Google's AI Overviews can answer part of a legal question before a prospective client reaches a law firm's website. They can also introduce sources, prompt another search, or sit beside ordinary organic, local, and paid results.
That changes the interface and the evidence a marketing team can collect. It does not change the firm's underlying job: become discoverable for work it can handle, publish information worth trusting, make the firm and its services clear, and turn appropriate interest into a useful conversation.
The practical response is not an “AI rewrite” of the whole site. It is a focused review of the queries, pages, sources, controls, reporting, and client decisions that matter to the practice.
Begin with what Google's AI feature changes
An ordinary result can move a person directly to a webpage. An AI Overview may summarize information first and offer links or expandable elements within the answer. The prospective client may:
- read the summary and leave;
- expand the answer;
- follow a cited source;
- continue to organic or local results;
- refine the question; or
- remember a firm or idea and return through another route.
Those are different behaviors. An impression is not a citation. A citation is not necessarily a click. A click is not a qualified inquiry. A mention is not an endorsement.
For a law firm, the strategy question becomes: where does accurate information about the issue and the firm enter this longer decision path, and what evidence can we responsibly observe?
Use Juris Digital's click study for the narrow lesson it supports
Juris Digital published a first-click exercise in February 2026. It showed 140 respondents one screenshot for “best dui lawyers in chicago” and asked where they were most likely to click first.
Scroll sideways to review every column.Each row is shown as a labeled card.
| First choice | Respondents | Share of 140 |
|---|---|---|
| Traditional organic result | 71 | 50.7% |
| AI “See More” | 26 | 18.6% |
| AI citation | 21 | 15.0% |
| AI mention | 16 | 11.4% |
| Other | 6 | 4.3% |
The total is 140. AI-related choices total 63, or 45.0%: 26 + 21 + 16. “See More” does not necessarily send someone to a website.
This was one stated first-click choice on one screenshot. It was not observed search traffic, did not include ads or a local map pack, and cannot provide a market-wide legal click-through rate. Read the study and its method.
The narrow lesson is useful: in this exercise, both AI elements and traditional organic results attracted attention. A law firm should preserve its ordinary organic foundation while learning how AI-feature visibility appears for its own priority searches.
Audit decisions, not imagined prompts
Choose 20 to 30 search situations tied to one priority practice. Cover a decision path rather than compiling every wording variant:
- recognizing the legal problem;
- understanding urgency or deadlines;
- comparing possible routes;
- preparing documents or questions;
- evaluating a lawyer or firm; and
- finding local help.
For each situation, record a dated observation:
Scroll sideways to review every column.Each row is shown as a labeled card.
| Field | What to record |
|---|---|
| Query and purpose | Exact wording and the client decision it represents |
| Conditions | Date, location, device/account state where relevant |
| Result state | Whether an AI Overview appeared and which other result types were visible |
| Answer coverage | Topics, qualifications, and important omissions |
| Visible sources | Cited pages and how they support the answer |
| Firm state | Cited, mentioned without link, present elsewhere, absent, or described incorrectly |
| Best next action | Preserve, correct source facts, improve a page, investigate, or no action |
Do not call 25 checks a visibility percentage for the whole market. The sample is intentionally chosen and search experiences can change. The worksheet identifies patterns and concrete source problems.
For example, a family-law firm might find that AI Overviews appear consistently for general process questions, occasionally for document-preparation questions, and rarely for direct “lawyer near me” searches in its sample. That would support different content and local priorities. It would not prove that every user sees the same interface.
Keep the technical and editorial foundation
Use the technical SEO guide for crawl and rendering controls and how AI search finds and cites content for the evidence and passage-level job.
Google's current guidance says its established SEO fundamentals also apply to AI features. Pages need to be indexed and eligible to appear in Search with a snippet. Google also says no special AI text file or schema markup is required, and it does not recommend breaking content into tiny fragments or building pages around every imagined prompt. Read Google's guidance on AI features and websites.
That directs the work toward durable questions:
- Can Google access and understand the important page?
- Does the page clearly state the service, jurisdiction, attorney, date, and source where those facts matter?
- Does it answer a distinct client decision with enough context to be safe and useful?
- Can the firm support every experience, result, credential, office, and outcome claim?
- Do internal links connect the explanation to relevant services, attorneys, offices, and deeper sources?
- Is the contact path clear if the reader needs counsel?
Google describes “query fan-out,” in which an AI feature may issue related searches across subtopics and data sources while assembling a response. That makes coherent depth useful, but it does not justify manufacturing one page for each predicted subquery.
Improve passages by adding decision value
A conventional legal article often opens with a definition, repeats a statute at a high level, and ends with “contact an attorney.” An AI summary can make that commodity explanation even easier to consume without a click.
