For UK small businesses, the most important SEO change is not that search has suddenly stopped sending customers to websites. It is that the route from question to decision is becoming less predictable. AI-generated answers, product comparisons and local recommendations can now sit between a searcher and the familiar list of links. A potential customer may receive a concise answer, refine their question, compare options or visit only one cited source rather than browse five results.
That changes the job of SEO. Visibility is no longer just about occupying a blue-link position for a target keyword. It is about making your business easy to understand, credible enough to be selected as a source, technically available to search systems, and useful enough to win the visit or enquiry when a customer needs more than a summary.
This is especially relevant in the UK. Google’s AI Overviews and AI Mode are within the scope of the Competition and Markets Authority’s Strategic Market Status designation for Google Search. The CMA has since imposed requirements covering publisher controls, attribution, fair ranking and greater transparency. For owners, this does not remove the need to improve their websites. It does mean the rules and reporting available to them are changing at the same time as search behaviour changes.
The practical response is not to chase every new acronym or publish hundreds of generic AI-written pages. It is to protect the assets you control: first-party expertise, accurate business information, accessible pages and a measurement model that separates brand demand from genuine new-customer discovery.
AI search changes the shape of discovery, not the need for SEO
Google’s own guidance is clear that conventional SEO remains relevant to generative search because AI features use core Search ranking and quality systems to retrieve up-to-date pages from Google’s index. Its systems can also expand a question into related searches before compiling an answer. In plain English, a searcher asking for “the best boiler service plan for a Victorian terrace” may trigger searches around boiler type, local availability, pricing, maintenance and safety. A thin service page aimed at one exact phrase is less likely to meet that wider need than a useful, well-maintained resource from a real local specialist.
Google’s guide to generative AI search advises owners to focus on unique, useful content, crawlability, clear structure and a good page experience. It also rejects several supposed shortcuts: there is no special schema type required for AI visibility, no need to create an llms.txt file for Google Search, and no ideal “chunk” or page length.
That is good news for small businesses. You do not need a separate AI-search strategy that competes with sensible SEO. You need a sharper version of the strategy you should have had already: answer real customer questions better than generic websites can, make key information unambiguous, and remove barriers that prevent a page being discovered or used.
Think beyond the click, but do not dismiss the click
An AI answer can satisfy a simple factual query without a website visit. That is most likely where the customer only wants a definition, a rough comparison or a quick opening-hours check. But a person planning an expensive purchase, seeking a local provider, checking suitability, assessing risk or requesting a quote still needs evidence and confidence. The website’s role shifts towards proving expertise and helping the customer act.
For example, an independent Bristol accountant should not rely solely on a page titled “Accountants in Bristol”. It should also make it easy to establish who the firm helps, which tax problems it handles, the qualifications of the relevant team members, its current fees or pricing approach, its contact route and practical explanations of changes affecting owner-managed companies. Those details are useful to a human visitor and provide clear, current signals about what the firm actually knows and does.
First priority: build first-party expertise that cannot be commoditised
AI makes generic content easier and cheaper to produce. That also makes generic content less differentiating. A page that simply repeats what competitors, directories and AI tools already say gives neither a customer nor a search system much reason to prefer it.
Instead, identify the knowledge that exists inside your business but is absent from the wider web. This is first-party expertise: the observations, processes, evidence, products, local experience and professional judgement created through doing the work.
Turn customer conversations into useful pages
Start with your sales calls, reception queries, site visits, email questions and post-purchase support requests. Build a list of questions that genuinely influence a decision. Then assign each question to the person in your company best placed to answer it. A writer or agency can shape the page, but the raw material should come from someone with direct experience.
- A Manchester kitchen installer could explain realistic lead times, what a survey reveals, how customers should prepare their home, and the trade-offs between common worktop materials.
- A cybersecurity consultancy could publish an anonymised account of recurring problems found in small firms, the practical controls that prevented them, and what its assessment process covers.
- An independent retailer could provide genuine fitting guidance, care advice, stock availability principles and original product comparisons based on returns or customer feedback.
- A specialist manufacturer could document tolerances, testing methods, compatibility limits and common installation mistakes, rather than relying on a catalogue description alone.
Use plain language, but be specific. Name the author or responsible expert where appropriate. Include dates when information could change. Explain exceptions rather than presenting every answer as universal. Link related service, product and contact pages so a reader can move from research to action without starting again.
