“I'll know it when I see it” is not a failure of the client. It is a completely normal thing to say about a discipline you do not practise.
Nobody expects a client to arrive with a specification. What they arrive with is a business problem and a feeling, and the job of the first conversation is to convert both into something that can be scoped and priced without either of us guessing.
Ask about the business, not the website
“What do you want the site to do” gets you a list of pages. “What happens in the business when this works” gets you the actual brief — more enquiries of a particular kind, fewer support emails, a sales conversation that starts further along.
Every design decision downstream can then be argued against that sentence, which is what stops a project drifting into taste.
Get them pointing at things, and asking why
Show three sites and ask which feels closest — but the useful part is never the choice, it is the reason. “It feels serious” and “it doesn't look like it's trying too hard” are specifications, once you write them down as such.
A client who cannot describe what they want can almost always describe what they do not want. Start there, it narrows faster.
Wireframe before anything is beautiful
Colour and type will absorb the entire review if you let them into the room too early. Structure first, in grey, so the conversation stays on what goes where and why.
Changes are cheap at this stage and expensive later. Making that explicit — out loud, in the first call — is what earns you the room to spend two weeks on structure.
Write it down and send it back
End the call by writing what you heard in a page and sending it over. Half the time the reply corrects something important, which is exactly the point. The other half you have a scope both parties have now agreed to in writing.
Want this run on your site?
A fixed-fee performance audit, delivered in a week.