Three worked scripts: fit, almost, and wrong
The state-machine idea stays abstract until you watch it run. Here are three short conversations on the same pricing page, each ending in a different bucket, so you can see how two or three questions do all the sorting.
The strong fit, booked at midnight. A visitor has been on the pricing page for ninety seconds. The agent opens: "Looks like you are weighing up the plans, happy to help, what are you trying to solve?" The reply: "We are a 30-person agency drowning in manual reporting." That is the fit signal and the problem in one answer. The agent asks the only remaining load-bearing question: "Are you looking to move on this soon, or scoping for later?" The answer, "this quarter," closes the qualification. The agent books a held slot, writes the CRM record, and confirms, all before the founder wakes up. Three turns, one booked call, zero manual triage.
The good fit who is not ready. Same opener, same page. The reply: "Honestly just researching, we will not have budget until next year." Fit is right, problem is real, but timing is wrong, and forcing the call here is how you turn a future customer into a lost one. The agent does not push. It captures the email, drops the person into a nurture sequence, and sets a clean expectation: "Makes sense, I will send you the one-pager and check back nearer your timeline." The relationship warms at its own pace. Activating more leads is the sequence this hands off to.
The wrong fit, deflected kindly. The reply: "I am a solo freelancer looking for something free." That is a clear miss on the buyer profile, and the most profitable thing the agent can do is not waste an hour pretending otherwise. It points them somewhere genuinely useful, a self-serve resource or a free tool that actually fits, and ends warmly. No call booked, no calendar burned, and the freelancer leaves with a good impression rather than a sales pitch they did not want.
The pattern across all three is the same: the agent asks fit, then problem, then timing only if the first two clear, and the conversation exits the instant the outcome is decided. Notice what it never does. It never asks all three questions of the freelancer who failed the first. It never drags the not-ready buyer onto a call to hit a number. The script is short because the state machine cuts every question that would not change the destination, which is the whole craft of it.