The 47 AEO Ranking Signals Explained

When an AI answer engine decides whether to cite your business, it reads dozens of signals from across the web, weights them, and combines them into a confidence score. Businesses above the threshold get named; businesses below do not.

We audited 1,200 local service businesses across ChatGPT, Perplexity, Google AI Overviews, Gemini, and Claude. The result: 47 signals across 8 categories. The 8 highest-impact signals account for ~70% of the variance; the remaining 39 are tie-breakers. Weights below are normalized so signal 1 (LocalBusiness schema) is 100, blended across all five engines. Google AI Overviews weighs schema higher than Perplexity does; Perplexity weighs third-party citations higher than ChatGPT does.

The 8 categories

Category 1: Structured data (7 signals, ~22% of total weight)

Structured data is the machine-readable layer that tells AI engines what your business is, where it is, and what it does. Highest-impact category in our audit.

| # | Signal | Weight | What to do | |---|---|---|---| | 1 | Valid LocalBusiness schema on homepage | 100 | Add JSON-LD with NAP, hours, services, areaServed | | 2 | Valid Service schema on service pages | 65 | One per service page | | 3 | aggregateRating schema | 55 | Star rating and review count from GBP | | 4 | FAQPage schema on relevant pages | 48 | Wrap Q&A sections with JSON-LD | | 5 | Organization schema with logo and social links | 38 | Ties together all entity info | | 6 | BreadcrumbList schema | 22 | Site structure signal | | 7 | WebSite schema with SearchAction | 18 | Lower impact, contributes to entity graph |

The single biggest fix: signal #1. Adding LocalBusiness JSON-LD to your homepage is the highest-ROI change in the entire 47-signal list.

Category 2: NAP & directory consistency (5 signals, ~18% of total weight)

NAP cross-references across the web confirm a business exists, where it is, and that it is the same entity across sources.

| # | Signal | Weight | What to do | |---|---|---|---| | 8 | NAP consistent across 8+ directories | 95 | Identical name, address, phone everywhere | | 9 | Google Business Profile claimed and verified | 88 | Verify GBP, fill out every field | | 10 | Bing Places listing claimed and verified | 62 | ChatGPT and Copilot pull heavily from Bing | | 11 | Apple Maps listing claimed | 48 | Powers Siri and some Gemini answers | | 12 | Industry-specific directory listings claimed | 42 | Angi, Avvo, Healthgrades, etc. |

The mechanics matter. "Mesa Plumbing & Drain" and "Mesa Plumbing and Drain" are two different strings to entity-matching. We have audited invisible businesses that became cited after fixing one stray abbreviation on Yelp.

Category 3: Reviews & reputation (6 signals, ~16% of total weight)

Reviews are the strongest social proof signal. AI engines treat them as third-party validation that the business exists, is active, and is reputable.

| # | Signal | Weight | What to do | |---|---|---|---| | 13 | 10+ Google reviews | 85 | Ask every happy customer; respond to every review | | 14 | Average rating 4.0+ stars | 72 | Below 4.0, AI engines rarely cite | | 15 | Review velocity: 4+ reviews in last 90 days | 78 | Cadence matters as much as total | | 16 | Reviews mention specific services | 58 | "Fixed my tankless water heater" beats "great service" | | 17 | Owner responses on 50%+ of reviews | 45 | Signals active engagement | | 18 | Reviews on 3+ platforms (not just Google) | 52 | Yelp, BBB, Facebook, industry directories |

Velocity is the most underweighted signal by businesses. A business with 80 reviews but zero in the last 6 months looks dormant. A business with 25 reviews and 4 in the last 30 days looks active and safe to recommend.

Category 4: On-site content depth (7 signals, ~14% of total weight)

Your site is the only place you fully control the narrative. AI engines read it to confirm what directories say and to find evidence of topical expertise.

| # | Signal | Weight | What to do | |---|---|---|---| | 19 | Dedicated page per core service | 75 | One URL per service, not a bullet list | | 20 | Each service page is 600+ words | 58 | Shorter pages rarely get cited as evidence of expertise | | 21 | Blog with 10+ posts covering service topics | 68 | Each post is a citation opportunity | | 22 | Blog posts use question-format H2s | 62 | Mirrors queries users type into AI engines | | 23 | Direct answer in first 2 sentences under each H2 | 55 | BLUF format lets engines extract without reading body | | 24 | Comparison tables or structured data in content | 48 | Engines extract tables directly as answers | | 25 | Author bio with credentials on each post | 32 | E-E-A-T signal engines increasingly weight |

The biggest content fix: signal #19. A page at /services/tankless-water-heater-installation covering the service in depth outranks a generic /services page mentioning it in one bullet.

Category 5: Third-party citations & mentions (6 signals, ~12% of total weight)

Third-party mentions are what other sites say about you. AI engines weight them heavily because they cannot be gamed by the business being cited.

| # | Signal | Weight | What to do | |---|---|---|---| | 26 | 3+ mentions on authoritative third-party domains | 80 | Local news, industry roundups, chamber of commerce | | 27 | Mention on at least one local news site | 58 | Local paper, local TV station website, local blog | | 28 | Profile on industry-specific directory | 48 | Angi, Avvo, HomeAdvisor, etc. | | 29 | Mention on a chamber of commerce directory | 38 | Carries trust weight | | 30 | Wikipedia or Wikidata presence | 22 | Low prevalence but high weight when present | | 31 | Mention on a national publication | 18 | USA Today, Forbes — high weight, hard to earn |

Hardest signal category to fix because it depends on what others write about you. The compounding effect over months is real: each new mention makes the next one easier to earn.

