A firm owner tells me their website is not working and asks whether AI could help. I ask what the site is supposed to do. There is a pause, and then several answers: look credible, explain the services, bring in the right kind of client.

Three goals, no order between them. No tool solves that.

This comes up often enough that it is worth stating plainly. Most firms reaching for AI are not short of technology. They are short of a decision.

What does AI actually do to an unclear offer?

It reproduces it, faster and in more places.

Point a drafting tool at a fuzzy value proposition and you get fifty variations of a fuzzy value proposition. Add a chatbot to a site that cannot say who it is for, and visitors now have a faster way to reach a confusing answer. Automate a follow-up sequence built on the wrong message and the wrong message arrives on schedule.

These tools multiply whatever they are given. If the input is unresolved, the output is unresolved and there is now more of it.

What has to be settled first?

Three things, and they are not technical.

Who you serve, specifically enough that someone reading it knows whether they qualify. "Individuals and families" does not qualify anyone.

What problem you take off their plate, in their words. Not a service list. The thing they were worried about at eleven at night before they called you.

What you want a visitor to do next, singular. Most sites offer a visitor four options of equal weight, which is the same as offering none.

Write those three on a page. If your team cannot agree on them in a sitting, that disagreement is the actual project, and no tool will resolve it for you.

How do you tell which problem you have?

Ask two people in your firm to describe, separately, who the ideal client is. Then read your homepage and ask what a stranger would say the firm does after five seconds.

If the answers line up and the site says it well, and you are still not getting traction, you may have a distribution or a mechanism problem, and AI could plausibly help with parts of it.

If the answers do not line up, you have a clarity problem. That one is cheaper to fix than a technology project and it makes everything after it work better, including the AI.

Where does AI fit once the clarity is there?

In the ordinary places. Drafting content you then edit. Summarizing research. Handling the repetitive work around a system that already knows what it is for.

That is the sequence we use in our own work: settle the message, build the mechanism, then automate the parts that are safe to automate. It is described under Considered Build, and the AI-specific side under Clarity AI.

None of this is an argument for going slowly. It is an argument for doing the cheap thing first. An afternoon spent settling three sentences will improve every piece of marketing you produce for the next two years, whether or not a machine writes any of it.

If you want an outside read on how clearly your site answers those three questions today, The Second Look is free and takes about four minutes.