How I Get Professional Services Firms Cited by AI Search

The new referral: a buyer asks ChatGPT for an expert

Referrals used to start with a colleague. “Who did you use for that?” A growing share now start with a prompt instead.

“Who are the best fractional CMOs for consulting firms?” “Which advisors help B2B companies get found in AI search?” The assistant answers with a short list of names and a line on why each one made it. That list is the new referral, and unlike a results page, there’s no page two.

And it doesn’t stop at the prompt. Even when a referral still comes from a colleague, more and more buyers check it with AI before they pick up the phone. When I raised this during my talk to consulting firm founders in Montreal, heads nodded around the room.

They ask whether the firm is legit, or whether it’s a scam. They look for reviews and testimonials. They want to know which clients it has worked with, and whether it has worked with companies like theirs, or even their competitors.

I saw this firsthand with a marketing agency I worked with recently. When I tested the questions its buyers ask, Gemini didn’t take the agency’s word for it. It ran its own searches on the agency’s name, its city, and one of its household-name clients, checking the same things a cautious buyer would.

What it could find wasn’t all flattering. The agency’s Google Business Profile described it as serving small businesses and listed services it doesn’t offer. Two different addresses showed up online. And its case studies, the first thing its prospects read, had no numbers in them. Those became the first fixes in our next phase.

So AI has become a second opinion on every referral. If it can’t confirm that you’re real and proven, a warm introduction can cool before you ever hear about it.

I’ve worked in organic visibility for most of my career, and this is the biggest shift I’ve seen in how expert firms get found.

SparkToro’s research found that 68% of U.S. Google searches in early 2026 ended without a click to anywhere. So clicks were the wrong scoreboard from the start. What matters is whether your firm’s name is in the answer.

Why expert-led firms are invisible to AI

Most professional services firms I look at know far more than their website shows. The real thinking lives in proposals, workshop decks, client calls, and the partners’ heads. AI can’t cite a deck it can’t read.

Then there’s the site itself. Service pages that promise “tailored solutions” and “a client-first approach” give a model nothing to repeat. Which is a positioning problem before it’s a search problem.

That’s why I start with Focus, the first pillar of Power Positioning. If a firm can’t say in one sentence who it serves and what it’s known for, a model can’t either. In my experience, firms that try to be everything to everyone get described vaguely, or not at all.

Two more gaps show up often. One is messy entity data: the firm’s name spelled three ways, partner bios that disagree, directory listings nobody has touched in years. The other is third-party silence. No lists, no interviews, no reviews. AI cross-checks what you say about yourself, and if nobody else says it, the claim carries less weight.

What AI engines cite when buyers ask who to hire

“Who should I hire” questions pull from different sources than “how do I” questions. When I track buyer prompts, the answers to “who” questions tend to lean on a familiar mix. The firm’s own service and about pages, LinkedIn profiles, “top 10” ranking lists, directories and review platforms like Clutch, and podcast or interview pages.

I learned this on my own name first. When I tracked prompts about who helps companies show up in AI search, several of the sources the assistants cited were agency ranking lists and directories. I wasn’t on them. My own pages had the expertise, but the lists had the vote.

A word of caution, though. Getting onto the right lists is different from manufacturing them. Publishing your own “best firms” list with yourself at #1, or buying your way onto lists that exist mainly to sell spots, is the kind of shortcut I wrote about in this week’s Fortin File. Lily Ray found that Google’s AI Overviews now leave the publishing brand out of its own self-ranked list 83% of the time.

What holds up is being on lists a buyer would trust. Independent rankings, directories with real client reviews, and publications that chose to include you. They take longer to earn, and that’s exactly why they carry weight.

One more thing. Showing up in a source list isn’t the same as being recommended. A model can use your content to build its answer and then recommend someone else, which is why I read Prominence in two depths, cited and recommended.

How I read it: the Triple-P Model

Before I change anything, I measure. I run a fixed set of buyer questions across the major engines, more than once, because AI answers shift from run to run. The full method is in how I run an AI visibility audit.

Then I read the results through the Triple-P Model. Presence asks whether the firm shows up for the questions its buyers ask. Prominence asks whether it gets cited and recommended when it does. Performance asks whether any of it turns into brand searches, direct traffic, and pipeline.

The layer that breaks tells me where to work. Weak Presence? The firm needs pages that own its claims. Strong Presence, weak Prominence? It needs proof and third-party footprint. Visibility with no demand behind it usually points back to positioning.

The work itself

The fix tends to come in four parts, and most firms need some of each.

First, citable pages. I rebuild key service and framework pages with SOME, the four moves I teach for making content AI can quote: Summarize, Organize, Modularize, Emphasize. Each page opens with a summary a model can lift whole, and the firm’s methods get real names instead of generic labels.

Second, entity consistency. The firm’s name, description, people, and core facts should match across its site, schema markup, LinkedIn, directories, and Wikidata where the firm qualifies. When I set up my own Wikidata entry, there were already five other Michel Fortins in there. Firms with common names or founder-named brands run into the same problem.

Third, earned mentions. This means getting onto the lists, directories, podcasts, and publications AI already cites for the firm’s category. It’s the slow part, and it’s the part most firms skip, because nobody owns it.

Fourth, original proof. AI quotes specifics. In my FORCEPS framework, that’s Factual and Evidential proof: a benchmark from your own client work, a named method, a finding nobody else can claim. Generic advice gets paraphrased. Distinctive data gets attributed.

One consulting firm I worked with shows how this plays out. When I started, the founder had spent years building real search equity, including books, podcasts, named frameworks, and over a thousand published articles. But rankings had started slipping as AI search changed how buyers found consulting expertise.

The broken layer was Presence. The library was deep, but too few pages made claims a model could lift. So that’s where the work went first. I led the rewrite of 100 core articles, merged overlapping ones, restructured them for AI retrieval, and added schema. Nearly 200 articles were rewritten or consolidated in total.

The firm’s sessions from AI search grew 924% year over year. New inbound leads also started telling sales they’d found the firm through ChatGPT and Gemini. That’s one firm with its own starting point, so I don’t treat it as a forecast. But it shows what happens when you fix the broken layer before anything else.

What it looks like as an engagement

I call this work The Vector, my visibility amplification engagement. It usually starts with The Gauge, my diagnostic, because the right fix depends on which layer is broken. Some firms need pages. Others have great pages and no footprint anywhere else.

The details for both are on my services page. And for firms that need someone to own marketing as a whole, not just AI visibility, that’s where my fractional CMO work for consulting firms comes in.

Your buyers are already asking AI who to hire. The question is whose name comes back.

---

Frequently Asked Questions