The firm's page should earn attention with material the reader still needs:
- a decision table showing which facts change the next step;
- a document checklist for a first consultation;
- a clear distinction between two commonly confused processes;
- attorney explanation of a recurring tradeoff;
- a dated primary-source analysis;
- a diagram or calculator whose assumptions are visible; or
- a precise account of the firm's process and fit.
Suppose an employment page says, “A severance agreement is a contract between an employer and employee.” A stronger section could help an executive identify the review questions: deadline, release scope, equity treatment, restrictive covenants, benefits, tax issues, return of property, and negotiation authority. It should explain which issues require jurisdiction-specific legal review rather than pretending a checklist gives advice.
That improvement is useful whether the visitor comes from an AI citation, an ordinary result, a referral, or an ad. Our guide to non-commodity legal content develops this editorial approach.
Audit the information environment around the firm
AI Overviews may rely on sources beyond the firm's website. Review the places a prospective client or search system may use to understand the firm:
- attorney and office pages;
- Business Profiles and major directories;
- bar, court, professional, and association records;
- reviews and client feedback where publication is appropriate;
- reputable news, legal, and local coverage;
- published research, speaking, and expert contributions; and
- inconsistent or obsolete pages still accessible online.
If an AI result describes the firm incorrectly, capture the query, date, location, wording, and visible sources. Then correct factual errors on sources the firm controls and request appropriate corrections elsewhere. Do not create fake citations or unsupported profiles.
A source correction can improve the information environment. It cannot guarantee that a generated answer changes, appears, or cites the firm.
Check the current Search generative AI control deliberately
Google now documents a Search generative AI control in Search Console. Its documentation says the control rolled out worldwide on August 31, 2026, covers inclusion or exclusion from the relevant generative AI features, and can inherit a setting from a parent property. Google distinguishes this Search control from ordinary Search inclusion and from AI-training controls. Read Google's Search generative AI control documentation.
The person responsible for the property should record:
- the property and parent relationship;
- the effective setting and whether it is inherited;
- the date checked;
- the business reason for leaving or changing it;
- the expected scope; and
- the verification and review date.
Do not change the control because a vendor says “AI visibility is the future.” Inclusion can create discovery opportunities; exclusion may reflect the firm's content policy or other business judgment. Make the decision consciously.
Juris Digital's actual property setting was not inspected for this draft, and no claim is made about it.
Use the Generative AI performance report without double counting
Google's current Generative AI performance report provides impression data for AI Overviews and AI Mode, with views such as pages, countries, dates, and devices. Google states that the data is already included in Web search performance reporting. Read Google's Generative AI performance report documentation.
That means:
- use the dedicated report to investigate AI-feature visibility;
- do not add its impressions to Web impressions and call the sum total exposure;
- compare pages and dates within the available scope;
- preserve low-data and missing-value uncertainty; and
- connect the observation to clicks and intake only through evidence your systems actually hold.
An example report might show increasing AI-feature impressions for a preparation guide and no measured organic clicks from that page. The correct next questions are whether the page is being cited, whether ordinary search behavior also changed, whether brand discovery or later visits appear, and whether the guide still serves a valuable reader. “AI stole our traffic” is not established by those two observations.
Juris Digital's actual Generative AI report was not exported or inspected for this draft.
Run one 30-day AI Overview review
A practical first cycle can fit inside the existing SEO program:
Week 1: choose and observe. Select one wanted practice and 20–30 client-decision searches. Record the result state and visible sources under consistent conditions.
Week 2: inspect the owned pages. Check technical eligibility, page role, current sources, attorney contribution, firm facts, internal path, and contact experience. Identify one to three consequential gaps.
Week 3: improve and verify. Correct source facts or strengthen the highest-value passage. Preserve the old state, obtain legal and brand review, deploy through the normal process, and verify the public page.
Week 4: establish measurement. Check the effective generative control, inspect the dedicated report if available, annotate the change, and connect future reviews to the firm's qualified-inquiry process. Set a realistic next review; do not promise a citation deadline.
At the end, choose one decision: expand the pattern, correct another source, improve a different page, preserve the current work, or wait for more evidence.
Keep Google AI Overviews inside the broader search strategy
Compare the practical AI-search visibility actions with the shared SEO-and-GEO strategy before creating a separate backlog.
This article is about Google's AI Overviews, Search controls, and reporting. The law firm guide to generative engine optimization covers broader visibility across AI answer systems.
Juris Digital's law firm SEO service combines technical foundations, useful legal content, firm and local facts, authority development, and outcome measurement. If your team is reacting to AI Overviews with prompt-page production or unsupported citation promises, bring us one priority practice, the pages and queries that matter, and the Search Console evidence you have. We can assess the actual visibility path and define a practical next scope; the proposal and agreement will govern the work.
Last updated:
