Most SEOs have a keyword list and a judgment call. They look at the top two or three ranking pages, decide they understand the intent, and build something. Sometimes that works. More often, it produces a page that targets the right topic with the wrong format, a service page where Google is rewarding a guide, a listicle where Google is rewarding a comparison table, and the page never ranks regardless of how well it’s written.
AI doesn’t fix this automatically. Most people are using it for the wrong job, asking it to write content instead of asking it to read the SERP.
This article is the workflow that does it correctly, including the actual prompts, applied to the kind of service business keyword sets CFM works with every day.
The problem is not knowing what search intent means. It is diagnosing it accurately across a full keyword set, consistently, without reading twenty SERPs by hand.
Most operators do this manually. They look at one or two competitor pages, make a judgment call, and move on. For unambiguous queries like “personal injury attorney Los Angeles,” that judgment is usually correct. For everything adjacent, it fails.
Take two queries from the same firm, same vertical: “lemon law attorney” and “do I have a lemon law case.” The first is commercial. A user knows what they need and wants a firm to call. The second is informational. A user suspects they have a problem and needs to understand it before they consider hiring anyone. Court House Lawyers, a personal injury firm CFM works with in Los Angeles, has pages targeting both. Building a service page for the second query, the way a firm would for the first, means competing on the wrong playing field. Google is not rewarding service pages for “do I have a lemon law case.” It is rewarding guides that explain the criteria.
Without a system, intent gets guessed. Guessed intent produces pages that do not rank, and at scale, pages that do not rank compound into index bloat, diluted crawl budget, weakened authority, and a site full of content that never converts.
The fix is applying search intent mapping as an extraction process, not a judgment call. Here is how to use AI to run it.
This is SERP analysis at scale, not keyword research. The goal is to understand what Google is currently rewarding for a specific query; the format, the structure, the question being answered before you write a single word.
Open your LLM of choice. Claude and ChatGPT both work well for this. Pull the top five to eight URLs ranking for your target query. You do not need to scrape the full content. The URL, page title, and meta description are often enough for the LLM to classify the format. For ambiguous queries, paste the introduction or first two sections.
Use this prompt verbatim or adapt it to your context:
You are an SEO analyst. I am going to give you a list of URLs and the query they rank for. For each URL, tell me: (1) the content format — guide, service page, FAQ, comparison, listicle, or other; (2) the primary intent the page is serving — informational, commercial, transactional, or navigational; (3) any secondary intent the page also addresses; (4) what a new page targeting this query would need to do differently to compete with the existing results. Query: [paste query]. URLs: [paste URLs].
The output gives you a format map for that query. Not a guess. A pattern extracted from what is already ranking.
This is one of the tasks AI handles reliably in an SEO workflow, pattern recognition across structured inputs. The interpretation of what to do with the output still requires human judgment.
Once you have the SERP reading method, scale it. Export your keyword set from Google Search Console; the queries report gives you real data on what your site is already being served for. Combine that with your target keyword list. Paste the full set into your LLM.
Use this prompt:
Here is a list of [X] keywords for a [personal injury law firm / dental practice / service business] in [city]. For each keyword, tell me: (1) the primary intent — informational, commercial, transactional, or navigational; (2) the most likely content format Google is rewarding for this query — guide, FAQ, service page, comparison, or other; (3) whether the intent is ambiguous or split across two types. Output the results as a table with four columns: keyword, intent, format, and ambiguity flag.
The table output is non-negotiable. A paragraph of observations is not actionable. A table is. You can sort by intent type, filter for ambiguous queries, and immediately see which keywords need informational content, which need service pages, and which are hybrid.
One practical note: navigational intent queries — branded searches, searches for a specific location, or contact details — rarely require new content. They resolve to Google Business Profile optimization or directory presence, not a blog post. The classification step surfaces this before you spend time writing pages that should not exist.
The classification output is only useful if you act on it. This is the step most operators skip. They classify intent correctly and then build the wrong format anyway, usually because their content process defaults to the same template regardless of what the classification says.
