Pattern A1, A2, and C: the three displacement plays in AI citation

Pattern A1, Pattern A2, and Pattern C are three citation behaviors we documented across 12 months of cross-client AEO audits. Pattern A is ChatGPT's tendency to cite a small, stable preferred roster of sources on agent-recommendation queries — picking the same handful of directories and personal sites repeatedly across thousands of buyer questions in a category. Pattern A splits into A1 (personal sites or branded properties) and A2 (vertical directories like FastExpert, Avvo, NerdWallet, Healthgrades, G2). Pattern C is the displacement play: a single well-engineered pillar page that displaces generic authority sources across an entire vertical, often outperforming the dedicated directories that normally dominate. The three patterns aren't mutually exclusive — most successful AEO programs use A2 directory presence as the foundation, then layer A1 owned-property optimization on top, then deploy Pattern C displacement pages to capture specific category queries directly. This article documents each pattern, shows real examples from our audit data, and lays out how to play each one tactically.

What are Patterns A1, A2, and C?

The patterns emerged from observing which sources AI assistants — primarily ChatGPT, with similar behavior on Perplexity — actually cite when answering buyer-recommendation queries. We ran the same category queries against the same AI assistant repeatedly over 90-day periods and tracked which sources got cited. Three citation patterns dominated:

Pattern A — The preferred roster. For any given category query ("best luxury real estate agent in Pacific Palisades," "best AI marketing platform for B2B SaaS," "best personal injury attorney in Boston"), ChatGPT cites a small stable set of 4-7 sources repeatedly. The roster shifts slowly (over months, not weeks) and largely the same 4-7 sources appear across thousands of variant queries within the category. The preferred roster splits further into:

• A1: Personal sites and branded properties. A specific real estate agent's site, a specific lawyer's firm site, a specific consultant's blog. These are owned brand properties that ChatGPT has come to trust as authoritative within the category.
• A2: Vertical directories. FastExpert and HomeLight for real estate; Avvo, Martindale-Hubbell, and Justia for legal; Healthgrades, Vitals, and Zocdoc for healthcare; G2, Capterra, and TrustRadius for B2B SaaS; Clutch and Agency Spotter for marketing agencies. ChatGPT treats these as authoritative third-party validators.

Pattern C — Displacement. A single well-engineered pillar page on a specific buyer-decision topic ("how much does it cost to repaint a 3,000 sq ft house in San Diego") that displaces the generic authority sites (HomeAdvisor, This Old House, Yelp) that normally dominate the query. Pattern C pages "displace" because they're more extractable than the generic authority pages despite having less brand authority — the content geometry wins over the brand weight.

The patterns aren't theoretical. They're empirical regularities in citation behavior. We see them consistently across verticals.

Pattern A: the preferred roster behavior

When we instrumented our audit pipeline to run 30+ category queries against ChatGPT for the same vertical, we noticed two things consistently. First, the citation set was much smaller than the universe of available sources. ChatGPT didn't cite 200 different real estate agents across 30 queries; it cited 12-15 repeatedly with the same 5-7 dominating across most queries. Second, the citation set was stable. Repeating the audit 60 days later returned mostly the same sources with minor turnover.

The behavior makes sense mechanically. ChatGPT's training corpus + retrieval system both reinforce sources that the model has high confidence in. A directory like FastExpert that the corpus represents heavily becomes the default citation for "best agent" queries because the model has thousands of training examples associating FastExpert with that category. New sources have to cross an evidence threshold before joining the roster.

The implication for brands: you can either join the preferred roster or you can be invisible. There is no middle ground — being cited on 3% of category queries is functionally the same as being cited on 0%. Getting into the roster is a step-function transition, not a gradual ramp.

Pattern A1: when ChatGPT cites personal sites

A1 — personal site or branded property citation — happens when ChatGPT considers an individual brand authoritative enough in its category to cite directly. This is rarer than A2 (directory citation) but more valuable when it happens — A1 citations drive direct attribution and traffic to the brand's own property.

Three things characterize brands that successfully achieve A1 citation:

High brand mention frequency. The brand appears across the open web in trusted publications, podcasts, trade press, and other authoritative sources at a frequency that makes the training corpus represent them densely. A real estate agent featured 30 times across LA Times, Mansion Global, and Forbes over 18 months crosses the threshold. An agent with comparable transaction volume but zero press doesn't.

Strong content citation geometry. The owned property publishes substantive content that's structurally extractable — 134-167 word passages, question-format H2s, sourced statistics, declared author entity. The agent's site reads like a citable answer for buyer questions in their category.

Clean entity signals. Wikipedia or Wikidata entity, complete Knowledge Graph presence, consistent NAP. The model can disambiguate confidently before citing.

