Ranking well on Google and appearing in an AI recommendation are two different things, and, unfortunately, they don’t always overlap. A business like yours can sit at position one for its core keywords (congratulations) and still be completely absent when someone opens ChatGPT and types “which financial adviser in Manchester should I contact?” or when Google’s AI Overviews are asked to recommend a plumber in their area, a different set of signals determines what comes back.
Prompt research is how you understand what those signals are.
Research covering 2025 found that the average ChatGPT prompt runs around 60 characters, compared to 3.4 for a Google search. People use AI tools the way they’d use a knowledgeable colleague, with a full question, real context, and no patience for a list of ten links to scroll through. Keyword research wasn’t built to account for that.
Our UK State of Digital Marketing Report 2026 found that 38% of UK marketing professionals say they need more training in how to optimise for AI-generated results, making it one of the top identified skills gaps this year. If you’ve been nodding along to conversations about AI search without being entirely sure what you’re supposed to do about it, you’re in good company. This is what it really means in practice.
Key takeaways:
- Prompt research identifies how real users ask questions to AI tools like ChatGPT, Perplexity and Google AI Overviews, then uses those patterns to shape content strategy.
- The average ChatGPT prompt is 17 times longer than a Google search query. People approach AI with full questions, context and an expectation of a direct answer.
- Prompt research focuses on the moments when AI is comparing options and making recommendations, which is when visibility matters most commercially.
- 38% of UK marketing professionals say they need more training in optimising for AI-generated search results, according to our UK State of Digital Marketing Report 2026.
- SEO isn’t being replaced. The strongest strategies combine traditional search optimisation with content built for AI citation.
Table of contents:
- What is prompt research?
- Quick AI terms glossary
- Why prompt research matters for UK businesses
- How to do prompt research
- Measuring success in AI search
- Prompt research FAQs
What is prompt research?
Prompt research is the practice of studying how real users phrase questions to AI systems, then shaping your content around those patterns so your business appears in the answers those systems generate.
It serves the same role for AI visibility that keyword research serves for traditional SEO, but the unit of measurement is different.
- Keyword research tracks short search terms and measures ranking positions.
- Prompt research tracks full, conversational questions and measures visibility in AI-generated answers.
The prompts that matter most commercially aren’t the informational ones. They’re the ones where AI is actively comparing options: “best IFA for small businesses in Manchester,” “top-rated emergency plumbers in Leeds for same-day callouts,” “which Italian restaurant in Birmingham has the best reviews?” When AI answers those questions, it names two or three businesses. If yours isn’t one of them, you didn’t make the shortlist. The customer moved on before you even knew they were there.
A quick AI terms glossary
These terms are everywhere in marketing conversations right now, so here’s a short rundown. Our AI marketing glossary covers all of them in more detail.
AI Overviews (AIO)
Google’s AI-generated summaries at the top of search results, pulling from multiple sources. A growing share of queries now resolve here before the user ever reaches an organic listing.
Generative Engine Optimisation (GEO)
Structuring and optimising content so AI platforms like ChatGPT, Google AI Overviews and Perplexity can extract, cite and recommend it. Prompt research tells you what questions to write content around. GEO shapes how you write and structure it.
AI search
ChatGPT, Perplexity, Google Gemini, Microsoft Copilot: platforms that generate direct answers rather than lists of links.
This is the one thing that should be said clearly: GEO doesn’t replace SEO. The signals that earn AI citations are largely the same ones that drive strong organic rankings. A well-structured, authoritative page is doing both jobs at once, which is a big win-win in digital marketing terms.
Why prompt research matters for UK businesses
Traditional keyword research tracked standardised terms with measurable search volumes. It’s still useful, but it doesn’t capture what happens when someone asks ChatGPT to recommend the best pizza place in their city or asks Google which pension option suits a self-employed person.
In AI search, your content either gets cited or it doesn’t. There’s no position four or five to fall back on. The businesses that tend to get cited share a few things in common:
- Their content directly answers the specific questions their audience asks
- They’ve built enough topical depth that AI platforms can draw on them across related queries
- They’ve got the authority signals that tell a platform the source is worth trusting
For UK SMEs, that’s actually good news. A local accountancy firm with specific, well-structured content answering what its clients search for has a real edge over a national competitor with broad, generic service pages.
In AI search, relevance and depth outperform domain size, which means smaller businesses can absolutely compete.
Our research into the top marketing challenges for UK businesses consistently shows that staying visible as search evolves is one of the most pressing concerns for marketing teams right now, and prompt research is the most direct response to it.
How to do prompt research
Step 1: Start with what you’ve already got
Your Google Search Console data contains long-tail queries of ten or more words that already behave like AI prompts. Filter for them. These conversational, intent-rich phrases are sitting in your data right now, and they’re your fastest route to a starting library without spending a penny on new tools.
Step 2: Mine real customer language
The most useful raw material is real language from real people:
Google’s “People Also Ask” sections: The questions already surfacing around your topic
Reddit threads in your sector: Unfiltered, unscripted customer language
Your own inbound emails and FAQs: If a customer asked it before buying, it’s worth building content around
The way a client describes their problem in an email to your office is closer to an AI prompt than anything a keyword tool will produce. That email is gold.
Step 3: Translate keywords into prompts
Take existing keywords and reframe them as the questions behind them:
“accountant London” → “What should I look for in a small business accountant in London and how much should I expect to pay?”
“private dentist Bristol” → “What’s the difference between NHS and private dental care in Bristol and how do I choose?”
“IFA near me” → “How do I find a financial adviser I can trust and what questions should I ask at the first meeting?”
