How Perplexity Decides Which Businesses to Cite

Most AEO advice treats Perplexity, ChatGPT, and Google AI Mode as interchangeable. They're not -- and nowhere is that gap more expensive than in how you optimize for Perplexity specifically.

Perplexity is a real-time search engine that synthesizes answers from live web results. Every query triggers a fresh retrieval pass across the web. What that means for local businesses is very different from what it means for businesses chasing ChatGPT citations -- and the fix sequence looks almost nothing like the generic "create listicle content" advice that dominates this space.

Perplexity Is a Search Engine That Answers, Not a Knowledge Base

ChatGPT's local citation behavior is largely determined by training data and structured data partnerships -- Foursquare, Yelp, Thumbtack -- that feed its business database. When you optimize for ChatGPT, you're mostly fixing your presence in databases that were assembled months or years ago.

Perplexity works differently. It fetches approximately 10 pages per query and cites 3 to 4 of them. When someone asks Perplexity for a local plumber or dentist, Perplexity runs a live web search and synthesizes from whatever it retrieves. The structural implication is significant: a fix you make to your website or directory listing today can show up in Perplexity citations within days, not the 30-60 day window you'd expect from platforms that rely on training data updates.

This also means that what Perplexity cites for local queries looks nothing like what ChatGPT cites. The question is: what does Perplexity actually retrieve?

What 25,000 Citations Show About Local Services

Most AEO content assumes that listicle-formatted pages dominate AI citations because that pattern holds for general commercial queries. For local services, that assumption is wrong.

In our July 2026 methodology rec (filed 2026-07-24, based on DeltaV Digital's citation study of 25,337 AI engine citations tracked from April to July 2026), the data showed that homepages account for 55% of AI engine citations for local services queries. For Perplexity specifically, a separate local business query study found that Perplexity cites business websites directly 73% of the time for "best [service] in [city]" queries.

The comparison format breakdown makes the divergence clear:

| Query type | Top cited page type | Perplexity's local services pattern | |---|---|---| | B2B technology | Listicles (61% of citations) | Not applicable | | Local services | Business homepages (55%) | Direct business site 73% of the time | | Informational | FAQ / how-to content | Varies |

This is not a minor nuance. The standard AEO fix plan says "create listicle-formatted service pages." For a local plumber, HVAC contractor, or dentist trying to appear when someone asks Perplexity for a recommendation, building listicles is largely wasted effort. The optimization target is your homepage.

The LLMClicks confirmation of the same pattern (filed in our July 2026 investigation): for local services queries, first-party websites and business listings account for 86% of citations, while Reddit represents just 2%. This is the inverse of Perplexity's general citation behavior, where community discussion sites typically rank second behind news and editorial sources.

What "Homepage Optimization for Perplexity" Actually Means

If Perplexity cites your homepage directly 73% of the time for local services queries, the question shifts from "what content should I create" to "can Perplexity actually read and extract value from what's already on my homepage."

Based on the 2026-07-24 methodology analysis, the variables that determine whether Perplexity can use your homepage are:

**Structure comes first.** Your primary service and city need to appear in the first 100 words. Perplexity retrieves and chunks content, and a homepage that buries its core offering below a hero image and a tagline about "excellence" does not produce a citable answer to "best plumber in [city]."

**Question-form H2s.** Headers structured as answers to likely queries ("24/7 Plumbing Services in Burlington" performs better than "Our Services") give Perplexity's retrieval a direct extraction target.

**LocalBusiness schema.** Schema markup on the homepage -- not just a separate contact or about page -- gives Perplexity's indexer structured signals about service type, service area, and location.

**PerplexityBot allowed.** Robots.txt must not block PerplexityBot. A surprisingly common fix for businesses whose homepage content is solid but whose Perplexity citations are thin. JavaScript rendering blocks are also a common culprit -- if your homepage content only appears after JavaScript executes, Perplexity's crawler may not see it.

