How Perplexity Picks Local Businesses (and Why It's Different)

Perplexity has a fraction of ChatGPT's users. It is still the assistant most worth studying, for one reason: it shows its work.

Every Perplexity answer lists the exact pages it read. For a local business owner trying to understand why an AI recommends a competitor, that citation list is the most useful free diagnostic available.

How Perplexity works

Perplexity is a search engine wearing an assistant's interface. Every question triggers a live search. It retrieves pages, reads them, and writes an answer with numbered citations pointing back at the sources.

It does not lean on training data for factual questions the way ChatGPT does. There is very little "what the model remembers" in a Perplexity answer about local roofers. There is almost entirely "what these six pages said."

Three consequences that matter:

It moves fast. No training cycle to wait for. A page that starts ranking today can be cited today.

It is fully attributable. You can see exactly which pages produced the recommendation.

It inherits search rankings closely. What ranks is what gets read. Classic SEO transfers to Perplexity more directly than to any other assistant.

What Perplexity actually reads for a local query

Run a local service query and read the citations. The pattern is consistent across cities and trades:

1. Directory and aggregator pages — Yelp, Angi, Thumbtack, HomeAdvisor, BBB, trade-specific directories
2. "Best X in [city]" roundup articles — local blogs, news sites, content-farm listicles
3. Google Business Profile data, usually surfaced through a map or directory page
4. Individual business websites — typically only a couple, and often only the ones that rank for the exact phrasing
5. Reddit threads, when the query resembles how a person would ask another person

Notice the ordering. Your own website is usually the fourth thing on that list, sometimes absent entirely.

This is the single most important thing to internalize about Perplexity: your website is rarely the source of your Perplexity recommendation. The directories and roundups are.

The three levers that actually move Perplexity

1. Get onto the pages it already reads

Run the query yourself. Read every citation. For each one, ask: am I on this page, and how am I presented?

If Perplexity is citing a "Top 10 Plumbers in Sacramento" post on a local blog, and you are not on it, that is a concrete, addressable gap. Contact the author. Many of these roundups are maintained by small local publishers who update them.

If it is citing Angi and your Angi profile is a stub, fill it in. If it is citing BBB and you are unaccredited with no profile, that is a fixable afternoon.

This is unglamorous work and it beats every clever tactic.

2. Make sure PerplexityBot can read your site

Perplexity operates PerplexityBot for indexing plus a user-triggered fetcher for live questions. Check your robots.txt. Blocking it is a real and common own-goal — several popular WordPress security and "AI protection" plugins block AI crawlers by default, and nobody told the business owner.

Also confirm your content exists in the raw HTML. Perplexity fetches and parses; it does not run a full browser for every page.

3. Rank for the actual phrasing people use

Because Perplexity mirrors search results closely, ordinary SEO on conversational phrasing works unusually well here. Not "plumbing services Sacramento" — "who to call for a burst pipe in Sacramento," "how fast can a plumber get here Sacramento," "cost to replace a sewer line Sacramento."

Those long conversational queries have low competition and map directly onto how people talk to an assistant.

Using Perplexity as a research tool

Set aside the optimization for a moment. Perplexity is the best free competitive research tool a local business has.

Ask it:

• "Best [your trade] in [your city]" — read every citation
• "Who should I call for [specific problem] in [your city]"
• "Is [your business name] any good?" — see what it finds and what it gets wrong
• "[Competitor name] reviews" — see what the internet says about them that it does not say about you

Twenty minutes of that gives you a clearer picture of your local digital footprint than most paid audits.

What to do this week

1. Run your top three customer questions in Perplexity. Open every citation.
2. Make a list of every directory or roundup cited where you are absent or thin.
3. Check robots.txt for PerplexityBot.
4. Fix the two most-cited directory profiles first.
5. Ask Perplexity "is [your business] any good?" and note every factual error it makes. Those errors exist somewhere public.

The wider picture

Perplexity is one of five systems answering your customers, and it is the most forgiving. The others weight training data and brand familiarity more heavily, which makes them slower to move and more dependent on off-site reputation.

Our free scan tests all five at once — twenty-five real customer questions, every verbatim answer shown, every competitor named in your place listed. Free, about fifteen seconds, no account.