Using Branded Search Data to Gauge a Defense Firm's AI Visibility

Defense firms keep asking a reasonable question: how do we know whether ChatGPT, Gemini or Perplexity ever mention us? The assistants don't send a report. Referral traffic from them is real but often small, and plenty of people who read an AI answer never click a link at all. They read the firm's name, close the app and search for it later.

That later search leaves a trace. When more people type your firm's name or your lawyers' names into Google, Search Console records it. Branded search data won't prove that an AI assistant recommended you, but read carefully over time it is one of the more useful indirect signals a defense firm has. This article explains how to pull it, what can distort it and how to pair it with other checks.

Why branded searches can follow AI answers

Think about how someone uses an assistant after an arrest. They describe the charge and the city and ask who handles cases like it. The answer names a few firms. The person doesn't call from inside the chat. They open a browser, search the firm's name to see reviews and the website, and then decide.

That pattern turns AI mentions into branded searches on Google. It also happens with word of mouth, local news, billboards and directory listings, which is why branded data is a signal and not proof. The job is to separate the AI-related change from everything else as well as you reasonably can.

Pulling branded queries from Search Console

In the Performance report, add a query filter. Use "Custom (regex)" and build a pattern that matches your firm name, common misspellings and each lawyer's surname. A made-up firm called Hale and Ortiz Defense might use something like hale|ortiz|haleortiz. Test the filter and look at the queries it returns to make sure it isn't catching unrelated searches, such as a common word that happens to be a surname.

Set the date range to the last 16 months, which is the most Search Console keeps, and compare month over month. Export the data to a spreadsheet so you have it after those months roll off. Record two numbers each month: total branded impressions and total branded clicks.

Things that distort the numbers

Several ordinary events can move branded search without any AI involvement. Keep a simple log beside the data so you can explain jumps:

• A lawyer appears in the news, on a podcast or in a local event.
• The firm opens an office, adds a lawyer or changes its name.
• A paid campaign or billboard starts or ends.
• A high-profile case brings attention to one of the lawyers.
• Seasonal patterns, since some arrest types rise around holidays.

If branded impressions rise in a month where none of those happened, that is worth a closer look. It doesn't confirm an AI cause, but it narrows the list.

Pair the data with direct checks

Branded data becomes much more useful next to a record of what the assistants actually say. Once a month, ask the main assistants the questions a client might ask, such as who defends DUI cases in your county, and write down which firms appear. Use the same wording each time so the results are comparable. Results vary between sessions, so run each question more than once.

Then look for a pattern across months. If your firm starts appearing in answers in March and branded impressions climb in March and April with nothing else in the log, you have a reasonable working connection. If branded searches rise while the assistants never mention you, look elsewhere for the cause.

Direct traffic data from analytics helps too. Visits referred from chatgpt.com, perplexity.ai and similar domains appear in referral reports, and developers can go further with measuring AI crawler visits and AI referrals at the server level.

Reading the trend without overreading it

Small firms have small numbers. A change from 40 branded impressions to 55 in one month is noise as often as it is signal. Look at three-month rolling averages instead of single months, and only draw conclusions from changes that hold.

Also compare branded clicks with intake. Ask new callers how they found the firm and record the answer. If several mention "I asked ChatGPT" or "Google's AI thing said your name," that is the most direct evidence you will get, and it is worth more than any chart.

Questions about branded search and AI visibility

Can Search Console show which searches came from an AI assistant?

No. It reports queries typed into Google, not where the person first heard your name. The connection to AI answers has to be inferred from timing and other evidence.

What if our firm name is a common word?

Filter on the full name in quotes and on each lawyer's full name rather than single surnames. Review the matched queries by hand each month to remove false matches.

How long should we track before drawing conclusions?

At least six months for a small firm, so you can see a baseline and how it moves around normal events.

For more on reading search data and measurement for legal marketing, see the legal marketing analytics and search data guides.

SBD Marketing works only with criminal defense firms, one per city, and helps with website design, SEO for the charges a firm defends, visibility in AI tools such as ChatGPT, Gemini, Claude and Perplexity, and SEO audits.