How AI Assistants Decide Which Businesses to Recommend
AI assistants like ChatGPT, Gemini, and Perplexity recommend businesses based on three signals: domain authority and topical relevance (how credible and focused your content is), structured data that explicitly tells the AI what you do and where you operate, and citation frequency from other authoritative sources. Unlike search rankings, AI recommendations cannot be purchased—they are earned through content quality, schema markup, and authority signals that AI crawlers can verify.
What Signals Do AI Assistants Use to Evaluate Businesses?
When you ask ChatGPT 'best plumbers in Denver' or Gemini 'who does logo design in Austin,' the AI is not running a real-time search query in the way a search engine does. Instead, it draws from patterns learned during training and, where available, cross-references live data sources against three core evaluation layers. Understanding these layers is the difference between being invisible to AI assistants and appearing in their recommendations.
The first signal is domain authority and topical focus. AI models were trained on large portions of the public web, so they carry forward implicit authority patterns from that training data. Modern AI assistants with live retrieval capabilities can also access current web indexes and business databases. A business with strong domain authority—earned through consistent, high-quality content, inbound links, and topical depth—starts with a credibility advantage. Topical focus matters equally: a dentist with many pages specifically about dentistry will generally be weighted more heavily for dental queries than a general health site that mentions dentistry in passing.
Signal 1: How Does Domain Authority and Topical Relevance Affect AI Recommendations?
Domain authority reflects how trustworthy and established a website is, built over time through consistent high-quality content, inbound links from other authoritative sites, and a sound technical foundation. For AI assistants, authority signals come from multiple sources: the depth and age of the domain's content history, the quality and quantity of inbound links, and the breadth of coverage on a specific topic.
Topical relevance is equally critical. If you are a plumber and your website has many pages covering plumbing services, emergency repairs, maintenance guides, and local service areas, you are more likely to appear in AI recommendations for plumbing queries than a general contractor with a single plumbing page. AI assistants recognize topical clusters—they treat a site with deep, interconnected content on a single subject as more authoritative than a site that touches many topics shallowly. This is why content silos and topical mapping matter not just for traditional SEO but for AI recommendation visibility.
Signal 2: How Does Structured Data and Schema Markup Influence AI Recommendations?
Structured data is machine-readable code embedded in your website that explicitly tells AI crawlers what you do, where you operate, your contact information, hours, and qualifications. Unlike prose content written for human readers, schema markup is a formal vocabulary (JSON-LD is the current standard recommended by Google and schema.org) that AI systems can parse with high reliability. This is the difference between an AI inferring 'this website appears to be about plumbing' and the website explicitly stating its business type, service area, and license number in a structured format.
The most important schema types for local and service businesses are LocalBusiness, Service, and Organization. LocalBusiness schema includes your name, address, phone, hours, service radius, and aggregate ratings. Service schema describes what you offer in detail. Organization schema establishes your company's identity, logo, and credentials. When AI crawlers ingest a page with complete, accurate schema, they have structured, verifiable information about your business rather than having to infer it from prose.
- Choose your primary schema type
Visit schema.org and identify your business category. If you are a local service business (plumber, electrician, salon), use LocalBusiness or an appropriate subtype. If you offer a specific service, also add Service schema. Copy the JSON-LD template for your type from schema.org or Google's Structured Data documentation.
Why: Schema.org provides the canonical definitions that AI systems and search engines recognize. Using the correct type ensures crawlers categorize your business accurately rather than guessing.
✓ Checkpoint: You have a JSON-LD template that includes at minimum the name, address, telephone, and openingHoursSpecification fields.⚠ Pitfall: Using outdated Microdata or RDFa formats instead of JSON-LD. Google's developer documentation recommends JSON-LD; older formats may be parsed inconsistently. - Fill in core business fields
Complete name, address, telephone, website URL, and service area. For service area, use the areaServed property to list cities or regions you actually serve. For hours, use the openingHoursSpecification format (for example, Monday 09:00–17:00).
Why: These fields are the minimum AI crawlers use to verify your business exists and operates in the claimed location. Incomplete fields reduce the signal value of your schema.
✓ Checkpoint: Your schema includes name, address, phone, and areaServed with no blank or placeholder values.⚠ Pitfall: Listing service areas you do not actually serve, or using outdated hours. AI systems cross-reference schema against citations and reviews; inconsistencies reduce trust signals. - Add credentials and qualifications where applicable
If your business operates in a licensed or regulated field, add license numbers, certifications, or professional memberships using the hasCredential or makesOffer properties. For example, a licensed contractor would include their license type and number.
