Three enquiries that didn’t look like our usual ones reached us between late August and mid-September.
None came through a referral, and none came from anyone we knew. Each came from someone who had asked ChatGPT to recommend an agency.
That isn’t how most of our new work arrives. Most of it comes through referrals. Someone we know mentions our name, and the prospect looks through our website before the first call. Three enquiries from a different source in a few weeks made us stop and pay attention.
What follows is what happened and what we make of it, including the parts we don’t know.
They arrived already knowing what we do
What struck us was where each conversation started.
Usually, a cold enquiry starts from scratch. We explain what we do, who we do it for, and whether it’s relevant to the person asking. Much of the first call is spent getting to a real conversation.
These three skipped that. Each arrived already knowing what we do and who we do it for. One said up front that they were interested in our financial services and insurance experience, which is not how a cold enquiry normally opens.
None of the three needed us to explain what we do or which sectors we work in. The first call with a stranger is usually about whether there’s any fit at all. Here, the conversation started further along.
With a referral, the person you trust does the vouching, which lets the first call move quickly. These three had no introduction from anyone. What they had instead was an answer from a tool, and they treated it as reason enough to get in touch.
What we think made the difference
We can’t see how an AI search tool puts an answer together, so this is our reading rather than a finding.
All we can say is what we had done. We reworked some parts of our own website earlier in the year so that it says plainly who we work with, sector by sector, and answers the questions buyers tend to ask, in their words.
Our guess is that this gave an AI tool something clear to repeat. A site that says plainly who it’s for is easier to summarise than one that doesn’t. If the language is vague, there’s nothing to pick up.
It’s worth asking what an AI tool has to work with when someone asks it to recommend an agency. A page that says it helps ambitious firms grow gives it very little to repeat, and the same sentence could sit on almost any competitor’s site.
It also changes how we think about the website. In a referral, the site is where a prospect checks what they’ve been told. In these three cases, it may also have been where the recommendation started.
Firms often describe themselves in the language they use among themselves, while buyers ask in different words. That gap can be easy to miss from the inside.
Anyone in financial services marketing will recognise the other half of the problem. Asset managers, fintechs and insurers often describe themselves in the same safe words, and a page that could belong to any competitor tells a reader very little. It might also tell an AI tool nothing, and a tool with nothing to repeat has no reason to name you.
The honest part
It’s been quiet since, and none of the three has turned into a project. One decided to do more of the work in-house. One chose an agency whose portfolio and visual ideas were closer to what they wanted. One hasn’t come back to us.
That’s worth saying plainly. An answer from a tool can put you in the running, and that may be all it does. With nobody vouching for us, we had more to prove once we were up against other agencies, and by then the website had done its job.
We don’t know yet whether those three enquiries were a run of luck or the start of something.
Three is a small number and a few weeks is a short time. AI search is also new enough that we’d be wary of drawing firm conclusions from either.
There’s also a lot we can’t see. We only know about the people who get in touch, not how many asked the same question and went somewhere else. We can’t say whether the three would have found us another way either.
Nobody can promise that AI search works on demand, and we’d be sceptical of anyone who does. It’s tempting to turn three enquiries into a strategy, but we’d rather say what we saw.
If you lead marketing at an asset manager, a fintech or an insurer, you may well be asking yourself whether any of this applies to you. We can’t tell you yet. A fair question is what your own site would give a tool to repeat.
Where this leaves a regulated firm
Financial services marketing runs through compliance, and that tends to push language towards the cautious and the general. It’s easy to see why. Every claim has to be defended, so the safest sentence is the one that says the least. That isn’t a criticism. In a regulated firm, caution about wording is part of the job.
But clear, specific claims that compliance has already approved are exactly what these tools can quote. Firms that say plainly what they do, who it’s for and which sectors they serve, in wording that’s already been signed off, have done much of the hard work.
So that’s where we’re at. Three enquiries is an observation rather than a result.
We can’t say ChatGPT will send buyers your way. But a B2B website that says clearly who it’s for gives it something worth repeating.
