Most chatbot qualifying flows ask too much, lose half the conversation, and then blame the chatbot. The fix is not to ask nothing — it is to ask the right four things in the right order. This piece walks through the minimum viable qualifying flow for a service business: which four lead qualifying questions to ask, the order they belong in, the one most chatbots ask that they shouldn't, and three vertical-specific scripts you can copy on a Sunday afternoon and ship on Monday.

Why most chatbot qualifying flows kill conversion

Form-length data is brutal. Every additional field you ask for costs you a measurable slice of conversion, and chat compounds the problem because every message is a fresh chance for the visitor to leave the tab.

These are form numbers, not chat numbers, and you should not pretend they translate one-for-one. But a chatbot is just a multi-step form with a friendlier UI. A vendor like Drift or LeadTruffle that asks twelve questions in chat is a twelve-field form with a chat skin on top — and the data on that pattern is consistent: each added question buys you a little less context at the cost of a measurable share of completed conversations.

There is a contrarian read worth naming. CXL has argued that more fields can sometimes mean better leads — fewer total submissions, but each one more qualified, which can still net out positive. That is a real effect for high-ticket B2B sales with expensive sales-team time on the other end. For most service businesses (trades, home services, agencies under £5k/mo retainers), the maths flips: you want more conversations and a fast follow-up, not fewer conversations and a longer dossier per lead.

So: shorter is better, but only if the four things you keep are the right four.

The four questions that actually qualify a service-business lead

A service-business chatbot has one job at the qualifying stage: find out whether you can do this work, for this person, in their area, in their timeframe — and get a way to call them back. That is exactly four questions: where, what, when, who. In that order.

Question 1 — Where (service area)

The phrasing matters more than people think. "Please provide your address" reads like a form. "Just to make sure we cover your area — what's your postcode?" reads like a person doing the visitor a favour by checking up front.

Ask this first, not last. If you cannot serve the area, you save the visitor (and yourself) from the rest of the flow and can offer a referral or a list of partner trades. Most chatbots ask for postcode somewhere in the middle, after extracting an email — which means they collect a lead they cannot fulfil and then disappoint the visitor by email.

Question 2 — What (the actual problem)

You do not need five sub-questions to learn that a kitchen tap is leaking. One free-text box plus a "tell me more" follow-up if the answer is ambiguous gets you the same context with a fraction of the friction.

The pattern to avoid is the structured intake form pretending to be chat. A plumbing intake that asks about water heater type, leak location, fixture type, age of the property, and previous work all in sequence is a five-field form, not one question — and the conversion cost of the four extra steps is real.

If your jobs do split cleanly into three or four buckets (e.g. boiler / plumbing / drainage), three single-tap options outperform free text — visitors tap faster than they type, and the structured answer is easier to route.

Question 3 — When (urgency)

Urgency is what tells your team whether to ring this lead in the next sixty seconds or queue it for the morning. It is also the easiest question for a visitor to answer — four buttons, one tap.

Question 4 — Who (contact, last)

Identity goes last because it is the question with the highest perceived cost. Asking for name and email first is the conversion-killer covered in the stop asking for name and email first piece — that argument is the prequel to this one.

Ask for name plus one contact channel — phone for urgent jobs, email for routine ones. Never both. Never a phone-then-email-then-form sequence; that is a flow designed to comfort the sales team, not the buyer.

The question most chatbots ask that they shouldn't

Budget. Or "what's your timeline?", or any scoring question that benefits the vendor and not the visitor. A chatbot is not a sales rep on a discovery call, and the BANT-style "qualify out the small fish" instinct that sales teams pasted into chat does not survive the conversion maths.

The Imagescape number is the one to remember. Eleven fields to four was not a small change, and the lift was not a small lift. The fields they cut were the "qualifying" ones — budget bands, project timelines, role within company. The fields they kept were name, email, what-do-you-need, and how-to-reach-you. Same four shapes as the flow above.

If you genuinely need budget data — for example because you sell three pricing tiers and the wrong tier wastes everyone's time — ask it on the call, not in the chat. The visitor who picked up the phone is twenty times more committed than the visitor still typing into a widget.

The order matters more than the questions

Where → What → When → Who is not arbitrary. Questions that confirm fit (we cover your area, we do this kind of work) come before questions that extract value (your contact details). Reverse that order and abandonment spikes, because the visitor is being asked to pay before they know they're being served.

The "lead quality" row is the honest caveat. An 8-question flow that pre-filters on budget will give you fewer but slightly more pre-qualified leads. For most service businesses that's the wrong trade — you want the call, not the dossier — but the trade is real. Run the maths for your own deal size and call-handling capacity before you commit.

Three example flows you can copy

Three verticals, four questions each. Phrasings are conversational, mobile-friendly, and tappable wherever possible.

When not to qualify in chat

Two situations where four questions is not enough — and you should know it before you set the flow live.

If you're not sure which side of that line you sit on, default to the 4-question flow. The penalty for asking too few is a slightly noisier inbox. The penalty for asking too many is a 50% drop in completed conversations — and you cannot get the lost ones back.

Measure what matters — conversion, not engagement

Most chatbot dashboards show "engagement" — messages exchanged, average session length, that kind of thing. Engagement is the wrong default metric for a qualifying flow. A long, engaged conversation that ends without a contact detail is a failed conversation.

Three numbers worth measuring instead:

• Leads-per-conversation — what share of started chats produce a name and contact channel.
• Qualified-leads-per-conversation — same, filtered to the leads your sales team would actually call back. (You'll need a tagging convention; "in-area + clear ask + reachable" is a fine starting set.)
• Drop-off-by-question-position — where exactly in the flow visitors leave. If the drop is concentrated at Who, your Who phrasing is too aggressive. If it's at Where, your visitors aren't your buyers.

That 10–20% is the bar you're aiming at, not a guarantee. A 4-question flow with the right order, the right phrasings, and a fast follow-up — the 391% within-one-minute number lives or dies on whether someone actually picks up the phone — is how you get there. The vendor demos that quote 35% conversion are usually counting different denominators; trust your own dashboard, not theirs.

This is the same logic behind why after-hours leads cost service businesses so much: a perfectly qualified lead at 9pm with no one to call them back is worth roughly the same as no lead at all. Qualification and follow-up are the same problem viewed from two ends.

Further reading

The four questions are the floor, not the ceiling. You can layer behaviour on top — branching by answer, triaging emergencies, A/B-testing phrasings — but only after the four-question backbone is in place and converting. Start with the floor, then dig into the sources below.

Capture AI Team

Chatbot specialists

We design and ship lead-qualifying chatbots for service businesses. Every recommendation in this post comes from flows we run for real clients — names redacted, numbers not.