The Question Behind the Question: Mapping AI Search Sub-Queries for UK Small Businesses
Photo: Francisclarke, CC BY-SA 4.0, via Wikimedia Commons
AI-powered search has changed the unit of competition. A customer may still type a broad question into a search box, but the system behind it may not treat that question as a single request. It may fan out into narrower sub-queries: definitions, costs, comparisons, suitability checks, local availability and aftercare. For UK SMEs, the commercial risk is not always a dramatic ranking fall. It is a quieter problem: the business has a service page, but not the evidence needed at each step of the fan-out.
That distinction matters for search strategy. A page that says a firm offers boiler installations may satisfy a head term. It may not answer the sub-queries that determine whether an AI-generated answer can cite the firm, or whether a conventional result set surfaces it for a more specific need. The task in 2026 is to audit the questions behind the question.
What fan-out means in practice
Fan-out is the process by which a broad query is decomposed into related retrieval tasks. The exact behaviour varies by platform and changes over time, so SMEs should not chase a single algorithm. The practical pattern is more stable: a customer asks one thing, but the decision requires several smaller things to be true.
Consider a homeowner asking whether a heat pump is worth it. The visible question is short. The underlying sub-queries may include retrofit suitability for an older property, insulation requirements, running costs, grant eligibility, noise, space, warranty and local installer availability. A business that answers only the headline question leaves the rest of the evidence set to competitors, aggregators or generic advice sites.
Why a single service page is rarely enough
Many UK small-business websites are organised around services rather than questions. That is understandable. Service pages are easier to write, sell and maintain. But fan-out rewards pages that resolve specific uncertainties. A service page can be technically relevant and still be thin in the eyes of a retrieval system because it lacks the detail needed for a sub-query.
Thinness is not only a matter of word count. A 900-word page can be thin if it repeats sales language and never states a cost range, a comparison, an eligibility rule or a local example. A 300-word answer block can be strong if it directly resolves a sub-query and sits within a clear page structure. The audit should therefore assess answer completeness, not just page length.
A five-question mapping exercise
Take five real customer questions. Use sales calls, enquiry emails, front-desk notes, site search logs and Search Console queries. Avoid invented questions that flatter the existing site. The aim is to capture the wording customers actually use, then map the likely fan-out beneath each one.
For illustration, consider a fictional Leeds-based heating and plumbing firm. Its five questions might be:
| Customer question | Likely fan-out sub-queries | Evidence the site needs |
|---|---|---|
| Why does my boiler keep losing pressure? | Common causes of pressure loss; how to repressurise a combi boiler; when pressure loss needs an engineer; whether pressure loss is covered by warranty | A troubleshooting guide with safety caveats; a clear engineer-call page; warranty guidance |
| How much does a new boiler cost in the UK? | Combi boiler replacement cost in 2026; system boiler versus combi cost; boiler finance options; grant or scheme eligibility | A cost-range page with assumptions; a comparison page; finance and grant explainers |
| Is a heat pump suitable for an old house? | Heat pump retrofit for terraced properties; insulation requirements; running costs; Boiler Upgrade Scheme eligibility; space and noise considerations | A suitability guide; a local case study; a grant explainer; a clear next-step assessment page |
| Do I need a power flush? | Symptoms that suggest a power flush; power flush cost; power flush versus chemical flush; when a power flush is not advisable | A diagnostic page; a price guide; a comparison table; risk and suitability notes |
| Which boiler brand is reliable? | Worcester Bosch versus Vaillant; boiler reliability surveys; warranty comparison; local installer recommendations | A brand-neutral comparison table; warranty explainer; installer selection guidance |
For each row, add three audit columns: the existing URL, whether the answer is present, partial or absent, and the format of the evidence. A single URL may serve several sub-queries, which is efficient. The point is not to create a page for every variation. It is to confirm that the site contains a clear, crawlable answer to each meaningful step.
Reading the gaps
The output of the exercise is usually a mixture of gaps. Some sub-queries are missing entirely. Others are thin: the page mentions the topic but does not resolve it. Some answers exist in a PDF, a staff-only document or an image, which makes them poor candidates for retrieval. Others are orphaned: the answer exists but is not linked from the page a customer or crawler would reach first.
A useful audit labels each gap rather than treating all absences as a content brief. A missing answer may require a new section. A thin answer may require a table, a cost range or a suitability caveat. An orphaned answer may require internal links and clearer headings. A PDF-only answer may need a web page summary. This prevents the common mistake of commissioning a batch of new pages when the site already contains fragments of the answer.
Fixing the map, not chasing the machine
Once the gaps are labelled, prioritise by commercial relevance. A sub-query that appears in sales calls and Search Console data matters more than a speculative question with no evidence of demand. Add concise answer blocks near the top of relevant pages, use descriptive subheadings, and place comparisons in tables where possible. Include local proof such as service area, response times, case examples or accredited installer status. Link related answers so that a customer journey through the fan-out remains on the site.
The audit should be repeated quarterly. Customer language changes, regulations change and AI search behaviour changes. The goal is not to predict every sub-query. It is to build a site that is legible when a broad question is broken into narrower ones. For UK SMEs in 2026, that is a more durable search strategy than treating fan-out as fallout.
