Ask any adviser where their best clients come from, and you’ll hear the same answer almost every time: a referral. A happy client, an accountant, a solicitor, somebody the prospect already trusted, quietly vouching for somebody they hadn’t met yet.
For decades, that’s how trust travelled in financial services. It moved through people, one conversation at a time.
Here’s a question that would have sounded absurd five years ago: what happens when your next referral isn’t a person at all?
The Rise of the Invisible Referrer
Digital marketing has been through a few identity crises. For years, it was about being found. Then it became about being ranked. Now, in a shift most firms haven’t quite clocked yet, it’s about being recommended, and increasingly by something that has never attended a networking breakfast in its life.
The scale of that shift is no longer a curiosity; it’s a documented behaviour change. BrightLocal’s 2026 Local Consumer Review Survey found that 45% of consumers had used an AI tool to find a local business or service in the past year, up from just 6% the year before, a sevenfold jump that already puts AI ahead of Yelp and TripAdvisor as a discovery channel.
UK research points in the same direction: a 2026 Investor Index study of 1,100 UK adults with at least £10,000 invested found 73% believed ChatGPT could give reliable financial advice in the future, and 42% of investors aged 18 to 34 had already used it for exactly that.
Rather than spending half an hour comparing five websites and a PDF brochure, a growing number of prospects now simply ask something like: “Who’s the best adviser for a business owner selling up?” The AI reads what’s out there, filters it, and hands back a shortlist. By the time a human prospect lands on your website, the audition may already be over, and nobody thought to tell you it was happening.
An Old Idea Wearing a New Coat
Referrals have always run on the same basic mechanic: somebody else does the filtering, and trust transfers before the first phone call happens. A friend says “use this person.” An accountant says “speak to her.” That endorsement removes the risk before you’ve even said hello.
AI is quietly stepping into that same role. Not perfectly, and not always accurately, but increasingly as a first-pass recommendation engine that never sleeps, never retires, and never expects a Christmas card in return.
The Problem: Most Firms Sound Identical
Picture an estate agent whose every listing reads “beautiful home, tailored to your needs, exceptional service.” You’d learn nothing, and you’d stop reading after the second one.
That, unfortunately, is how most financial services websites read. Client-focused. Trusted. Personal service, naturally. You’ll find some version of all three on nearly every site in the sector, which is the trouble. It’s the marketing equivalent of describing every house on the street as “nice”: true, harmless, and completely useless for telling one from another.
Humans, to be fair, quite like this kind of language. It feels warm. AI models could not care less. They respond to structure: clear positioning, a defined audience, and plain answers to who you help, what you specialise in, and why it matters. A gorgeously designed website full of soft, interchangeable phrases is nearly invisible to a system trying to match a prospect’s question to a genuine answer.
What Actually Earns the Recommendation
If AI is filtering the way a good referral partner always has, the firms most likely to get chosen are the ones that make that filtering easy. Three things tend to matter most.
Specificity beats scope every time. “We help business owners” says very little. “We help business owners in the two years either side of an exit” says almost everything, giving both the AI and the prospect something to recognise themselves in.
Then there’s structure. Clear service pages, FAQ-style content answering real questions in plain English, and consistent information across your website, LinkedIn and any directories you sit in. This matters even when the client arrives via a traditional referral: research from Wealthtender found that 72% of prospects still visit an adviser’s website before making contact, even after being personally referred. The website is doing more of the convincing than most firms assume, human or AI referral alike.
And finally, proof that isn’t generic. A five-star rating with no context tells an AI model almost nothing. A review describing a client who needed help navigating a business sale, and how that actually went, tells it a great deal.
None of this asks a firm to become less distinctive. It asks it to say what makes it distinctive clearly enough that a machine can repeat it back accurately. Done properly, that’s exactly what a good referral has always done.
Where the Real Risk Sits
The uncomfortable part isn’t that AI recommendations get things wrong occasionally. It’s that most firms will never know they’ve been quietly left off the list. Nobody emails you to say a prospect asked an AI assistant for a recommendation and heard someone else’s name. They just don’t call, and the silence gets blamed on everything except the real cause.
The Marketing Hedge Takeaway
The goal used to be getting found. It’s shifting toward getting recommended by a referral partner that never turns up to a networking breakfast and never sleeps. Firms that a machine can repeat accurately won’t just rank higher. They’ll already be on the list before the conversation starts.
Click here to see the original Marketing Hedge newsletter.
