Your Law Firm Is Measuring AI Wrong (And It’s Costing You Clients)

Your analytics are undercounting AI’s influence by as much as 10x — because AI shapes decisions before anyone clicks anything. The fix isn’t a new tool. It’s a question.

Ross Hudgens, CEO of Siege Media, posted something on LinkedIn recently that stopped me mid-scroll.

He wrote: “5% of our closed-won deals come from LLMs.” Then he called himself wrong.

“5% of your currently attributable deals come from LLMs,” he clarified. “The real number is far more likely to be in the 30–60% range.”

His reasoning: most firms look at referral traffic to measure attribution. But referral traffic only captures the moment someone clicked a link from an AI tool. It misses everything that happened before that click — the research, the comparison, the recommendation that shaped the decision before the person ever touched a search bar.

I reached out to Ross. He confirmed: for Siege, 5% of traffic comes from ChatGPT. But when they added a self-attribution prompt to their lead form — simply asking “How did you hear about us?” — 50% of leads credited an LLM. Ten times the default assumption.

That gap is not a rounding error. It is a blind spot baked into how most firms measure their marketing. If you have not read our breakdown of SEO vs. GEO, that is the right place to start.

What the Data Actually Shows

Datos, a Semrush company, spent all of 2025 publishing quarterly State of Search reports built from clickstream data across tens of millions of real desktop users in the US, EU, and UK — developed in collaboration with Rand Fishkin, CEO of SparkToro. Actual device behavior, not surveys.

Here is what a full year of that data reveals.

AI tools are a small but growing share of total search

Across 41 major domains analyzed in Q4 2025, traditional search engines still account for roughly 80% of all desktop search activity in the US. AI tools as a category represent about 3.2%. Commerce sites like Amazon account for around 10%. Social platforms come in near 5.5%.

AI has not replaced Google. But it has started to change how people use Google — and that distinction matters enormously for law firm marketing.

According to the Q4 2025 State of Search report, Google desktop searches per user fell nearly 20% year-over-year in the US. Rand Fishkin’s analysis points to a clear cause: AI-powered answers are resolving questions before users need to search again. They get what they need faster, click less, and return to search engines less.

In Europe, that same decline was only 2–3%. The behavioral shift is happening faster in the US — which puts the US legal market on the leading edge of this disruption.

ChatGPT leads, but the AI field is fragmented

Within the AI tool category, ChatGPT remained dominant throughout 2025, reaching roughly one-quarter to one-third of desktop AI users in the US and over 40% in Europe. Google Gemini grew steadily as a distant second, overtaking DeepSeek by mid-year. Perplexity, Copilot, and Claude maintained niche positions with no breakout adoption.

DeepSeek spiked to 6–7% usage share in February 2025, then declined to around 3% by mid-year. That volatility is a reminder that the AI tool landscape is not settled. This is part of why building AI visibility across platforms matters.

Visits are not the same as usage

One of the most underreported findings from the Datos research: only about half of people who visit ChatGPT actually enter their own prompts. Many arrive to view a conversation someone else shared. Raw traffic data overstates active AI usage significantly.

If your team is looking at ChatGPT referral sessions in Google Analytics and calling that your AI attribution number, you are measuring the wrong thing. You are counting people who clicked a link from an AI conversation — not the much larger group whose decision was shaped by one.

Why Referral Traffic Lies

The way most people use AI tools before hiring a lawyer does not produce a referral click.

They open ChatGPT or Gemini and ask: “What should I do if I was injured in a car accident in Denver?” or “Is a DUI a felony in Colorado?” The AI answers. It may mention a firm by name. The person reads, closes the tab, and opens Google — typing the firm name or a search term directly. That journey produces zero AI referral traffic. Google Analytics records it as direct or organic. The AI’s role is invisible.

This is not a hypothetical. It is the most common way people use AI for high-stakes research decisions. Legal services are exactly that kind of category. This is the same dynamic we described in our post on why most law firm content strategies are broken: firms optimize for signals they can easily measure, while the signals that actually drive decisions stay invisible.

What We Do at LGD — and What We Recommend for Clients

At Legal Guardian Digital, we use tools like CallRail and Attributer.io for lead and form tracking. They help. They are not perfect. Data from these tools can tell you one thing while the truest picture of attribution is something else entirely.

The most reliable signal we have found is also the simplest: we ask every prospect, “How did you find us?” in the first conversation. We recommend every client do the same. No technical setup. No budget. Most people are happy to answer — and the answers are consistently more accurate than anything analytics platforms report.

Ross Hudgens at Siege Media made this formal by adding a field to their lead form. See it live at siegemedia.com/business-form. A simple “How did you hear about us?” field — optional or post-conversion — gives you far more truth than referral data alone.

What This Means for Law Firm Marketing

The data from Datos and SparkToro is clear: AI tools are growing, search behavior is shifting, and the gap between what attribution software captures and what actually drives decisions is widening. Three things worth building around this:

Optimize for AI visibility, not just search rankings. If ChatGPT or Gemini recommends your firm by name, that is a referral you will almost never see in your analytics. Generative Engine Optimization (GEO) is about making your firm the answer AI tools give, not just the result Google ranks.

Treat your content as a trust signal for AI. Thin pages and templated practice area descriptions are invisible to these systems. Content built around real attorney frameworks and specific client scenarios is what gets cited and recommended. Our breakdown of why law firm content strategies fail goes deeper on this.

Ask the question. Self-attribution is imperfect, but it is currently the best tool available for closing the gap between what your analytics show and what is actually driving new clients through your door. Build it into your intake process. Make it a habit on every new matter.

The firms that adapt their attribution thinking now will be the ones who see clearly while everyone else is still looking at the wrong numbers. If you want to talk about what AI visibility looks like for your firm, let’s have a conversation.

Sources

Datos & SparkToro — State of Search Q1–Q4 2025
Rand Fishkin, SparkToro — New Research: Search Happens Everywhere (March 2, 2026)
Rand Fishkin, SparkToro — 20% of Americans Use AI Tools 10x+/Month (August 2025)
MarTech — Google Searches Per U.S. User Down Nearly 20% (Jan 28, 2026)
Search Engine Land — Google Searches Per US User Fell Nearly 20% YoY
Ross Hudgens, Siege Media — LinkedIn, March 2026. Form: siegemedia.com/business-form

Written By Austin Hunt

Austin Hunt is the CEO & founder of Legal Guardian Digital, a trusted agency specializing in SEO for lawyers and AI search optimization. He has devoted his life’s work to helping attorneys connect with their audience in an effective, ethical, and lasting way—educating, guiding, and building strategies that endure.