Law Firm AI Search Visibility Buyer’s Guide: GEO and AEO

An agency sends a screenshot showing your firm in an AI answer. It may be useful. It is not, by itself, a reason to buy a program.

First ask what the screenshot shows. Was your firm merely mentioned? Was your page visibly cited? Did the answer recommend the firm for a particular matter in a particular market? Did the person visit your site, contact the firm, pass intake, and become a client? Those are different events, owned by different systems, and they call for different work.

GEO and AEO are common labels for work around AI-search visibility. The label does not tell you whether a provider is proposing a factual-content audit, technical access review, citation monitoring, content production, a public-relations effort, or a reporting dashboard. Buy the work behind the label.

This guide gives an owner or marketing leader a way to compare those proposals without treating an AI appearance as a case result.

Start with a decision, not a platform list

Choose one business question before you choose tools. For example:

• A personal-injury firm wants prospective clients to find accurate answers about a narrow case type before they choose whom to call.
• A business-law firm wants its practices, attorneys, and offices represented accurately when a founder compares counsel for an acquisition.
• A growing firm wants to learn whether AI-assisted research is producing identifiable visits that the intake team can classify.

Each objective needs a different page set, question set, and measurement path. “Improve our AI visibility” is too broad to evaluate a provider or a pilot.

Write the decision down in one sentence: For [practice] in [market], we want to learn whether better public information helps people researching [specific task] reach a page or contact path that the firm can evaluate. The provider should be able to show how its work relates to that sentence.

Separate the evidence before someone rolls it into an “AI leads” number

The table below is a useful reality check during a sales call.

Google's current guidance says its generative Search features use core Search systems and can retrieve relevant pages from its Search index. It also makes clear that meeting eligibility requirements does not guarantee indexing or serving. Google's AI-search guidance is useful for evaluating claims that a vendor can guarantee citations.

Do not let a report combine screenshots, citation counts, referral visits, consultations, and retained matters into one number. Ask for each metric's definition, source, date range, coverage, and known gaps.

Ask the provider to show the baseline before proposing the cure

A credible baseline is more than a collection of favorable prompts. It should preserve:

1. The practice, market, and prospect task being tested.
2. The platform or reporting source and the date checked.
3. The exact or carefully grouped question set, including nonbranded questions.
4. Location, language, account state, and other conditions that could affect an observed answer.
5. The answer, cited page, or exported data that supports the observation.
6. Negative and ambiguous results, not just wins.

Suppose a Phoenix employment-law firm wants to know whether people researching severance agreements encounter useful firm information. A reasonable baseline might include a limited, documented set of research questions; the current related service and attorney pages; and any identifiable visits or intake comments the firm already records. It would not claim that every person asking a related question sees the same answer.

If a provider changes the question set during a pilot, it should retain the old version and explain why. Otherwise, a report can look better simply because it added easy branded questions.

Compare deliverables, not acronyms

Ask a prospective GEO or AEO provider which of these jobs it will actually do and who owns each one.

Monitoring does not automatically include implementation. Content production does not automatically include technical access work. A public-relations project may produce credible outside material but cannot control whether any AI system cites it. Make the proposal say which work is included, excluded, and dependent on the firm.

Check platform claims at the primary source

Current product language changes quickly. Treat a vendor's slide deck as a starting point, then ask for the official documentation behind a platform-specific assertion.

If the provider claims its own monitoring, scoring, or prediction system is accurate, ask what task was evaluated, which test set was used, what counted as an error, how large the sample was, which cases were excluded, and when the evaluation last ran. A favorable screenshot demonstrates an output; it does not validate the system.

Google now gives site owners a Search Console control for inclusion in supported generative Search features. The default can include the site's links and content; exclusion prevents those features from using or displaying them, but does not block ordinary Search or control AI training. The control's effect and inheritance are documented by Google, and its setting for Juris or any other site should be inspected before anyone changes it. Read Google's control documentation.

Google also says it does not require special AI files, special markup, tiny content chunks, or pages for every possible prompt variation. A provider may still recommend a clearer page structure for readers, but it should not sell those tactics as a Google requirement. Google's guidance explains the limits.

For Bing, AI Performance reports aggregated citation activity across supported AI experiences. It labels grounding queries as grouped phrases rather than full user prompts, and it says citation activity is not a ranking, authority, traffic, or quality measure. Its newer intent, topic, citation-share, and compare views can help a team see patterns; they do not explain why a number changed. Read Bing's documentation.

OpenAI distinguishes OAI-SearchBot, GPTBot, and ChatGPT-User. They have different documented roles. A provider should identify the relevant control and the actual business consequence instead of treating every crawler setting as one decision. Read OpenAI's crawler documentation.