The intent, context and qualifying detail surrounding the core term are what make a prompt, and that’s what AI systems are actually responding to.
Step 4: Build a prompt library
Organise prompts by where customers are in their decision:
Research stage: “What’s the difference between a will and a lasting power of attorney?”
Comparison stage: “Which type of pension is best for self-employed people in the UK?”
Decision stage: “How do I book an initial consultation with a financial adviser in Birmingham?”
This becomes the foundation of a content plan that serves both traditional and AI search, and makes it very clear where the gaps are.
Step 5: Test your visibility now
Open ChatGPT, Perplexity and Google in separate incognito windows and search for your category and location the way a customer would. Does your business appear? Are competitors being cited instead? Is the information shown accurate?
That’s your baseline, and it costs nothing. For more structured tracking over time, SISTRIX’s prompt research tool and SEMrush’s AI visibility suite make it possible to monitor this systematically across platforms.
Step 6: Create content AI wants to cite
Content that consistently gets cited tends to share the same characteristics:
- It answers the question directly at the start of each section
- It uses clear headings that reflect the question being answered
- It includes FAQ sections throughout
- It’s written by, or credited to, someone with genuine authority on the subject
- It cites trustworthy sources where relevant
Google’s guidance on helpful, reliable, people-first content points in exactly the same direction. The bar for what counts as “good content” has risen, but it’s still a bar that any business willing to put in the work can clear.
Quick wins for prompt research
Lead with the answer
Start every section with the direct answer, then elaborate. AI systems reward answer-first structure, and so do busy readers who are scanning for what they need.
Add FAQ sections with schema markup
AI tools are question-and-answer engines by nature. FAQ schema helps them identify and surface your content in generated responses.
Write for entities, not just keywords
Use your full business name, location and specialism consistently throughout. “Manchester-based independent financial adviser” builds a much clearer picture for AI systems than a generic “we.”
Build topical clusters
A pillar page supported by well-linked related articles signals depth of authority to both Google and AI platforms. One broad page doesn’t achieve what a thoughtful cluster of specific pages can.
Earn mentions and reviews
AI systems weigh brand mentions, third-party reviews and coverage in trusted publications as credibility signals. These aren’t just nice to have.
Check your robots.txt file
Make sure AI crawl bots, including GPTBot, ClaudeBot and PerplexityBot, aren’t inadvertently blocked from your site. It’s a five-minute check that businesses are surprised to find has been the problem all along.
Measuring success in AI search
Rankings mean less when AI generates answers above organic results. Here’s what’s worth tracking instead:
- AI referral traffic in GA4: Traffic arriving from ChatGPT, Perplexity and similar platforms appears as referral traffic and can be segmented. It’s the most direct measure of AI search performance available right now.
- Citation frequency for target prompts: How often does your business appear when you run your core prompts across ChatGPT, Perplexity and Google AI Overviews? Track this manually or through specialist tools as they develop.
- Share of voice: For the questions your customers are most likely to ask at the point of choosing a provider, are you consistently showing up? Tools like SEMrush and SISTRIX are building prompt tracking into their platforms, so the ability to measure this properly is becoming more accessible every month.
Is this worth tackling alone?
For most business owners who are already managing a hundred other things, probably not. Prompt research, GEO and AI Overviews represent a real skills challenge, and the expertise to do it well is still relatively scarce across UK marketing teams. The 38% training gap from our own research reflects that.
The most useful first step for most businesses isn’t building a full prompt library from scratch. It’s understanding where you currently stand. A free SEO audit covers the technical and content foundations that determine both traditional and AI search visibility and gives you a clear picture of what’s actually worth prioritising.
Alternatively, contact us today to find out how our team of award-winning SEO experts can help you improve your visibility across all aspects of search.
Prompt Research FAQs
What’s the difference between prompt research and keyword research?
Keyword research identifies the terms people type into search engines and tracks their volume. Prompt research looks at how people phrase full, natural-language questions to AI tools and uses those patterns to shape content strategy. Keywords show what people search for. Prompts show how they ask when they want a direct, considered answer back. Both matter in 2026.
How do I get my business to appear in Google AI Overviews?
Google’s AI Overviews pull almost exclusively from top organic results, so strong traditional SEO is still the foundation. Beyond that, clear content structure, FAQ sections, direct answers to the specific questions your audience asks, and well-cited authoritative content all improve the likelihood of being cited. Our guide to optimising for AI-powered search covers this in detail.
Do I still need SEO if I’m optimising for AI search?
Yes, and the two are more closely aligned than they might look. The signals that determine AI citation, clear structure, topical authority and direct answers, are largely the same ones that drive strong organic rankings. Businesses that invest in one tend to benefit in the other.
What is GEO and how does it relate to prompt research?
Generative Engine Optimisation (GEO) is the practice of structuring content so AI platforms can extract, cite and recommend it. Prompt research identifies what questions to build that content around. They work in sequence: prompt research tells you what to write, GEO shapes how to write and structure it.
How quickly will I see results from prompt research?
AI search visibility builds over time, much like traditional SEO does. The most useful starting point is a visibility check to understand where you stand right now. Content and structural improvements tend to show measurable impact within weeks as AI platforms re-evaluate their sources.
Related Articles
Non-Commodity Content: What It Is and Why Your Business Needs It
How much does SEO cost in the UK? A 2026 pricing guide
Is Your Business Visible in ChatGPT, Gemini and Perplexity? Here's How to Check
5 Examples of LOCALiQ UK Delivering Outstanding SEO Results for Clients
What is Local SEO? A Practical Guide for Businesses