Every law firm evaluating a GEO agency eventually asks the same question: what does the work actually look like? Not the framework diagram — the work. What happens in week one, who touches what, and what lands on the managing partner's desk at the end of the month. Most agencies answer with abstractions, which is exactly why firms deliberate for a year or more before hiring anyone in this category. This piece answers the question literally. It walks through the first 30 days of a Legal Torch engagement, week by week, using the VERDICT™ methodology — what gets measured, what gets fixed first, what gets written, what gets reviewed, and what the first PROOF™ Report contains. It is specific enough that a prospective client can picture the engagement before signing, and specific enough that a skeptic can audit the method against it. Both readings are welcome.
Why the first 30 days decide the next eleven
Month one of a GEO engagement has one non-negotiable job: establish the baseline. Every number that matters later — citation lift, prompt-coverage delta, share-of-voice trend — is a comparison against the Month 0 measurement. An engagement that skips the baseline and jumps straight to "optimization" is flying blind, and the client has no way to distinguish work that moved the numbers from numbers that moved on their own.
This is the reason the VERDICT pillars run in a fixed order — Visibility before Entity, Entity before Relevance, Relevance before Digital PR. The ordering isn't ceremony. Entity work changes what AI engines retrieve; content work changes what they quote; PR work changes what corroborates it. Run them out of order and each layer's effect contaminates the measurement of the last. Run them in order and every change is attributable.
The other thing month one establishes is the working rhythm: what the firm sends us, what we send back, who reviews what, and when the monthly report arrives. Law firms are process businesses. An agency that can't describe its own process in operational detail is asking the firm to extend a trust it hasn't earned.
Week one — the baseline: buyer-intent prompts, multiple engines, no assumptions
Week one is measurement. We build a prompt set of 50–100 buyer-intent questions tuned to the firm's practice mix and geography — the questions a prospective client actually types into ChatGPT, Perplexity, Gemini, or Google AI Overviews when they need what the firm does. The set follows a documented taxonomy: Who questions ("who's a good immigration lawyer in Irvine"), What questions ("what's a wrongful-termination case worth in California"), How questions ("how do I file an immigration appeal"), and Why questions ("why was my visa denied"). Each practice area gets its own cluster, because PI buyers, immigration buyers, and family-law buyers ask structurally different questions — a point covered in more depth in how ChatGPT picks which lawyers to recommend.
We run the full set across the engines and log three things per prompt: whether the firm is cited, in what position, and with what framing. We log the same for the firm's named competitors. The output is a citation share-of-voice score — the firm's share of citations across the tracked prompt set, by engine — plus an inventory of which prompts the firm loses and to whom. The numbers are granular by design: cited in 3 of 12 personal-injury prompts in the target city while a named competitor takes 9 of 12 is a finding a partner can act on; "your AI visibility is low" is not.
Week one also captures the entity-recognition snapshot: does each engine resolve the firm as a coherent entity, with its attorneys, practice areas, offices, and languages attached — or as a string of text it sometimes encounters? The distinction drives everything in week two.
Week two — entity consolidation starts before content does
Week two begins the highest-leverage repair work: making the firm and each attorney resolve as the same entity everywhere AI engines look. That means directory hygiene across Avvo, Justia, FindLaw, and Martindale-Hubbell; schema markup on the firm's site (LegalService, Person, FAQPage, BreadcrumbList); reconciling name, address, and practice-area data across every surface; and merging or retiring stale duplicate listings — the fragmented profiles that make an engine treat one firm as two ambiguous ones.
The reason this precedes content work is mechanical. AI engines pull from a curated corpus of authority surfaces, and they cite entities they can resolve. A firm can publish excellent content and still lose citations because the engine can't confidently connect the content's author to the entity being asked about. Fixing retrieval before display is the core argument of our piece on entity consolidation — in audits to date, firms with high domain authority and low AI citation rates almost always have an entity-resolution problem at the root.
Operationally, week two produces a written inventory: every surface where the firm appears, every inconsistency found, every fix applied or queued. The inventory becomes part of the month-one report — the client sees the fragmentation that existed at baseline and exactly what was consolidated, item by item.
Week three — the content-gap analysis and the prompt map
Week three connects the baseline to the firm's content. Every prompt the firm lost in week one gets mapped against the firm's existing pages: does content answering this question exist at all, does it exist but fail to answer within the first passage, or does it answer well but sit on a page the engines don't retrieve? Those are three different problems with three different fixes — write, restructure, or consolidate.
The restructuring work is passage-level. AI engines quote passages, not pages. A practice-area page that takes six paragraphs to reach "how much does an immigration consultation cost" loses to a competitor page that answers in one sentence and then elaborates. Week three produces the re-engineering queue: which pages get clear question-answer structure, which get schema-tagged FAQs, which get attorney-attributed statements that an engine can quote with a name attached.
