Someone at the firm opened a keyword tool, typed "personal injury lawyer Denver," saw a number, and decided that was the target. Two years later the firm ranks eleventh for it, the pages are all written around it, and nobody can explain why the phone rings less than expected. Meanwhile "car accident lawyer no insurance Colorado" and "should I talk to the adjuster after an accident" bring calls to a competitor who never chased the big phrase.
Keyword research for a law firm is the work of finding out what prospective clients actually type, in what order, and with what in mind, and then deciding which of those phrases each page should own. Done well, it produces the site plan. Done badly, it produces a list of phrases with volume numbers next to them. Here is how we approach it within our SEO for lawyers work.
The mistake: starting from "lawyer" plus your city
"[Practice area] lawyer [city]" is the phrase everyone targets, so it is the one where you face every competitor, every directory, and every lead-generation site at once. It is also not how most clients begin. Most begin with a situation ("my landlord is keeping my deposit"), then a question ("can a landlord keep a deposit for painting"), and only then, if at all, a lawyer search.
That does not mean the head phrase does not matter. It means it should be one target among many, owned by the practice page, while the situation and question phrases are owned by dedicated pages that funnel to it.
Where the real phrases come from
Keyword tools (Semrush, Ahrefs, Google Keyword Planner) are useful for confirming and sizing, not for discovering. Discovery comes from four sources closer to the client.
First, intake notes and consultation questions. Every "how do I," "can they," and "what happens if" a prospective client asks is a search phrase. Second, Google Search Console for your own site, filtered to queries with impressions but low clicks; these are phrases Google already thinks you are relevant for. Third, Google's own interface: the autocomplete suggestions, the "People also ask" boxes, and the related searches at the bottom of the page for each head phrase. Fourth, the practice area's forums and communities (Reddit legal advice threads, Avvo questions), where people describe their situation in their own words before a lawyer reframes it.
Sorting phrases by intent, stage, and device
Once you have a few hundred phrases, sort them by three things. Intent: is the person looking for information, looking for a lawyer, or looking for a specific lawyer by name? Stage: is this before an event, right after it, or during a matter? Device and moment: is this a desktop search by a business owner during the workday, or a phone search by someone in a parking lot?
Then look at the actual search results for each phrase before committing. If the first page is entirely Nolo, state court self-help pages, and national directories, a local firm page may not break in, and the phrase is better handled as a section within a broader page. If the first page shows local firm pages and a map pack, the phrase is winnable.
Mapping phrases to pages, and retiring the rest
The output of keyword research is not a list. It is a map: each page on the site, the one primary phrase it owns, and the handful of related phrases it should also capture. No two pages should own the same primary phrase, because they will compete with each other and both lose.
Practice pages own the head phrases and their close variants (DUI lawyer, DWI attorney, drunk driving defense). Question pages own the situation and question phrases. Local pages own the court and place names. Attorney bios own the attorney names. This mapping becomes the content plan; our page on law firm content strategy describes how it turns into a publishing schedule and a review process.
Phrases that fit no page get one of three treatments: a new page is planned, the phrase is folded into an existing page as a section, or it is retired because the results show it is not winnable or the searcher is not a client.
The phrases you cannot see: "near me" and AI prompts
Two categories of searches do not show up cleanly in keyword tools. "Near me" searches (and their unspoken cousins, where Google infers location without the searcher typing it) depend on the searcher's position and the firm's Google Business Profile far more than on page content. The research step here is checking the map results from several points in your service area, using a rank-tracking tool with a location grid or simply a phone in different neighborhoods. Our page on ranking for "lawyer near me" searches explains the factors involved.
The second category is questions typed into ChatGPT, Gemini, or Perplexity. There is no volume data, but the questions are the same ones from your intake log, phrased more conversationally and often longer. Ask those tools your own head phrases and situation questions and note which firms and which sources they cite. That tells you what content the tools are drawing from and what your pages need to contain. Our page on how to get your law firm recommended by ChatGPT covers what to do with that.
Questions we get
How often should we redo keyword research?
The initial map is a project. After that, a quarterly review of Search Console (new queries appearing, pages losing position) and a look at the intake log for new questions is enough. Practice areas shift with new laws and court decisions, so a major change in your area is also a trigger.
Do we need paid tools?
Not at the start. Search Console, Google's autocomplete and related searches, and your own intake notes give most firms more than they can act on. A paid tool becomes useful when you want to size a list, check competitors' pages, or track positions across many phrases.
Should we target phrases in Spanish or other languages?
If your clients search in those languages, yes, and the research process is the same, with one addition: a native speaker should review the phrases, because machine translation of legal terms rarely matches what people actually type.
If you want a keyword map for your practice area rather than a list, tell us what you are working with.