Brands in A1 tend to be top-decile within their vertical for at least one of three metrics: press coverage frequency, original research published, or category authority via trade-association recognition. They tend to not be the highest-transaction-volume brands, which counterintuitively underweights commercial activity in the citation decision.

A1 citation is what most brands actually want from an AEO program: their own site cited directly. It's the most aspirational outcome and the slowest to achieve. Realistic timeline to A1 from a cold start: 12-24 months of consistent investment across content + brand mention + entity signals.

Pattern A2: the directory citation pattern

A2 — directory citation — is the dominant citation pattern across verticals. When ChatGPT answers "best [profession] in [city]" queries, it cites the vertical's authoritative directory more often than it cites any individual brand. The directories that occupy the A2 slot per vertical:

A2 citation works as a proxy: ChatGPT cites the directory, which contains a profile for the brand. The brand gets attributed indirectly. The user follows the directory link and then potentially finds the brand within it.

The leverage play: claim, complete, and optimize every relevant A2 directory profile. This is the highest-ROI same-week AEO action for almost every brand. A complete FastExpert profile takes 2-3 hours to build and meaningfully lifts the brand's effective citation surface in ChatGPT for years. The lift compounds because A2 directory presence also signals to other AI assistants (Perplexity, Claude) that the brand is category-relevant.

Critically, brands NOT in A2 directories are systematically invisible. We've audited multiple brands with strong content and brand mention frequency who scored poorly on ChatGPT because they weren't listed on the directories the model favors for their category. The single fix that lifted them most: claiming the missing directory profiles.

Pattern C: displacement through pillar content

Pattern C is the displacement play. A single well-engineered pillar page on a specific buyer-decision topic outperforms generic authority sources for that query. The pillar "displaces" — replaces the source ChatGPT would otherwise default to.

The mechanics: ChatGPT's preferred-roster behavior assumes generic queries. "Best painter in San Diego" returns generic-authority sources (Angi, HomeAdvisor, Yelp). But a query like "how much does it cost to paint a 3,000 sq ft house in San Diego" is specific enough that ChatGPT prefers the most extractable answer over the most generally authoritative source. A pillar page that answers the specific question with extractable passages, sourced cost ranges, comparison tables, and FAQ schema can displace the generic authority site.

Three things characterize successful Pattern C displacement:

Specificity over breadth. The pillar targets a specific buyer-decision question, not a generic category page. "How does the FAIR Plan work for Westside Los Angeles homeowners after the 2025 fires" is Pattern C. "Insurance information" is not.

Sourced specificity. Real numbers, real dates, real local context. Generic authority pages tend toward generic content; Pattern C wins by being concretely specific. A Pacific Palisades brokerage page with current median lot prices, named neighborhoods, dated permit data, and links to LADWP / FAIR Plan / city planning sources outperforms a generic "Pacific Palisades real estate" page from Realtor.com.

Comparison and decision structure. Pattern C pages that compare options (X vs Y), provide decision frameworks, and answer specific sub-questions outperform pages that simply describe a topic.

The Tony's Painting case study illustrates Pattern C cleanly: three pages on the site directly compare Tony's Painting vs CertaPro / Five Star / 360 Painting with sourced criteria. Those comparison pages displaced generic painting-contractor authority pages for buyer-decision queries in the San Diego market. The displacement worked despite Tony's Painting having far less brand authority than the national franchises being compared.

Why these patterns matter

The three patterns combine into a coherent strategic framework:

A2 is the foundation. Without directory presence in your vertical's A2 set, your brand is undercited regardless of other investment. Same-week priority.

A1 is the long-term ambition. Direct brand citation requires sustained investment across content, mentions, and entity signals over 12-24 months. The right target for AEO programs that can budget that horizon.

C is the tactical accelerator. Pattern C displacement pages can produce specific-query wins within 60-90 days of publishing, well before A1 status is achieved. They also reinforce A1 over time by adding brand-cited content to the corpus.

A complete AEO program runs all three patterns in parallel: A2 directory claims in week 1, Pattern C displacement pillars starting in week 4, A1 investment ongoing across content + mentions + entity for the long haul.

How to play each pattern

Playing Pattern A2

1. Identify the 3-5 dominant A2 directories for your vertical (see table above).
2. Claim profiles where they exist; submit new profiles where they don't.
3. Maximize profile completeness — bio, photos, services, reviews, sold/closed data where relevant.
4. Maintain consistency — same brand name, same NAP, same descriptions across all directories.
5. Re-audit quarterly to catch new directories joining the A2 roster or old ones falling out.

Time to first measurable effect: 2-4 weeks after profile claim. Time to peak effect: 2-3 months as the AI assistants' retrieval systems incorporate the new directory entries.