Google specifically recommends original viewpoints and content based on first-hand experience rather than material that merely restates existing sources. Its guidance also warns against producing high volumes of pages simply to capture every conceivable variation of a query. Useful, non-commodity content is therefore a better investment than an AI content factory.
Show the evidence behind the claims
Expertise should be visible, not implied. Add team biographies, credentials, relevant memberships, case studies with customer permission, before-and-after examples, clear methodologies, original photography and answers to limitations. For regulated, financial, medical or legal subjects, have a qualified person review content and state when it was last checked. Do not use vague phrases such as “leading experts” where you can show the reader why the business is qualified.
This is not just an AI-search tactic. It makes a small company more trustworthy when a prospect lands directly from a search result, recommendation, social post or email. The more consequential the buying decision, the more that evidence matters.
Second priority: make every important fact current and consistent
Generative search heightens the cost of stale information. A wrong price, discontinued product, former staff member, old service area or outdated policy can confuse customers and weaken confidence. For a local business, the basics are often the conversion point: phone number, opening hours, service radius, booking availability, delivery terms, returns process and whether you are taking on new work.
Create a simple “facts register” for the website. It can be a shared spreadsheet, but it should record each critical fact, the page or system where it appears, the business owner responsible for it, and the date it was last checked. Include the website, Google Business Profile, Merchant Center feed where applicable, booking platform, major social profiles and key directory listings.
Set a maintenance rhythm
Review high-risk information monthly: hours, contact details, stock status, promotions, prices, appointment availability and emergency-service coverage. Review core service and advice pages quarterly, and immediately after a regulatory, product, staffing or policy change. Add a visible “last updated” date to guidance where freshness matters, but do not change dates cosmetically. Update the substance, examples and links too.
For ecommerce, ensure that the product page, structured product data and product feed agree on price, availability, delivery and condition. For local services, ensure the service-area wording does not promise travel arrangements you cannot fulfil. Google notes that product and local business information can appear in generative responses, and points owners towards Merchant Center and Google Business Profile as useful supporting products. Keeping these business details accurate is a practical visibility task, not mere admin.
Third priority: make pages technically accessible
Excellent expertise cannot help if search systems cannot reliably find, render and index it. AI search does not bypass technical SEO. Google says that, to be eligible for generative features, a page must be indexed and eligible to appear in Search with a snippet. It also needs to be included in generative AI features in Search Console where that setting is available.
Give a developer or SEO supplier a focused technical brief rather than asking for a vague “AI optimisation” package. Ask them to check the pages that produce revenue or enquiries first.
- Can Google crawl the page, or is important content blocked by robots rules, login walls, consent tools or broken internal links?
- Does the page return a successful response and have a self-consistent canonical URL, rather than competing duplicate versions?
- Is the main text present in the rendered page, including on mobile, rather than appearing only after a user interaction that search systems may not reproduce?
- Are key pages included in an XML sitemap and linked from relevant navigation or content pages?
- Do titles, headings and visible copy clearly identify the service, product, location and audience without keyword stuffing?
- Does the site load acceptably, work on a phone, and present its main content without intrusive pop-ups or layout disruption?
- Are images useful, original where possible, properly described and supported by surrounding text?
Use semantic HTML and logical headings because they improve readability and accessibility. Use structured data where it accurately represents eligible content, such as products, events or local business details, but do not treat it as a magic AI citation button. Google says structured data is not required for generative AI search and that no special AI schema exists; its value is mainly in supporting normal Search features such as rich results.
Search Console should be part of the routine. Check the Pages report for indexing problems, inspect representative URLs after major site changes, and monitor Core Web Vitals and mobile usability where relevant. If a redesign or JavaScript framework is involved, require a pre-launch crawl and a post-launch check. The official technical guidance for generative Search stresses crawlability, JavaScript SEO, duplicate-content management and page experience.
Fourth priority: stop treating old SEO reporting as the whole story
Traditional reporting often celebrates total organic sessions, a handful of rank positions and year-on-year traffic. Those figures still matter, but they can hide the changing mechanics of discovery. A drop in click-through rate may reflect an altered results page, an AI response, a richer local result, a change in demand or a weaker snippet. Equally, a stable ranking number does not guarantee stable prominence because results pages contain many different elements.