The mapping is direct:
Informational + guide format — long-form structured article with H2 questions, direct answer paragraphs, FAQ block at close. The user has a problem to understand, not a vendor to hire. Do not put a sales section above the fold.
Commercial + comparison format — structured table or side-by-side breakdown. The user is evaluating options. Give them the comparison they came for, then contextualize why your offer wins.
Transactional + service page — short, specific, CTA-forward. The user is ready to act. Every element that delays that action is friction.
Hybrid — a guide with a commercial section embedded. The user is research-first but open to conversion if the right moment arises. Put the informational content first, the commercial offer second.
The Court House Lawyers example is concrete: “lemon law attorney Los Angeles” is commercial intent, service page format. “Do I have a lemon law case” is informational intent, guide format. Same firm, same vertical, opposite format requirements. Building both incorrectly or building one and ignoring the other leaves significant visibility on the table.
For law firm SEO specifically, this distinction matters more than in most verticals because the research-to-hire journey is long and trust-dependent. The informational pages do the early relationship work. The commercial pages close it.
Google’s helpful content guidance frames this as content that “serves the searcher’s actual goal” — the format question is not a technical preference, it is the practical expression of whether the page serves the query or serves the publisher.
For service businesses, how local service pages should be structured follows the same format logic — the intent classification determines the structure before any copy is written.
After classifying your keywords and reading the SERP, there is one more question to ask: what intent is present in this query that none of the top-ranking pages fully serve?
This is where intent analysis shifts from defensive — building the right format to compete — to offensive. Secondary intent gaps are citation and ranking opportunities. A page that cleanly answers a question the top five results all leave partially unanswered becomes a citation target for both traditional ranking and AI-generated answers. In generative engine optimization, a page that uniquely serves an underserved intent segment has a disproportionate chance of being cited precisely because no other source has covered it clearly.
Use this prompt after completing the SERP analysis in Step 1:
Based on the SERP analysis above, identify: (1) any intent signals present in the search query that none of the top five results directly address; (2) any content format types absent from the SERP that could serve a meaningful segment of searchers; (3) specific questions a user searching this query likely has that none of the top results answer explicitly. List each gap as a one-sentence opportunity.
The AI-first SEO strategy article we published identified its content gap through exactly this process: no competing article included a verified, named client AI Mode citation. Every competitor was writing theoretically. That absence was a gap. Filling it with a real result is what makes the page differentiated rather than derivative.
Two additional filters to apply after the gap analysis: does filling the gap build topical authority in a category you are trying to own, or does it scatter your content footprint? And does the page you would build have the digital PR and link authority to be cited once it ranks? Intent alignment and content quality are necessary but not sufficient. Off-site authority is the third variable. The gap analysis is incomplete without acknowledging it.
Pages that slip in intent over time, drifting from what the SERP now rewards, also show up through this process. That is content decay, and the same workflow that identifies new gaps identifies existing pages that have lost their intent alignment.
The workflow above is repeatable. It takes roughly forty-five minutes to run properly for a single keyword, including the SERP pull, the classification pass, and the gap analysis. For one article, that is a reasonable investment. For a content program covering eight city pages, six practice areas, and twenty supporting blog articles, it becomes the ongoing operational layer that most businesses cannot maintain consistently.
The constraint is not capability. Most practitioners who read this can run it. The constraint is time and discipline running it every time, for every page, without shortcuts, across a full site.
CFM operationalizes this workflow through the Authority Engine™—an integrated lead generation SEO system for every client engagement. Every page that goes into production has been through the intent classification, the format mapping, and the gap analysis before a brief is written. That includes the internal linking structure that connects every piece to every other piece in the content cluster.
If you are managing a single site with a contained keyword set, the four steps above are enough to run this yourself. If you are managing a service business with multiple locations, practice areas, and ongoing content needs, the system matters more than the workflow.
If the workflow above reveals more gaps than you have time to fix, that is exactly what the Authority Engine™ is built for.
© Curious Fortune Media – All Rights Reserved – Terms and Conditions – Privacy Policy – Accessibility Statement