Why: Credentials are a high-trust signal, particularly for regulated fields. AI systems weight recommendations toward verifiable professionals when queries involve health, legal, financial, or safety topics.
✓ Checkpoint: Your schema includes credential fields with real, verifiable identifiers—not generic claims.⚠ Pitfall: Listing credentials you do not hold or exaggerating qualifications. Public license databases can be cross-referenced; inaccurate claims undermine your overall trust signals. - Include ratings and reviews schema accurately
Add aggregateRating schema with your average rating, review count, and rating value (1–5 scale). Only include ratings you can verify from your actual review sources (Google, Yelp, Trustpilot). Keep the count current.
Why: Review signals are a trust factor. AI assistants weight businesses with verified, recent reviews more heavily than those without.
✓ Checkpoint: Your aggregateRating reflects your actual current review data from a named platform. The count is updated at least monthly.⚠ Pitfall: Inflating review counts or ratings, or using stale data. AI systems can cross-check against public review platforms; discrepancies damage your recommendation standing. - Embed schema in your page head
Place your completed JSON-LD schema inside a <script type="application/ld+json"> tag in the <head> section of your website. If you use a CMS like WordPress or Webflow, use the platform's schema field or a dedicated schema plugin rather than pasting into the visible body.
Why: Placing schema in the head ensures crawlers parse it early in the page load sequence and do not miss it.
✓ Checkpoint: Viewing your page source (Ctrl+U or Cmd+U) shows a valid <script type="application/ld+json"> block with your schema data in the <head> section.⚠ Pitfall: Placing schema in the visible body or inside HTML comments. Crawlers may skip it or treat it as non-authoritative. - Validate your schema before publishing
Use Google's Rich Results Test (search.google.com/test/rich-results) or the Schema.org validator to paste your JSON-LD and check for errors. Fix any errors before deploying; review warnings to determine if they affect required fields.
Why: Validation catches syntax errors and missing required fields before the schema goes live, preventing wasted crawl cycles.
✓ Checkpoint: The validator shows no errors. Any warnings are for optional fields and do not affect core business information.⚠ Pitfall: Deploying unvalidated schema. Malformed JSON-LD may be ignored entirely by crawlers, eliminating the signal benefit. - Monitor schema performance in Google Search Console
In Search Console, navigate to Enhancements and look for your business schema type (for example, Local Business). Review error reports and resolve any issues promptly.
Why: Search Console shows whether Google's crawler successfully parsed your schema and how many pages include it. Errors here indicate the schema is not being used.
✓ Checkpoint: Search Console shows your schema type is detected with zero errors. Any warnings are documented with a resolution plan.⚠ Pitfall: Ignoring Search Console errors. Unresolved schema issues prevent Google and downstream AI systems that rely on Google's index from using your structured data.
Signal 3: How Do Citations and Third-Party Verification Affect AI Recommendations?
A citation in the context of AI recommendations is any mention of your business on an authoritative third-party source: Google Business Profile, Yelp, industry directories, local business listings, news articles, or other high-authority websites. AI systems use citations to verify that your business is real, that the information you publish about yourself is consistent, and that other trusted sources reference you. A business mentioned across multiple local directories with consistent name, address, and phone information is more likely to be recommended than a business with accurate schema on its own website but no external citations.
Citation consistency matters significantly. If your website lists one address format and Google Business Profile lists a slightly different format, AI systems may still recognize these as the same business but note the inconsistency. Minor variations (St vs Street, CO vs Colorado) are often normalized automatically. However, major discrepancies—different phone numbers, conflicting addresses, or different business names across platforms—can reduce recommendation weight. This is why maintaining consistent Name, Address, and Phone (NAP) information across all platforms is a foundational practice for AI visibility.
0/10 complete
How Do AI Assistants Weigh These Signals When Generating Recommendations?
When you ask ChatGPT 'best dental practice in Portland,' the assistant does not rank websites the way a search engine does. Instead, it retrieves candidate businesses from its training data and, where retrieval is enabled, from live sources, then evaluates them against the three signal categories. The relative weighting of these signals is not publicly documented by AI platform operators, so any specific percentages (such as '40% authority, 35% schema, 25% citations') would be speculative. What is documented through platform developer guidance and schema.org specifications is that all three signal types—authority, structured data, and citations—contribute to how confidently an AI system can identify and recommend a business.