Playing Pattern C

1. Identify 5-10 specific buyer-decision queries in your category that current authority sources answer poorly. Look for queries with vague-or-generic top results, missing local context, or no good comparison content.
2. Build a pillar page per query that's structurally optimized for displacement: 134-167 word extractable opening, question-format H2s on every sub-question, sourced statistics with dates AND inline source links, declared author entity, FAQPage schema, comparison tables for X vs Y questions.
3. Publish with strong internal cross-linking to your other content + outbound links to authoritative sources.
4. Wait 60-90 days for the page to mature into citation behavior.

Time to first measurable effect: 60 days minimum. Pattern C is not a same-week play; it requires real content investment.

Playing Pattern A1

1. Establish foundational entity signals: Wikipedia or Wikidata entity, Knowledge Graph presence via verified Google Business Profile, consistent NAP, schema markup site-wide with Person author entity.
2. Build content citation geometry across the site: 40-point AEO standard applied to every key page.
3. Run sustained brand mention engineering: daily HARO/Connectively/Featured pitching, podcast guesting, trade publication contributed essays, Tier 1 publication outreach (Forbes, WSJ, NYT, Mansion Global, American Lawyer, vertical equivalents).
4. Publish original data quarterly — proprietary indexes, market reports, surveys.
5. Re-audit semi-annually to track A1 citation share growth.

Time to A1 citation in a category: 12-24 months of sustained investment from a cold start.

Combined strategy: A2 + C as the practical AEO core

For most mid-market brands, the A2 + C combination produces the best near-term return. A2 directory presence ensures the brand isn't invisible in ChatGPT's default citation behavior. Pattern C displacement pages capture specific high-intent queries directly.

The combined sequence:

• Week 1-2: Claim or complete A2 directory profiles. Same-day fix.
• Week 3-6: Build 3-5 Pattern C displacement pillars on the highest-value specific buyer queries in the category.
• Week 7-12: Site-wide Article schema rollout with declared Person author. Cross-link new pillars to existing content. Begin brand mention engineering (HARO/Connectively).
• Month 3-6: Re-audit. Expand A2 coverage to second-tier directories. Build supporting articles around each Pattern C pillar. Sustain mention engineering.
• Month 6-12: A1 ambitions become realistic. Pursue Tier 1 publication features, Wikipedia/Wikidata eligibility, original data publication.

The A2 + C combination is what we recommend in the prioritized remediation roadmap of most audits. It produces the most measurable lift in the 90-day window without requiring the 12-24 month patience of pure A1 investment.

What about Pattern B?

Pattern B is intentionally absent from this taxonomy. Earlier observations suggested a possible "Pattern B" — ChatGPT citing aggregator/comparison sites — but on closer examination it collapses into A2 (aggregator sites are functionally directories) or into Pattern C (when the aggregator is also engineered for displacement). We dropped Pattern B from the framework to avoid forced taxonomy.

The patterns we publish reflect what we actually observe consistently. We don't add patterns to fit a structure; we name patterns that show up reliably across audits.

Frequently asked questions

Are these patterns ChatGPT-specific or universal across LLMs?

Pattern A2 (directory citation) appears across ChatGPT, Perplexity, Claude, and Gemini consistently — different platforms favor slightly different directories but the pattern of citing authoritative directories holds. Pattern A1 (personal site citation) is most pronounced on ChatGPT, weaker on Perplexity (which leans more on Wikipedia/entity confirmation), variable on Claude. Pattern C (displacement via specific content) appears across all platforms when the page geometry is strong.

How do I tell which A2 directories matter for my vertical?

Run 20-30 category queries against ChatGPT yourself and track which sources get cited. The 4-7 sources that appear repeatedly are your vertical's A2 directories. We document this in the 10-Point AI Citation Audit Phase 0 vertical detection. The table in this article covers the most common verticals as of June 2026.

Can the A2 roster change?

Slowly. We've seen specific directories enter or leave preferred-roster status over 6-12 month windows. The roster is more stable than search rankings but not permanent. Quarterly re-audits catch shifts.

Is Pattern C worth the effort if my A2 presence is strong?

Yes. A2 covers generic category queries; Pattern C captures specific buyer-decision queries. The two address different parts of the buyer journey. Brands strong on A2 alone tend to capture early-funnel awareness but lose mid-funnel decision queries to better-engineered displacement content.

How does Pattern C compete with the Brandlight 70%→20% finding?

Brandlight's research showed Google's top-10 organic links and AI-cited sources overlap dropping from 70% to below 20%. Pattern C exploits this gap: a page that ranks position 8-12 on Google but has superior citation geometry can win AI citation while losing the Google click. This is why Pattern C pages often look like SEO underperformers (mid-page rank) but produce significant AEO upside.

---

Companion guides: The Two-Track Law: content vs entity signals · The 10-Point AI Citation Framework · How AI assistants decide what to cite.