Building Entity Authority That LLMs Can Cite

Why "entity authority" replaced "domain authority" for local AI

For a decade, local marketers chased domain authority — backlinks, referring domains, PageRank proxies. That still helps traditional SEO. It does not fully explain why ChatGPT names a competitor with a weaker link profile but stronger entity clarity.

Entity authority describes how confidently automated systems — search indexes, map graphs, retrieval pipelines, and LLM grounding layers — can answer four questions about your business:

1. Who is this organization (legal name, brand, DBA)?
2. Where do they operate (address, service area, geo boundaries)?
3. What do they do (categories, services, credentials)?
4. How can a buyer reach them (phone, booking, hours)?

When those answers disagree across Google Business Profile, Apple Business Connect, Yelp, your website footer, and a stale Angi listing, entity resolution fails. Models hedge, omit you, or attach facts to the wrong location.

AEO, GEO, and LLM SEO converge on entity work because recommendation engines synthesize from entity-shaped data — not from keyword density on a single landing page.

This is technical marketing. The goal is not to trick an algorithm. The goal is to make your business easy to cite honestly.

Entity resolution — how machines pick one business among many

Entity resolution is the process of deciding that "Joe's Plumbing LLC," "Joe's Plumbing," and "Joes Plumbing Austin" at slightly different addresses are — or are not — the same economic unit.

Search engines solved this for the local pack. AI assistants inherit the same problem at larger scale:

• Training data may contain outdated names
• Browsing retrieves live pages with conflicting phones
• Retrieval-augmented generation (RAG) pulls snippets from directories you forgot you had
• Map APIs return place IDs that must align with web entities

High entity authority means fewer collisions. Low entity authority produces the failure modes owners describe in support forums: wrong hours in ChatGPT, a competitor's phone number, "permanently closed" when you are open, or complete invisibility despite ranking on Google.

Fixing entity authority is upstream of AI indexing — the cumulative process by which public facts about you enter corpora models may read, browse, or retrieve.

The entity stack local businesses actually control

You cannot edit OpenAI's weights. You can control verifiable public signals. Think in layers:

Layer 1 — Root listings (map graphs)

These are primary identity anchors:

• Google Business Profile — dominant for Gemini and Google AI Overviews
• Apple Business Connect — Siri, Apple Maps, Apple Intelligence local context
• Bing Places — Microsoft Copilot and Bing chat surfaces
• Yelp — persistent third-party corroboration in many US markets

Each root listing should share identical NAP (name, address, phone) formatting. "Ste 200" vs "Suite 200" vs "#200" is not cosmetic — parsers treat them as different strings.

Guide: Apple Business Connect for local AI visibility.

Layer 2 — Aggregators and vertical directories

Data vendors propagate your NAP to dozens of downstream sites. Industry-specific directories add category authority:

• Healthcare: Healthgrades, Zocdoc, Vitals
• Legal: Avvo, FindLaw, state bar directories
• Home services: Angi, HomeAdvisor, BBB
• Restaurants: TripAdvisor, OpenTable (where applicable)

AI systems treat multi-source agreement as confidence. A business named consistently across Google, Apple, Yelp, and a vertical directory is easier to recommend than one strong on Google alone with chaos everywhere else.

Layer 3 — Your website as canonical entity document

Your site is the canonical URL you control. Minimum technical requirements:

• JSON-LD LocalBusiness (or subtype: Dentist, Plumber, LegalService) with @id stable across pages
• Visible NAP in footer matching GBP exactly
• About page with sourced facts — founding year, service area cities, licenses, insurance — not vague superlatives
• Contact page with click-to-call matching primary listing phone
• llms.txt at domain root summarizing entity facts and linking to authoritative paths

Checklist: llms.txt, schema, and robots.

Layer 4 — Citable third-party pages

Entity authority extends beyond listings. Models grounding in live web search favor URLs with traceable claims:

• Local press covering a real milestone (expansion, community program)
• Merit-based awards with public criteria
• Data studies on your domain citing public datasets (permit filings, census tracts, industry benchmarks)

This layer supports GEO on retrieval-heavy engines — Perplexity, Claude with search, ChatGPT browsing — where a directory alone is insufficient.

