Sales & CRM Systems
Lead qualification: a verdict with reasons for every inquiry
One week of inquiries, four layers
Layer 1
Rules
Hard criteria you can write down: service area, job type, minimum size, timeline.
9 removed: 5 outside the service area, 3 work you don't do, 1 below minimum size
Layer 2
Enrichment
Fills in what the inquiry didn't say, where a reliable source has it: property type, business size, existing customer.
12 enriched, none removed
Layer 3
AI extraction
Reads free-text descriptions and call transcripts and turns them into fields the rules can check.
14 read; 2 more removed by the rules on the new fields · 4 flagged as borderline
Layer 4
A person
Looks at the borderline cases with everything above in front of them.
4 reviewed: 3 qualified, 1 not a fit
40 inquiries in12 not a fit · 0 unanswered
- Passed
- Removed, with reason
- Flagged for a person
Your criteria decide what each layer checks.
AI reads. The rules decide.
Free text
“Hi, we manage a 12-unit building near Wrigley and the boiler's been making a banging noise since Sunday. Not urgent yet but we'd like someone out this month. Can you quote?”
Extracted fields
- service
- boiler repair
- property
- multi-unit, 12 units
- location
- near Wrigley Field (ZIP inferred: 60613)inferred
- timing
- this month
- urgency
- not urgent
- request
- quote
“ZIP inferred” is marked as inferred, not confirmed. Routing treats it as provisional until the customer confirms the address.
Rule checks
- ✓ Service lineboiler repair → offered
- ✓ Service area60613 → inside
- ✓ Property typemulti-unit residential → served
- ✓ Timelinethis month → within booking window
Qualified · in area, offered service, within window
The model's job is narrow: turn messy language into fields, and say how sure it is. It doesn't decide who's worth calling. Your written criteria do that, the same way for every lead, which is what makes the outcome explainable and the extraction testable against real inquiries before it goes live.
Every verdict comes with its reasons
Qualified
Next: scoring and routing.
In the service area, a service you offer, within your booking window.
Not a fit
Next: a same-day reply, with a suggestion of who might help.
Commercial boiler install; you serve homes and residential buildings only.
Needs review
Next: a person decides, and the decision is recorded.
Large job with no address yet.
A fast no is part of the service
An inquiry that never gets a reply leaves a worse impression than one that gets a polite no. Because qualification records why a lead isn't a fit, the reply can be specific, immediate and approved in advance, without anyone writing it at the end of the day.
Thanks for getting in touch about the boiler installation. We work on homes and residential buildings, so we're not the right fit for a commercial installation. A commercial mechanical contractor will be able to quote it properly. If you ever need plumbing or heating for a home or residential building, we'd be glad to help.
Approved template · sent automatically for this reason
People can overrule it, and that's recorded too
01
Every change is recorded
Anyone with the right access can change a verdict. The change and the reason are saved with the lead.
02
Overrides are a signal
If the same rule keeps getting overruled, the rule is wrong, not the people.
03
Checked against outcomes
Criteria are reviewed against what happened next: which qualified leads booked, and which “not a fit” leads would have.
What qualification hands off
In
Out
Worth building when screening leads by hand costs real time
Good fit
- You can write your criteria down: area, service, size, timing.
- Enough inquiries arrive that reading each one is a job in itself.
- Poor fits currently wait days for an answer, or never get one.
Not the right fit
- Fit only becomes clear after a long discovery conversation, which is common in complex, high-value sales.
- A handful of inquiries a week, all read by the owner.
Questions about lead qualification
What's the difference between lead qualification and lead scoring?
Qualification decides whether a lead fits at all: yes, no or needs review. Scoring decides the order among the leads that do fit. Most teams need qualification first.
Can AI qualify leads on its own?
It can read and structure what a lead wrote, which is the hard part. The decision itself works better on written criteria that people own, with borderline cases going to a person.
What criteria should we start with?
The ones that already decide whether you take a job: where it is, what it is, how big it is and when they need it. Add more only when overrides show a gap.
What happens to leads that aren't a fit?
They get a prompt, specific reply from an approved template, and they're kept with their reason so you can see what you're turning away.