GuidesAI automation
Qualifying leads with AI: score, reasons and next step
The flow
- 1FormMessage, company, budget, timeline.
- 2ModelReads the lead against your written criteria.
- 3Structured outputScore, reasons, next step.
- 4CRMFields filled in, hot lead flagged.
The expected output
{
"score": 82,
"category": "hot",
"reasons": ["stated budget fits", "clear need", "deadline within 2 months"],
"action": "discovery call this week"
}
The schema is imposed on the model (a tool or structured output), then validated by code before writing to the CRM.
Criteria, in black and white
| Criterion | Example rule |
|---|---|
| Fit | The need matches an existing offer |
| Budget | Stated and compatible with your prices |
| Urgency | A dated deadline rather than “someday” |
| Decision-maker | The person can sign, or says who can |
Safeguards
- The score helps sort, it rejects no one: every lead gets a reply.
- The reasons are visible in the CRM, so they can be understood and corrected.
- A sample is reviewed every month to adjust the criteria.
For reliable output: reliable JSON from an LLM. To go further: AI automation.
Frequently asked questions
- What criteria does the model score on?
- The ones you write down: industry, size, stated budget, urgency, fit with your offer. Without written criteria, the score means nothing.
- Can I trust it?
- As a first sort. Compare its scores with a salesperson's on a hundred past leads before relying on it.
- Should I enrich the data?
- The company's website or a public business register add context. Respect GDPR: only fetch what you need to handle the request.