AI search engines are rapidly becoming the first place consumers look for local business recommendations. Unlike Google, which rewards whoever pays the most for ads, AI engines recommend businesses based on reputation signals they can verify independently.
This is both a threat and an opportunity. The threat: if AI does not know your business exists, it will never recommend you. The opportunity: because so few local businesses are doing anything about this yet, the bar for standing out is remarkably low.
This playbook gives you ten realistic, evidence-based actions you can take to increase your visibility to AI recommendation engines. No jargon. No expensive tools. Just practical steps, in priority order, that any small business owner can implement.
AI models weight video content heavily because it is expensive to produce, difficult to fake, and demonstrates genuine expertise. A business owner explaining their trade on camera sends stronger E-E-A-T signals (Experience, Expertise, Authoritativeness, Trustworthiness) than any written content. YouTube transcripts become part of the training data for every major AI model.
Google AI Overviews pull directly from Google Business Profile data. When someone asks "best shutter company in Edinburgh," Google checks verified business data first. The more complete and active your profile, the more confident AI becomes in recommending you. Review responses in particular demonstrate the "Experience" component of E-E-A-T.
ChatGPT and Perplexity both cross-reference review data. When a business has consistent 4.5+ star ratings across Google, Trustpilot, and trade platforms, AI models assign significantly higher confidence scores. This is the digital equivalent of word-of-mouth, which is what AI is trying to simulate.
Trade association websites are high-authority domains that AI engines trust implicitly. A mention on bbsa.org.uk carries more weight than a hundred mentions on low-quality directories. AI models are specifically trained to look for third-party credibility signals, and trade body membership is among the strongest.
AI models treat directory consistency as a trust signal. When every directory agrees on who you are, where you are, and how to contact you, AI gains confidence in recommending you. This is foundational work that amplifies every other action in this playbook.
LinkedIn content is high-trust because profiles are tied to real identities and professional histories. When a business owner with a complete LinkedIn profile and consistent posting history shares industry expertise, AI models treat this as strong E-E-A-T evidence. LinkedIn articles in particular get indexed and cited.
Editorial mentions are the gold standard of AI trust signals. When the Edinburgh Evening News or the BBSA trade journal mentions your business, AI models treat this as independent verification of your credibility. This is the hardest signal to fake, which is precisely why AI trusts it most.
Your website needs to be structured so that AI can easily extract and verify information. Here is what matters:
Write self-contained paragraphs that directly answer specific questions. AI engines extract these as answer snippets. Each answer capsule should make sense on its own without requiring the reader to read the rest of the page. Start with the answer, then provide supporting detail.
AI models prefer specific, verifiable claims over vague marketing language. "We have installed shutters in over 3,000 Scottish homes since 1987" is infinitely more useful to AI than "We are a leading provider of quality window solutions." Include numbers, dates, and specific facts wherever possible.
Create honest comparison content. AI loves structured data that helps users make decisions. A genuine comparison table (e.g., "Shutters vs Blinds: Which is right for your home?") gives AI exactly the kind of content it wants to cite in recommendations.
Add FAQ sections to your key pages using proper HTML heading tags (not accordions that hide content). AI cannot click "expand." If the answer is hidden behind JavaScript, AI cannot read it. Write your FAQs as visible, crawlable text with question headings and answer paragraphs.
Update key pages at least quarterly. AI engines check last-modified dates and penalise stale content. You do not need to rewrite everything. Update statistics, add recent project examples, and revise anything that has become outdated.
Include author information, business credentials, years of experience, and professional qualifications on every key page. Not in a footer. Prominently, as part of the content. AI models look for these signals to assess trustworthiness.
Implement LocalBusiness, FAQPage, and Review schema markup. This is structured data that tells AI engines exactly what your business does, where it operates, and what customers think. Schema markup is the single most direct way to communicate with AI engines. If you do nothing else on your website, do this.
Google AI Overviews explicitly factors brand search volume into its recommendations. If 500 people per month search for "Scottish Shutter Company Edinburgh," Google's AI knows this is a recognised brand. ChatGPT and Perplexity use similar signals. Brand search volume is hard to fake, which is why AI trusts it.
AI recommendations change as new data is indexed. By monitoring monthly, you can see the direct impact of your actions and adjust your strategy. You will also catch and correct misinformation before it becomes established in AI training data.