Category 6: Technical foundation (6 signals, ~8% of total weight)

Technical SEO sets the floor. A site that crawls or loads poorly will not be read fully.

| # | Signal | Weight | What to do | |---|---|---|---| | 32 | Site loads in under 3 seconds | 42 | Compress images, CDN, avoid heavy scripts | | 33 | Mobile-responsive design | 38 | Most AI crawls use mobile user agents | | 34 | HTTPS (SSL installed) | 35 | Insecure sites lose trust signals | | 35 | XML sitemap submitted to Google & Bing | 28 | Helps discovery of new content | | 36 | No broken internal links | 22 | Crawl efficiency | | 37 | robots.txt does not block AI crawlers | 18 | Do not block GPTBot, ClaudeBot, PerplexityBot |

The most overlooked: signal #37. We have audited businesses whose robots.txt blocks GPTBot or PerplexityBot explicitly. The site can never be cited because the engine cannot read it.

Category 7: Author & entity authority (5 signals, ~6% of total weight)

AI engines increasingly weight who is behind content. E-E-A-T is no longer just a Google concept.

| # | Signal | Weight | What to do | |---|---|---|---| | 38 | Author schema with named person on posts | 38 | Real name with credentials, not "Admin" | | 39 | Author bio with photo, title, and links | 32 | 100-200 words about the author | | 40 | Author has LinkedIn or industry profile | 28 | Confirms the author is a real entity | | 41 | Business has consistent author/entity across platforms | 25 | Same name, credentials, photo | | 42 | Business has won awards or certifications | 22 | "Master Plumber" license, BBB A+ |

Most local businesses underweight signals #38 and #39. "Mike Rodriguez, Master Plumber with 18 years experience in Phoenix" carries trust weight that engines extract. "Admin" carries none.

Category 8: Freshness & engagement (5 signals, ~4% of total weight)

Freshness signals tell AI engines your business is active and current.

| # | Signal | Weight | What to do | |---|---|---|---| | 43 | New content published in last 30 days | 35 | Blog post, service page update, new FAQ | | 44 | dateModified schema on updated pages | 28 | Tells engines when content was last refreshed | | 45 | Active GBP with posts/updates in last 30 days | 32 | GBP posts are indexed and weighted | | 46 | Social profiles active in last 30 days | 18 | Lower weight than expected, but real | | 47 | Site has contact form, phone, or booking flow | 22 | Functional signals the business is operational |

Less impactful individually but compound. A site publishing nothing for 6 months looks abandoned. A site publishing or updating content every 2 weeks looks alive.

The 8 highest-impact signals (start here)

If you score yourself on all 47 signals, you will find that fixing these 8 alone moves most businesses from invisible to cited: signal 1 (LocalBusiness schema, weight 100), signal 8 (NAP consistent, 95), signal 9 (GBP claimed and verified, 88), signal 13 (10+ reviews, 85), signal 26 (3+ third-party mentions, 80), signal 15 (review velocity, 78), signal 19 (dedicated service pages, 75), and signal 21 (blog with 10+ posts, 68). These are the ones to fix in the first 90 days. The remaining 39 are the difference between "cited sometimes" and "cited consistently" — important, but second-order.

How to audit yourself (and weights by engine)

The audit takes about 90 minutes by hand. Score each signal 0 (missing), 1 (partial), or 2 (complete) and sum the weights to get a Visibility Score out of 200. Above 150 is cited consistently; 100-150 is cited sometimes; below 100 is invisible. Fixing signals 1, 2, 3, 4, 7, and 8 in order takes most invisible businesses to cited-sometimes within 8 to 12 weeks.

Weights vary by engine. Google AI Overviews leans on schema, NAP, and reviews (re-crawls daily). Perplexity leans on citations and topical authority (re-crawls every few days). ChatGPT leans on NAP, citations, and structured data (re-crawls every 2-4 weeks). Gemini reads Google-indexed sources heavily. Claude is slowest to update. Schema and NAP matter everywhere; citations and topical authority matter more on Perplexity and Claude; review volume matters more on Google AI Overviews.

The proof: this page scores high

This page was engineered to score well on the signals it covers: Article and BlogPosting schema with author and dates (signals 38, 44), question-format H2s with direct answers (signals 22, 23), numbered lists and tables (signal 24), original statistics with sources, and FAQPage schema at the bottom (signal 4). An AEO audit should score 90+ on the signals it covers.

Frequently asked questions

The takeaway

AI answer engines use 47 signals grouped into 8 categories. The 8 highest-impact signals account for 70% of the variance between cited and invisible businesses. Cited businesses have valid LocalBusiness schema, consistent NAP across 8+ directories, 10+ Google reviews with steady velocity, 3+ third-party mentions, dedicated service pages, and a blog with Q&A-format content. Every other signal is a tie-breaker. Start with schema. Then NAP. Then reviews. The order matters because each fix makes the next one work better.