Layer 5 — Reviews as entity-attached evidence

Reviews bind to place entities, not homepages. Google reviews attach to GBP place IDs; Yelp to Yelp business IDs. Review text supplies theme evidence models quote: "same-day," "explains options," "fixed leak under slab."

Review strategy intersects entity work when review counts and themes differentiate you from a namesake competitor in another state.

Guide: Google reviews the right way.

JSON-LD LocalBusiness — what to implement (and what to skip)

Schema is not magic markup. It is structured disclosure of facts you already assert publicly.

Fields that matter for local entity clarity

Property	Purpose
@type	Specific subtype beats generic LocalBusiness when accurate
name	Match GBP legal/trade name
url	Canonical HTTPS homepage
telephone	E.164 or consistent national format — match GBP
address	PostalAddress with street, locality, region, postalCode
geo	Lat/long if you have verified coordinates
areaServed	Cities, counties, or GeoCircle for service-area businesses
openingHoursSpecification	Match live listing hours
sameAs	URLs of official social and directory profiles
hasMap	Link to Google Maps place URL where appropriate

Anti-patterns that waste time or harm trust

• Fake aggregateRating without visible on-page reviews matching the markup
• Service area spam — listing 40 cities you never visit
• Keyword-stuffed description with no verifiable claims
• Multiple conflicting LocalBusiness blocks on the same page
• Automated schema pulling old phone numbers from a database nobody updated

Validate with Google's Rich Results Test and manual spot checks after any office move or rebrand.

llms.txt — lightweight AI indexing signal

llms.txt is an emerging convention: a markdown or plain-text file at /llms.txt that tells AI crawlers:

• Official business name and canonical site
• Primary phone and address
• Service summary in factual prose
• Links to /about, /services, /contact, and any fact-dense resources

It does not replace schema or listings. It orients models that explicitly look for publisher guidance — similar in spirit to robots.txt but for human-readable entity summary.

Example structure (adapt to your facts):

# Business Name
> Short factual tagline — city, primary service, credential if relevant.

## Contact
- Phone: (555) 555-0100
- Address: 123 Main St, Suite 200, Austin, TX 78701
- Website: https://example.com

## Services
- Emergency plumbing — Travis County
- Water heater install and repair

## Key pages
- About: https://example.com/about
- Services: https://example.com/services
- Service area: https://example.com/areas-served

Keep it updated when NAP changes. Stale llms.txt is another conflict source.

NAP consistency — the highest-leverage entity fix

Most entity failures are boring: two phone numbers, an old suite number on Yelp, holiday hours updated on Google but not Apple, a 2019 location page indexed with prior address.

Run a NAP audit quarterly:

1. Export NAP from GBP as source of truth (or internal ops doc if you multi-location)
2. Compare Apple BC, Bing, Yelp, website footer, schema, llms.txt
3. Log conflicts in a spreadsheet with priority by traffic and AI sampling impact
4. Fix root listings first, then aggregators (may take 2–6 weeks to propagate)
5. Resample AI mention prompts after propagation window

AIrecommend.ai Listings + Apple BC (Growth Engine module #3) automates drift detection and queues fixes with client approval — critical when you lack staff to manually re-check forty directories.

When models state wrong facts despite your fixes, trace upstream sources: AI reputation repair for wrong facts.

Entity authority vs brand authority

Brand authority is human reputation — trust, recall, word of mouth.

Entity authority is machine resolvability — can systems attach accurate attributes to the correct node in a graph?

A beloved local brand can have weak entity authority if:

• A franchise namesake dominates training data
• A former location's listings still rank in directories
• The owner never claimed Apple BC
• The website redesign dropped schema

Conversely, strong entity authority without brand warmth produces mentions — but not conversions. You need both. This article focuses on the machine half because marketers already understand brand; they underestimate graph hygiene.

How different AI platforms use entity signals

No platform publishes weights. Observable patterns from cross-platform sampling:

Google Gemini and AI Overviews

Heavy reliance on Google's local graph — GBP completeness, Q&A, reviews, posts. Entity schema on your site supports Knowledge Graph alignment but does not replace GBP.

ChatGPT

Blends training memory, optional browsing, and third-party plugins/tools depending on user settings. Entity conflicts in browsed results produce hedged answers or wrong phones. Browsing-heavy sessions favor fresh, consistent directory agreement.

