, Claude, and other models describe your company. The audit will show you:

• What AI gets right
• What AI gets wrong
• What information is missing entirely
• How you compare to competitors in your sector

Step 2: Publish Structured Data on Your Careers Page

Structured data (schema markup) is the single most impactful thing you can do for AI visibility. It's a way of tagging information on your website so that AI models can reliably extract it. Key schema types for employers include:

• Organisation schema — company name, description, founding date, employee count
• JobPosting schema — role titles, salary ranges, locations, benefits
• FAQPage schema — common candidate questions and answers

OpenRole provides a free schema markup generator that creates this for you in minutes.

Step 3: Unblock AI Crawlers

Check your robots.txt file. If it blocks GPTBot, ClaudeBot, or Google-Extended, AI models can't access your careers page content. Many companies blocked these crawlers during the 2023–2024 AI panic without realising the recruitment implications.

Step 4: Publish Salary Ranges

Companies that publish salary ranges see a dramatic improvement in AI salary accuracy. If you're not legally required to publish ranges (the UK currently doesn't mandate it), consider publishing ranges anyway — it's one of the strongest signals you can send to AI models.

Step 5: Create an llms.txt File

An llms.txt file is a simple text file that tells AI models what your company wants them to know. Think of it as a structured briefing document for AI. OpenRole offers a free llms.txt generator to create one in minutes.

The Cost of Doing Nothing

Every month you wait, AI models serve thousands of candidate queries about your company using outdated or incorrect information. The cost of AI misinformation compounds over time:

• Candidates self-select out based on wrong salary data or inaccurate culture descriptions
• Your competitor's AI visibility improves while yours stagnates
• Correcting entrenched AI misinformation becomes harder the longer it persists

According to OpenRole's analysis, companies that address AI visibility issues in Q1 2026 will have a significant advantage over those that wait until 2027, because AI models learn and reinforce patterns over time.

How to Make This an HR Priority

Getting budget and attention for AI employer branding requires framing it in terms leadership understands:

1. Talent acquisition cost. If AI is sending candidates to competitors by recommending them more favourably, your cost-per-hire increases.
2. Brand risk. If AI is saying incorrect things about your company, it's a reputational issue that affects more than just recruitment.
3. Competitive intelligence. Your competitors in the UK AI employer visibility rankings may already be ahead of you.

The conversation with your leadership team should be simple: "80% of young candidates now research us via AI. We have no idea what AI says about us, and we're not doing anything to influence it. Here's what we need to do."

Frequently Asked Questions

Q: Do we really need to worry about AI search if we're a well-known employer brand?

A: Yes — in fact, well-known brands often have more AI visibility problems than smaller companies. AI models have more data about you, which means more opportunities for outdated or contradictory information. OpenRole's audit of 500 UK employers found that brand recognition does not correlate with AI accuracy. Companies like Tesco and PwC have significant AI information gaps despite being household names.

Q: How much does it cost to fix AI employer brand issues?

A: The basic steps — adding structured data, publishing an llms.txt file, unblocking AI crawlers — are free or very low cost. They require time from your web team (typically 2–4 hours for initial setup) and ongoing content maintenance. For a comprehensive AI employer brand strategy, see OpenRole's pricing for audit and monitoring tools.

Q: How quickly will changes affect what AI says about us?

A: It depends on the AI model. Google AI Overviews can reflect changes within days, because they pull from live web data. ChatGPT and Claude update their training data periodically — typically every few months — but also use web browsing for real-time queries. Structured data and llms.txt improvements can be picked up within weeks for browsing-enabled models.

Q: Should HR own AI employer branding, or should it sit with marketing?

A: AI employer branding sits at the intersection of HR, marketing, and IT. HR owns the content and strategy (what should AI know about us?), marketing owns the execution (how do we publish and optimise it?), and IT enables the infrastructure (schema markup, crawlers, llms.txt). In practice, the most effective approach is for HR to lead with support from marketing and IT.

Q: Is this just another SEO fad that will pass?

A: No. AI search is not a marketing channel — it's a fundamental shift in how people access information. Just as companies that ignored SEO in the 2010s fell behind, companies that ignore AI visibility in 2026 will find themselves increasingly invisible to candidates. The data is clear: AI adoption in candidate research is accelerating, not plateauing.