Directory presence remains secondary but non-negotiable. Our 2026-04-22 investigation (directory-presence-impact.md) confirmed Perplexity uses Yelp as a primary review source in every industry tested. For local businesses, a claimed and complete Yelp profile fills the cases where Perplexity doesn't cite your homepage directly -- the remaining 27% of local services queries where roundup pages and directories dominate the citation pool.

Why Perplexity Scores Change Faster Than Any Other Platform

Even after fixing your homepage and directory presence, Perplexity citation scores are inherently more volatile than other platforms. The SISTRIX citation drift study (April 2026, 82,619 prompts and 1,548,213 snapshots across 17 weeks) measured source rotation across platforms.

From our methodology rec filed 2026-07-30, the headline numbers:

- Google AI Mode rotates 56% of its source pool weekly - ChatGPT Search rotates 74% of its sources weekly - Brand domains (the stable anchor category) hold across all 17 weeks in only 43% of cases - Co-citations -- the surrounding sources next to cited brands -- show 89% weekly rotation

Perplexity's live-retrieval architecture means it inherits the volatility of web search ranking plus its own retrieval behavior on top. A business that was cited by Perplexity last Tuesday may not be cited this Tuesday through no fault of its own infrastructure -- query phrasing, competing content indexed since the last check, and platform-level algorithm shifts all affect individual results.

The practical implication: a single audit point or a weekly Perplexity citation check is a sample from a volatile distribution, not a stable measurement. Consistent Perplexity presence comes from consistent infrastructure signals -- homepage structure, directory accuracy, schema health -- not from chasing any single citation result.

A Data Integrity Problem Affecting Perplexity Scores Right Now

If you're using a third-party AEO monitoring platform to track your Perplexity visibility, there's a specific reason to be cautious about data from October 2026 onward.

Perplexity retired its Sonar Chat Completions API on September 27, 2026. Any AEO tool using the old Sonar endpoint to run Perplexity citation checks now receives an HTTP 403 error (`agent_api_migration_required`) instead of citation data. The new endpoint is the Agent API, which uses a different response structure -- citations appear in `output[]` rather than the previous format.

Our research vault tracked this edge case across 12 consecutive daily checks from September 27 to October 8, 2026 (edge-cases/perplexity-sonar-api-retirement-2026-09-19.md). The findings relevant to business owners:

Tools using the old citation parser on the new Agent API endpoint will silently return empty citation arrays. No error surfaces. The tool appears to work, but every Perplexity citation check returns zero.

As of October 8, HubSpot's AEO Grader -- which explicitly uses Perplexity as one of its three scoring models and has 205,000+ AEO platform customers -- had published no migration announcement. Twelve days post-retirement, no changelog entry, no blog post, no product announcement.

We found no widespread reports of AEO tools breaking across the 12-day monitoring window, which suggests most major platforms migrated before the deadline or use API gateways that handled the transition transparently. But the absence of public reports doesn't confirm successful migration. A tool can silently return zero Perplexity citations -- accurately reflecting its own failure to retrieve data, not your actual visibility -- without any error or alert surfacing to the customer.

If a platform-provided Perplexity score looks anomalously low after September 27, ask the vendor to confirm Agent API migration. A manual check on perplexity.ai for your category and city takes two minutes and tells you what the actual Perplexity results show, independent of any tool's API integration status.

Where to Start

For most local businesses, the priority order based on our July 2026 DeltaV and SISTRIX research:

Fix your homepage structure first. Direct answer to your primary service query in the first 100 words, question-form H2s, LocalBusiness schema, PerplexityBot not blocked. This addresses the 55-73% of Perplexity local citations that point directly at business homepages.

Claim and complete Yelp. It appears in every industry tested in our directory research. For Perplexity, it fills the non-homepage citation cases.

Verify your monitoring tool's data if you're checking Perplexity scores from any API-based platform. A manual perplexity.ai check for your category and city is the ground truth.

Sourcepull's Signal Check runs a live diagnostic across ChatGPT, Perplexity, Gemini, and Claude and shows you exactly which platforms are citing your business and from what sources -- without relying on a third-party API integration. If you want a baseline that isn't subject to the Sonar migration data gap, it's the right place to start.