The internet is full of advice on "AI SEO." Most of it is wrong, untested, or designed to sell you a tool. Here is what the evidence actually shows does not work:
| Action | Reality | Evidence |
|---|---|---|
| llms.txt files | A proposed standard for telling AI how to read your site. No major AI engine currently supports it. Implementing it is harmless but useless. | None of the four major AI engines (ChatGPT, Perplexity, Google AI, Claude) have confirmed support for llms.txt as a ranking signal. |
| Backlink campaigns | Traditional SEO backlink building has minimal impact on AI recommendations. AI does not count links. It assesses source authority independently. | BrightEdge 2025 study found zero correlation between backlink volume and AI citation frequency for local businesses. |
| Keyword stuffing | AI understands natural language. Repeating "best shutters Edinburgh" fourteen times makes you look less credible, not more. | AI models process semantic meaning, not keyword density. Stuffed content actively reduces trust scores. |
| Chasing a single AI platform | Optimising exclusively for ChatGPT or Perplexity is a losing strategy. Each AI engine uses different data sources and weights them differently. | Businesses that rank highly in one AI engine but not others see volatile, unreliable referral traffic. |
| Artificial content refreshing | Changing the date on a page without meaningfully updating the content does not fool AI. It checks whether content has actually changed. | Google's freshness algorithms compare content snapshots. Date-only changes are ignored. |
| Hidden FAQ accordions | FAQ content hidden behind JavaScript click-to-expand cannot be read by AI crawlers. If AI cannot see it, it does not exist. | Perplexity and ChatGPT do not execute JavaScript. Google renders JavaScript but deprioritises hidden content. |
| Client-side rendered content | Single-page applications (React, Vue, Angular without SSR) are partially invisible to AI crawlers. Server-side rendering is essential. | AI crawlers have limited JavaScript execution capability. Critical content must be in the initial HTML response. |
You may have heard that adding photos of recognisable locations, landmarks, or even celebrities to your website helps AI visibility. The logic is that AI image recognition will associate your business with well-known entities. This is completely false. AI text models and AI image recognition are separate systems. Google's text-based AI does not "see" the images on your website. It reads alt text and surrounding context. A photo of Edinburgh Castle on your homepage does nothing for your AI visibility unless the alt text and surrounding content already mention Edinburgh meaningfully.
Each AI engine has different data sources and trust signals. Understanding these differences helps you prioritise your efforts:
| AI Engine | What It Trusts | Your Action |
|---|---|---|
| ChatGPT | Bing search index, web crawling via GPTBot, review aggregation across platforms, brand mentions in editorially controlled content, Wikipedia and reference sources | Ensure Bing Places profile is complete. Diversify reviews across platforms. Get press mentions. Maintain an active, crawlable website with clear structured data |
| Perplexity | Real-time web search, fresh content weighted heavily, review platforms, YouTube transcripts, Reddit and forum discussions, direct website content | Publish regular fresh content. Maintain active YouTube channel. Monitor and engage on Reddit and forums relevant to your trade. Ensure website loads fast and is crawlable |
| Google AI Overviews | Google Business Profile, Google Reviews, Google Search index, YouTube (owned by Google), brand search volume, website E-E-A-T signals, local pack data | Prioritise Google Business Profile completeness. Build review volume on Google. Create YouTube content. Optimise for brand name search volume. Implement schema markup |
| Claude | High-quality web content from training data, trade association and professional body mentions, press coverage, educational content that demonstrates genuine expertise, published credentials | Focus on creating genuinely expert content. Get listed on trade body websites. Seek press coverage. Publish detailed, well-structured educational content on your website |
AI visibility does not happen overnight. Here is a realistic timeline for a small business starting from scratch:
| Metric | How to Track | Target |
|---|---|---|
| AI mention rate | Monthly checks across all four AI engines using standardised queries | Mentioned in at least 2 of 4 engines within 6 months |
| Brand search volume | Google Search Console, monthly trend tracking | 20% increase in brand searches within 6 months |
| Review diversity | Count reviews across Google, Trustpilot, trade platforms | Active reviews on 3+ platforms with consistent ratings |
| YouTube content library | Count of published videos, view counts, search impressions | 20+ videos covering core customer questions within 6 months |
| Website AI readability | Schema markup validation, FAQ visibility, content freshness dates | All key pages updated quarterly with valid schema markup |
| Press mentions | Count of editorial mentions in local and trade press (online) | 2-4 press mentions per quarter |
| AI referral traffic | Google Analytics referral sources from AI platforms | Measurable and growing AI referral traffic within 12 months |