Week three also scopes the citation-source work for months two and three — the placements on surfaces AI engines disproportionately retrieve from for legal queries. The target list follows the five citation sources every law firm should be in: the legal directories, Google Business Profile, and the state and local bar surfaces, plus the editorial surfaces (JD Supra and peers) that carry attorney-attributed commentary. Month one scopes and prioritizes this work; it does not rush placements out the door, because placements built on an unconsolidated entity waste their effect.
What the engagement asks of the firm
The other thing week one settles is what we need from the firm's side — and it is deliberately little, because partner time is the scarcest resource in any law-firm engagement. The intake list is specific: administrative access to the website (or an introduction to whoever maintains it), access to the Google Business Profile, credentials or delegated access for the firm's directory listings, the attorney roster with bar numbers and admission years, and thirty minutes with the managing partner to confirm the practice-area priorities and the competitor set worth tracking.
After intake, the recurring ask is bounded: one review touchpoint per week, typically under thirty minutes, where the firm's designated reviewer — usually the managing partner or the attorney who owns marketing — approves or redlines the outputs queued that week. Approval authority stays with the firm on every public-facing word. That is not a courtesy; it is how the compliance posture works. The firm's attorneys are the ones bound by the advertising rules, so the firm's attorneys hold the final pen.
What the firm should not expect to provide: content drafts, keyword lists, or hours of staff time chasing directory logins we can chase ourselves. Firms that have been through a traditional SEO engagement are often braced for a second job. Month one is structured so the firm's total time investment stays under four hours — intake included.
The compliance pass runs through all four weeks
Everything above passes through a compliance review before it ships — every schema description, every rewritten FAQ answer, every directory blurb, every attorney bio line. The review screens against ABA Model Rule 7.1 and the state-specific advertising rules in the firm's jurisdiction: superlative claims without verifiable third-party ratings, unsubstantiated specialization language, predictive outcome phrasing, testimonial-implying language, fee-quote phrasing that omits required context.
This is the pillar generalist GEO agencies don't offer — most don't know it's required. Law-firm web content is lawyer advertising under the California Rules of Professional Conduct, and AI-assisted content pipelines produce rule violations at volume precisely because the models are trained on promotional marketing copy from unregulated industries. The failure modes are documented in our piece on AI search and Rule 7.1; the operational answer is a named human reviewer, a written checklist, and a log.
Two things follow from running the review continuously instead of as a final gate. First, nothing accumulates in a pre-publication pile — flagged language is rewritten the week it's drafted. Second, the review log itself becomes a client asset: a dated record showing every output was screened before publication, retained for the firm's file. If the state bar ever asks, the firm has receipts. To be precise about roles — Legal Torch is not a law firm, and the compliance screen is a content review, not legal advice; the firm's attorneys decide and approve.
Week four — the first PROOF Report
Month one closes with the first PROOF Report — the same five-section format every retainer client receives every month, published openly on our methodology page as a redacted sample. Position: where the firm stands — citation share by engine, prompt coverage, entity-recognition status, at baseline. Receipts: everything shipped in the period — pages restructured, schema deployed, directory merges completed, inconsistencies fixed, itemized. Outcomes: what changed against baseline — in month one, typically the entity-layer fixes and early movement on branded queries, reported without inflation. Oversight: the compliance log — outputs reviewed, flags raised, resolutions, reviewer named. Forward: the hypothesis-driven plan for month two — what we expect to move, by how much, and what risk flags we're watching.
The format is designed for a managing partner to read in ten minutes without translating marketing language. The Forward section matters most: it states the next month's plan as testable expectations, which means every subsequent report can be checked against what the previous one predicted. That is the discipline that separates a methodology from a slide deck.
What month one deliberately doesn't do
Month one does not promise citation lift. AI engines re-crawl and re-resolve entities on their own schedules; directory changes and schema deployments propagate over weeks, not days. In engagements to date, the measurable movement tends to arrive in the 60–90 day window, after the entity layer settles and the restructured content starts getting retrieved — the published sample report shows the shape of that trajectory in a real engagement. An agency that promises first-month citation lift is either measuring noise or manufacturing it.
Month one also doesn't chase volume. No content calendar with twelve generic blog posts, no press-release syndication, no directory spam. The month is diagnostic and structural on purpose — because every later month measures against it, and because the work that compounds is the work done in the right order.
Bottom line
The first 30 days of a GEO engagement should produce three things: a defensible baseline across every engine that matters, an entity layer consolidated enough that later work is attributable, and a written report a managing partner can audit — including the compliance log. If an agency can't describe its month one at this level of detail, that is itself diagnostic information.
If you want to see what the baseline measurement looks like for your firm before committing to anything, the free AI Visibility Audit runs a compressed version of the week-one diagnostic — how the engines cite you today, and where the gap is. Month one starts with measurement either way. The firms that win AI search are the ones that know their starting point.
New to the topic? Start with the complete guide: GEO for Law Firms — what it is, what it costs, and what the citation data actually shows.