Judge content work by the uncertainty it resolves

Useful work makes a prospective client's next decision easier. It might correct an outdated office or practice fact, clarify the kinds of matters a firm handles, explain an attorney's experience without exaggeration, answer a recurring process question, or add original research with its method and limits.

Ask for a before-and-after sample. What was unclear? Which source supports the revised claim? Who reviewed it? Would the page still help a potential client if no AI system ever cited it?

That last question matters. Google's current guidance favors content that is useful, distinctive, and written for people. The same page should earn its place through a real reader task, not because it repeats small variations of an AI prompt.

Juris Digital's law-firm SEO service is the relevant commercial context for this work. The service page should not be read as a promise of a separate GEO product, a guaranteed citation count, or a fixed outcome. When the constraint is the site's information or user path, a website design and rebuild or legal-content writing decision may be more useful than a new monitoring subscription.

Price the workload, including the work your firm must do

There is no meaningful universal GEO price. Compare like with like:

• baseline research and monitoring frequency;
• number of practices, markets, pages, and factual records in scope;
• editorial research, drafting, attorney review, and publishing;
• technical/development changes and any approvals they depend on;
• software access, exports, and reporting time; and
• the firm's internal time for facts, reviews, intake feedback, and decisions.

Ask for setup, recurring, editorial, technical, and outside-placement costs separately. Ten revised pages, fifty monitored questions, and a monthly strategy call are not interchangeable units.

Also ask what you retain if the relationship ends. The firm should be able to keep approved copy, source records, change history, appropriate platform access, and exports needed to understand what happened. A dashboard alone is not a handoff.

Work through the buying decision before signing

Return to the hypothetical Phoenix employment-law firm. Assume it has a working website, one attorney who can review four hours of material each month, and a three-month test budget capped at $24,000. Its immediate problem is not general brand awareness. The firm wants to correct inconsistent severance information, understand how five existing pages appear for a defined set of executive-employment questions, and learn whether any identifiable visits become qualified inquiries.

Two hypothetical providers submit proposals:

The cheaper proposal is not automatically the better buy. Provider A may fit a firm whose pages, facts, analytics, and intake are already sound and whose only missing capability is monitoring. It does not fit this firm's stated problem because it cannot correct the inconsistent sources or test the client path.

Provider B fits more of the objective and remains under the cap, but it still needs two corrections before approval. “Up to three revisions” must identify who publishes them and what happens if attorney review exposes a fourth critical page. The separately estimated technical work needs a not-to-exceed approval step so the pilot cannot exceed $24,000 without a new decision.

The firm should choose Provider B only after those responsibilities are written into the scope. If the provider will not clarify them, the firm should not substitute enthusiasm about GEO for the missing controls.

The arithmetic is hypothetical and the prices are not Juris Digital pricing or market benchmarks. The point is to normalize the work against the firm's problem, internal review capacity, budget ceiling, retained assets, and decision rule.

Buy a pilot that can reach a decision

A useful pilot is small enough to inspect and specific enough to change a decision. It might cover one practice, one market, a defined group of pages, a documented baseline, agreed factual or editorial improvements, and a review date.

Before it starts, agree what each outcome would mean:

• Expand if the firm sees useful, repeatable evidence and has the capacity to act on the resulting inquiries.
• Repair if the baseline exposed factual, access, content, or intake gaps that prevent a fair test.
• Observe longer if the work is complete but evidence is too thin to decide.
• Stop if the proposed activity cannot be tied to a useful reader task, a trustworthy measurement source, or a realistic owner.

The goal is not to manufacture a favorable screenshot. It is to decide whether the next dollar belongs in public information, technical work, content, intake, paid demand, or a narrower AI-search experiment.

Use the same standard on Juris Digital

Matt Green's warning on Juris Digital's Non-Billable Hour podcast is useful here: do not buy work you “don't necessarily need or understand.” That is a procurement test, not a reason to ignore AI-search changes.

Ask us to show the same things you would ask another provider: the reader decision, baseline, proposed work, owner, source records, measurement limits, and review point. If the answer depends on a platform claim, ask for the current primary source. If the evidence cannot support a decision, the right answer may be to fix a more immediate constraint first.

An AI-search engagement is worth considering when it makes your firm's public information more accurate and useful, gives your team a clear record of what changed, and helps you decide what to do next. The acronym matters much less than that.

Casey Meraz Casey Meraz is an entrepreneur, SEO expert, investor, creator, husband, father, friend, and CEO of Juris Digital. Casey is a frequent speaker at industry events and the author of two books on digital marketing, including "Local Marketing for Personal Injury Lawyers" and “How to Perform the Ultimate Local SEO Audit”

Connect with Casey Meraz on LinkedIn