Lead Qualification Criteria: How to Define, Score, and Test Them
2026-09-20 · Julian Hartwell
Lead qualification criteria should map observable evidence to the next responsible action. Separate hard gates from prioritization signals, avoid fixed universal scores or SLAs, and recalibrate criteria on dated acceptance, rejection, override, progression, and reversal evidence from your own process.
Turn each criterion into an action rule
A lead qualification criterion is useful only when an observable fact changes the next responsible action. “High intent,” “good fit,” and a composite score are labels until a reviewer can identify the source, freshness, decision owner, and permitted route. Begin with gates that determine whether a record may proceed; add priorities only among records that already pass. Publish the six contrasts in one list: verified company/no problem, strong signal/unverified role, stale contact, capacity conflict, clear reject, and one borderline two reviewers must argue.
- Company identity is verified before any other criterion applies.
- Segment evidence must be observable and current enough for routing.
- Role evidence can support contact selection but does not prove authority.
- A documented business event can support a hypothesis but not purchase intent.
- An objection or suppression is a stop state, not a negative score.
- Missing required evidence routes to review, not automatic rejection or acceptance.
- Every override needs an owner and reason.
- Every criterion needs a retirement or recalibration trigger.
- Test the definition with a record that looks commercially attractive but lacks one mandatory identity fact; the correct state should remain unresolved.
A score cannot repair a failed gate
If identity, contact context, or a mandatory eligibility fact is missing, extra points from activity or firmographic fields should not authorize progression. Keep gates and priorities visibly separate. For a score cannot repair a failed gate, keep the eligible record, criterion version, reviewer action, override reason, later disposition, and reversal evidence together before recalibrating the qualification rule. Publish the six contrasts as a teaching file: verified company with no problem, strong signal with an unverified role, stale contact, capacity conflict, clear reject, and one borderline the team must argue. The file trains the gate; it is not a benchmark.
Calibrate with six contrasting records
Use a hypothetical six-record teaching set, not an invented benchmark. Record A has verified company and role evidence but no current problem signal. Record B has a strong event but an ambiguous person match. Record C clearly fails the segment. Record D is a referral to the correct owner. Record E carries an objection. Record F passes every stated gate but later disqualifies on timing.
- A passes eligibility and enters low-priority research.
- B is held for identity review despite its attractive event.
- C is rejected with the segment reason preserved.
- D is rerouted and the referrer context remains attached.
- E is suppressed from further outreach.
- F advances, then reverses without rewriting the original evidence.
- Reviewers classify independently before discussing disagreements.
- The calibration result is reviewer agreement and error learning, not a universal conversion rate.
- Ask reviewers to explain their decisions before seeing the other person’s label, then classify whether disagreement came from evidence, wording, or action mapping.
The disagreement file matters
When two reviewers disagree, retain both decisions and ask which criterion was ambiguous: source quality, freshness, threshold, or action mapping. Rewriting the label without recording the disagreement destroys the evidence needed to improve it. For the disagreement file matters, keep the eligible record, criterion version, reviewer action, override reason, later disposition, and reversal evidence together before recalibrating the qualification rule.
Follow one record through override and reversal
Record F appears eligible because the company matches the segment, the role is plausible, and a documented event supports a relevance hypothesis. A reviewer accepts it with a note that timing remains unverified. The recipient later states that evaluation is deferred. The lead moves to a dated recycle state; the team does not relabel the original evidence as wrong or keep the record active through points alone.
- Input: verified company, plausible role, dated event.
- Gate decision: pass identity and segment.
- Uncertainty: current timing is unknown.
- Allowed action: one bounded relevance check.
- Outcome: recipient supplies a timing correction.
- State change: recycle to the stated period.
- Owner: qualification manager reviews the timing criterion.
- Rule change: none unless the pattern repeats across a defined cohort.
- Preserve the recipient’s timing correction as a new observation and do not lower the historical score to manufacture consistency.
Overrides are new evidence, not secret exceptions
If a manager advances a record that fails a published rule, the override must state the additional evidence, owner, and expiration. Otherwise the team cannot distinguish judgment from favoritism or repair the criterion later. For overrides are new evidence, not secret exceptions, keep the eligible record, criterion version, reviewer action, override reason, later disposition, and reversal evidence together before recalibrating the qualification rule.
Measure false accepts and false rejects
Qualification quality appears downstream. Review accepted records that later prove wrong, rejected records later restored, time spent unresolved, override frequency, progression by reason, and reversals after new evidence. Keep the eligible-record denominator and observation window. Do not optimize one blended score when different errors have different costs.
- False accept: advanced without sufficient evidence.
- False reject: excluded despite later verified fit.
- Unresolved: waiting for a named fact or owner.
- Override: advanced under documented additional evidence.
- Reversal: state changed after a new observation.
- Progression: next action completed under the original rule.
- Suppression: contact prohibited regardless of priority.
- Latency: elapsed time from evidence arrival to responsible action.
- Use OKKI Go to review candidate companies and unlock only selected records, while the published qualification criteria remain owned by the team.
A threshold is local
The balance between false accepts and false rejects depends on sales capacity, research cost, deal value, risk, and the consequence of inappropriate contact. Publish those conditions instead of presenting one number as universal. For a threshold is local, keep the eligible record, criterion version, reviewer action, override reason, later disposition, and reversal evidence together before recalibrating the qualification rule.
Operationalize the criteria without surrendering judgment
Company search and reviewed unlock can support qualification, but a tool output remains a candidate. OKKI Go lets users describe target companies and exclusions, review candidates, unlock selected records, discover contacts, and confirm drafts. Attach the accepted evidence and criterion version to each transition; keep human authority for ambiguity, objections, and send decisions.
- Version the criteria and effective date.
- Train reviewers on accepted, rejected, and unresolved examples.
- Use OKKI Go to review candidates before selective unlock.
- Confirm recipient, subject, and body before sending.
- Route corrections back to the canonical record.
- Audit overrides and reversals each review period.
- Retire criteria that never change an action.
- Recalibrate after a material segment or workflow change.
- Schedule calibration after enough downstream dispositions exist for the local decision; do not borrow a universal threshold from another company.
The qualification standard
A defensible framework tells a reviewer what fact to inspect, what the fact permits, who owns uncertainty, and how later evidence can reverse the state. Precision comes from that operating record, not from a score with unexplained decimals. For the qualification standard, keep the eligible record, criterion version, reviewer action, override reason, later disposition, and reversal evidence together before recalibrating the qualification rule.
Lead qualification criteria should name the next responsible action. If a record cannot change that action, it is not yet a criterion.
Frequently asked questions
What is the difference between a qualification gate and a score?
A gate is an observable fact that changes the next responsible action. A score can rank records without telling anyone what to do when the fact is missing.
How do you catch false accepts in lead qualification?
Review accepted records that later prove wrong and rejected records later restored. Time spent unresolved and override frequency show whether the criteria are operational.
Should every team use the same qualification thresholds?
No. Local capacity, product fit, and follow-up ownership change the threshold. Copying another team’s scoreband without those conditions creates false precision.
What belongs in a six-record teaching set?
Contrasts: verified company with no problem signal, strong signal with unverified role, stale contact, capacity conflict, and a clear reject. The set is a teaching file, not a benchmark.