Query context shifts which signals matter most. For queries in regulated fields (medicine, law, finance), credential verification and citation authority from professional sources appear to carry more weight. For local or niche queries, topical relevance and recent citations matter more. Recency also plays a role: keeping your Google Business Profile, Yelp listing, and website information current signals to AI systems that your business is active. Similarly, recent reviews and fresh content indicate ongoing relevance. The practical implication is that AI recommendation visibility is not a one-time setup task—it requires ongoing maintenance.
What Mistakes Prevent Businesses from Appearing in AI Recommendations?
The most common mistake is publishing schema without supporting topical content. A business might add complete LocalBusiness schema but have only a thin homepage with no service pages, FAQ content, or local information. AI systems expect schema to be corroborated by real, detailed content. If your schema identifies you as a tax accountant but your website has no pages about tax strategy, deductions, or local tax considerations, AI systems lack the content evidence to confidently recommend you.
The second mistake is inconsistent NAP across platforms. A business listed as 'John's Plumbing' on Google, 'Johns Plumbing LLC' on Yelp, and 'John Plumbing' on its own website creates fragmented citation authority. AI systems may treat these as separate entities or flag the inconsistency as a trust issue. The fix is straightforward: choose one canonical business name and phone number, then standardize it across every platform where you have a listing.
The third mistake is neglecting review recency. A business with many reviews from several years ago may be weighted lower than a business with fewer but more recent reviews. AI systems interpret recent reviews as a signal of active, ongoing business operations. If your most recent reviews are old, your recommendation weight may decay over time. Consistently encouraging customers to leave reviews is part of maintaining AI recommendation visibility, not a one-time effort.
The fourth mistake is publishing inaccurate or exaggerated information in schema or citations. AI systems cross-reference claims against external sources. If your schema claims you serve an entire state but you have no reviews or citations outside a single city, AI systems may treat the claim as unverified. False credentials or inflated review counts can be detected through cross-checking with public databases and review platforms. Accuracy and consistency are the foundation of sustainable AI recommendation visibility.
What Is the Complete Workflow for Building an AI-Visible Business Presence?
Start by establishing topical authority through content depth: build out service pages, FAQ content, and local guides organized into a clear topical structure. Simultaneously, claim and optimize your Google Business Profile, Yelp listing, and relevant industry directories. Add complete, accurate schema markup to your homepage and all service pages. Then focus on citation consistency and review recency: encourage customers to leave reviews, audit your NAP consistency across all platforms, and update information promptly whenever anything changes. Finally, monitor your AI recommendation visibility quarterly by searching your business name and key service terms in ChatGPT, Gemini, and Perplexity.
How Can You Measure AI Recommendation Success?
Unlike search rankings, which you can measure directly in Google Search Console, AI recommendation visibility is harder to quantify because each AI assistant uses different data sources and refresh cycles. You can measure it indirectly through citation growth, review recency, and referral traffic. If your analytics show traffic arriving from ChatGPT or Perplexity (check referrer data), you are being recommended. Growth in citation count and review recency are leading indicators that your recommendation likelihood is improving.
A practical monitoring approach: once per quarter, search your business name and key service terms in ChatGPT, Gemini, and Perplexity. Note whether you appear in recommendations and in what position relative to competitors. If competitors appear but you do not, audit your schema completeness, citation consistency, and review recency. If you appear but not prominently, look for gaps in topical content depth or citation authority. Improvement in AI recommendation visibility typically takes longer than traditional SEO changes—allow several months before drawing conclusions from any single change.
FAQ: How AI Assistant Recommendations Work
No. ChatGPT, Gemini, Claude, and Perplexity do not accept payment for recommendations, as stated in their published platform policies. Recommendations are determined by domain authority, structured data, and citation signals—not advertising budgets.
What Are Your Next Steps for AI Recommendation Visibility?
If you have not yet optimized for AI recommendations, start with these foundational steps: (1) Claim your Google Business Profile and verify that your NAP is accurate and complete. (2) Add complete LocalBusiness schema to your homepage and service pages, then validate it using Google's Rich Results Test. (3) List your business in five to ten relevant directories and audit NAP consistency across all of them. (4) Encourage recent customers to leave reviews on Google and Yelp. This foundation typically takes two to four weeks to implement and positions your business for improved AI visibility over the following months.
If you already have citations and reviews in place, focus on content depth and recency: publish additional service guides, FAQ pages, and local content over the next sixty days, and ensure your citations are current. Then begin quarterly monitoring—search your business name and key service terms in ChatGPT, Gemini, and Perplexity, note your position relative to competitors, and use those observations to identify your next priority: more reviews, deeper content, or citation expansion.