Part of Buyer intent
Compare their jobs
| Dimension | Intent data | Lead scoring |
|---|---|---|
| What it is | Observed or inferred evidence | A rule for evaluation |
| Example | A current service request | A priority category based on need and fit |
| Main risk | Weak provenance or incorrect matching | Poor weights or hidden assumptions |
| Review question | What actually happened? | Why does this change the decision? |
A scoring system cannot create certainty missing from the source. Several uncertain fields do not become reliable because they are added together.
A useful sequence
Collect evidence, preserve provenance, review fit and context, then prioritize. Apply disqualification gates before calculating a score. Mark unknowns rather than treating missing values as positive evidence.
Keep a reviewer explanation next to the score. A simple category with a clear reason can outperform an elaborate model that nobody can audit operationally.
Avoid circular evaluation
If the score defines a qualified lead, reporting that high scores produce more qualified leads may simply repeat the definition. Evaluate against a later, independent outcome such as a viable opportunity.
Keep criteria stable long enough to learn and report counts with rates. Small cohorts can make arbitrary changes look successful.
When you do not need a score
A small queue of explicit requests may be handled with fit checks and a clear next-step decision. Add scoring when it improves consistency or manages a review constraint.
The goal is better prioritization. Software complexity is not evidence of better qualification.