Perplexity

Citation-forward — favors pages and directories it can quote with URLs. Entity listings plus citable domain content (studies, detailed service pages) outperform brochureware.

Strategy comparison: covered in depth in Perplexity vs ChatGPT local visibility.

Claude

Similar retrieval emphasis when search enabled. Values precise, sourced copy. Vague marketing superlatives ground poorly.

Grok and emerging engines

Treat as additional samples in mention-rate tracking — low overlap with incumbents. See the 11% platform overlap problem.

Implication: Entity authority is platform-spanning infrastructure. Optimizing only for Google leaves retrieval gaps elsewhere.

Multi-location and franchise entity architecture

Single-location guidance scales with discipline:

• One GBP per physical location — never merge distinct addresses
• Location pages with unique LocalBusiness @id URLs (/locations/austin/)
• Centralized NAP governance — franchisees should not edit phone without HQ sync
• parentOrganization schema linking locations to brand entity where appropriate
• Apple BC and Bing claims per location, not corporate HQ alone

Franchise systems fail entity resolution when each franchisee uses a personal cell as primary phone — models cannot unify routing.

Entity authority measurement

Technical fixes without measurement are guesswork. Track:

Metric	What it tells you
Mention rate by platform	Are prompts naming you at all?
Fact accuracy rate	Hours, phone, services stated correctly?
NAP drift count	Open listing conflicts
Schema validation status	Parse errors blocking rich understanding
Directory coverage score	Claimed vs unclaimed high-value directories

Baseline with a free six-platform scan. Resample monthly with the same buyer-intent prompt library. Compare trend lines, not single snapshots.

Full framework: What is AEO? Complete guide.

AIrecommend.ai Entity Profile module

Building entity authority manually is achievable for a single-location owner with technical comfort. It does not scale when:

• You operate multiple locations
• Directories drift faster than you can audit
• Wrong facts appear in AI answers and you need source tracing
• Marketing staff lack schema expertise

Entity Profile (Growth Engine module #4) generates and maintains:

• JSON-LD aligned to onboarding facts
• llms.txt with canonical NAP and page pointers
• About-page fact blocks optimized for grounding (not keyword stuffing)
• Refresh cadence when you submit change requests

Everything enters the approval queue — no silent publishes to your site or directories.

Growth tier ($4,997/mo) includes Entity Profile with Review Engine, Listings + Apple BC, monitoring, and Super Pixel attribution. Dominance tier ($9,999/mo) adds GBP Autopilot, Data Studies, Press Wire, Awards, and AI Accuracy Repair for retrieval-heavy GEO work.

Public pricing · AEO services · GEO services.

What honest entity work refuses to do

• Guarantee #1 AI placement — third-party platforms, no vendor control
• Fabricate sameAs links to profiles you do not own
• Inject fake reviews to boost entity-attached sentiment
• Publish schema ratings that do not match visible evidence
• Auto-change listings without client approval

Entity authority is infrastructure honesty. The businesses LLMs cite repeatedly tend to be the ones easiest to describe truthfully in public data.

90-day entity authority roadmap

Days 1–14: Audit and claim

• Run AI visibility scan and NAP audit
• Claim Apple BC, Bing, Yelp if unclaimed
• Document canonical NAP in one internal doc

Days 15–30: Website and schema

• Deploy or fix LocalBusiness JSON-LD
• Publish or refresh llms.txt
• Align footer, contact, and About NAP

Days 31–60: Directory propagation

• Fix root listing conflicts
• Submit aggregator corrections
• Add sameAs social and directory URLs to schema

Days 61–90: Corroboration and measure

• Publish one citable fact page (service area guide, sourced FAQ)
• Accelerate ethical review velocity for theme evidence
• Resample mention rates; log fact accuracy improvements

Adjust pace for your market — competitive categories may need ongoing Dominance-tier modules (studies, press) for Perplexity-class retrieval.

Related reading

• How AI assistants choose businesses
• AEO vs GEO vs SEO
• How to check what ChatGPT says about your business
• Zero-click AI searches for local business

Entity authority is not a one-time project. Listings drift. Models update. Competitors improve signals. Treat entity maintenance as operational AEO — the technical foundation LLM recommendations build on.