You typed your practice area and city into Perplexity, read a confident three-paragraph answer with small numbered citations, and your firm was not one of them. A competitor was, or worse, a national directory was. Now you want to know whether there is anything you can actually do about it.
There is. Perplexity, ChatGPT search, Google AI Mode and the rest are retrieval systems layered on top of language models. They fetch pages, pick a handful that answer the question well, and write from those. The work is making your site one of the pages worth fetching.
Where does Perplexity actually get its answers?
Perplexity runs a live search for every query. It maintains its own index, and it fetches pages with a crawler that identifies itself as PerplexityBot. The model then reads the retrieved pages and writes an answer, attaching a citation to each claim that points back to the page it came from.
That means two things for a law firm. First, there is no separate "Perplexity ranking" to earn; if your page is crawlable and answers the question well enough to be retrieved, it is a candidate. Second, the citations skew toward pages that state a fact plainly near the top. A page that spends four paragraphs on the firm's history before mentioning the statute of limitations rarely gets pulled.
ChatGPT's search mode behaves similarly, using its own crawler (OAI-SearchBot) alongside Bing's index. Google's AI Overviews and AI Mode draw from Google's index, so the ordinary ranking factors still apply there. We cover Google separately in our guide to AI Overviews and ChatGPT in how to get your firm recommended by ChatGPT.
Can these crawlers even read your site?
Before worrying about content, check the plumbing. Open your robots.txt (yourfirm.com/robots.txt) and look for rules that disallow PerplexityBot, OAI-SearchBot, ClaudeBot or GPTBot. Some security plugins and hosting firewalls add these blocks by default, and a blocked crawler cannot cite you.
Many law firm sites render the practice area content with JavaScript after the page loads. Search crawlers handle this unevenly, and AI crawlers handle it worse. Load a page with JavaScript disabled; if the body text disappears, that is the first fix.
Speed matters too, because crawlers give up. A page that takes eight seconds to respond is often skipped in favor of one that responds in one. If your site is slow, site speed work comes before any AI-specific effort.
What does a citable page look like for "car accident lawyer in Tampa"?
Pull up the Perplexity answer for your own target phrase and open every citation. You will usually find a mix: one or two firm pages, a legal directory, a state bar or court page, and sometimes a news article. The firm pages that made it almost always share a shape.
The heading matches the question. The first paragraph gives the answer in two or three sentences (what the deadline is, what the fee structure is, what the first step is). Then the page gets specific to the jurisdiction: the Florida statute number, the county courthouse, the PIP threshold. Then it says who the firm is and how to reach them.
The typical practice area page opens with "If you or a loved one has been injured..." and never states a fact a model could quote. Rewriting your ten most important pages in that shape is the highest-return task on this list. The AEO for lawyers page walks through that structure in more detail.
Which third-party sources appear next to you?
AI search engines lean on aggregators because aggregators are consistent. Avvo, Justia, FindLaw, Super Lawyers, the state bar directory and Google Business Profile all get cited for "best" and "top" queries where the engine wants a list rather than a single firm.
You cannot rewrite those sites, but you can make sure your entries on them are complete, current and consistent with your website: same firm name, same address format, same practice area language, same attorney names. When the model sees the same facts about your firm in six places, it treats them as reliable. When your site says "Smith Injury Law" and Avvo says "Law Offices of John Smith, P.A.", it has two weaker entities instead of one strong one.
Reviews count here as well. Engines summarizing "which firm should I call" often quote review themes. A steady flow of detailed Google reviews gives them material, which is why reputation management belongs in the same project.
How do you know if any of it is working?
Perplexity does not offer a Search Console equivalent, so measurement is manual and directional. Keep a list of twenty questions your clients ask, run them monthly in Perplexity, ChatGPT and Google AI Mode, and record whether your firm is cited and which page.
In your analytics, add a segment for referrals from perplexity.ai, chatgpt.com and copilot.microsoft.com. The numbers will be small, but the trend tells you whether the rewriting is paying off. Google Analytics for law firm SEO covers how to set up that kind of segment.
Server logs are the other source of truth. If PerplexityBot has never requested your practice area pages, no amount of content work matters until it does. Some firms also publish an llms.txt file as a low-cost signal; we explain its real status on that page.
Questions we get
Does Perplexity favor big national sites over local firms?
It favors pages that answer the question, and national sites have more of those. For jurisdiction-specific questions, a local firm page with the actual county procedure often beats a generic national page, because the national page cannot supply that detail. Aim your content where the big sites are vague.
Should we block AI crawlers to protect our content?
That is a business decision, but understand the trade: blocking PerplexityBot or GPTBot removes you from those answers entirely. Most firms we talk to want to be cited more, not less. If you block anything, block training-only agents and leave the search agents open.
Can we pay to appear in AI answers?
Perplexity has run sponsored placements and other engines are testing ads, but the cited sources in the organic answer are not for sale. Anyone promising guaranteed placement is promising something the platforms do not offer. Nothing on this page is a guarantee either; it is what makes citation more likely.
If you want a second set of eyes on how your pages read to these engines, tell us what you are working with.