Google itself cautions that average position is complex and can be misleading. It reflects the topmost link to your property averaged across impressions, not a simple, fixed place in a universal list. Some non-link or query-refinement elements do not occupy a position at all. Google recommends focusing more on trends in impressions and clicks than position alone. See its explanation of how impressions, clicks and average position are counted.
Separate brand demand from discovery demand
The priority metric for growth is often non-brand search performance: impressions, clicks, click-through rate and leads from queries where the searcher did not already ask for your company, product name or domain. Brand traffic can rise because of repeat customers, offline advertising, PR or reputation. That is valuable, but it can mask a decline in your ability to reach new buyers who are searching generically.
Eligible sites can now use Search Console’s branded and non-branded query filter. Google says branded searches include the business name, domain or closely associated product or service, while non-branded searches are more general. Build a monthly dashboard that compares both groups and segments them by the pages that matter: core services, locations, products, advice content and commercial landing pages.
- Track non-brand impressions to understand whether you are still being considered for generic searches.
- Track non-brand clicks and click-through rate to spot pages that are visible but no longer winning the visit.
- Track enquiries, calls, bookings, sales and qualified leads by landing page, not merely by total sessions.
- Record dates of major content, technical and pricing changes, then compare similar periods rather than declaring success after a single week.
- Review which queries and pages lost or gained the most, then check whether the page is current, technically sound and genuinely better than alternatives.
Use the filter intelligently. Google notes that it is unavailable for some low-impression sites, classifications can occasionally be wrong, and query reporting has privacy and data-limit constraints. It is a useful diagnostic, not an infallible ledger. The Search Console Performance report guidance also recommends examining page-level click-through rates and significant changes, while keeping the difference between Search, News and Discover reporting in mind.
Prepare for AI-specific reporting and choices
In June 2026, Google began testing a Search Console control for a subset of UK website owners to decide whether their content may appear in and ground generative AI features such as AI Overviews and AI Mode. Google says opting out means the site will not receive generative-feature traffic or impressions, and that this choice is not a ranking signal for results outside those features. It is also rolling out generative-AI performance insights, including impressions and page and country information. Owners should check Search Console rather than assume that every account has the feature immediately.
The choice should be commercial, not emotional. A business that depends on research-led discovery may prefer to remain included while measuring outcomes. A publisher with valuable content and a different revenue model may assess the trade-off differently. Keep a record of any setting changed, its effective date and the affected content so you can judge the result when reporting becomes available.
What the CMA’s requirements mean for small businesses
The CMA’s intervention is important because it recognises that AI search affects the relationship between Google, publishers and businesses. On 3 June 2026, the CMA imposed a publisher conduct requirement. It requires Google to provide effective controls over the use of publishers’ search content in generative AI, explain that use clearly, provide detailed engagement metrics, and take reasonable steps to ensure clear, accurate attribution and access to source content.
On 17 June 2026, the CMA imposed a separate fair-ranking requirement. Google must rank organic results using objective and non-discriminatory criteria, including in generative AI features; provide transparency and sufficient notice around material changes; and offer ways for publishers to raise concerns. The CMA says Google has six months from 17 June 2026 to implement the fair-ranking requirement, while the publisher requirement has a nine-month implementation window, although it expects important controls earlier.
For a small-business owner, the immediate implication is not that rankings are guaranteed or that every page will be cited. It is that you should retain evidence. Keep Search Console exports, screenshots of material changes, records of outages and correspondence with suppliers. If a serious visibility issue coincides with a documented platform change, you will be better prepared to use Google’s processes or contribute meaningful evidence through trade bodies.
Read the CMA’s publisher conduct requirement and fair-ranking conduct requirement rather than relying on headlines. The obligations are aimed at Google, but the resulting controls, attribution and reporting could give website owners more practical information and choice than older SEO reporting provided.
Conclusion: protect the parts of visibility you own
AI-generated answers make search less linear, but they do not make a strong small-business website obsolete. The businesses most likely to hold their visibility are those that publish real expertise, keep facts dependable, make their pages easy to crawl and use, and judge performance through non-brand discovery and commercial outcomes rather than rank reports alone.
Start this month. Audit your ten most valuable pages for accuracy and first-hand insight. Fix one technical barrier that prevents reliable indexing or mobile use. Set up a branded versus non-branded Search Console view. Then nominate one person to review your key business facts every month. Those actions will not manufacture visibility overnight, but they will give your business a more credible, measurable and resilient place in search as AI continues to reshape how customers discover it.
