What You Don't Measure, You Can't Improve

Qualification, depth of discovery, objection handling — and whether the agent ever actually asked for the business. It scores the behaviour that precedes a sale rather than only the sale: a good call on a bad lead is still a good call, and marking it down teaches people to avoid difficult leads.

• BANT, but measuredNeed, budget, authority and timeline each as their own field
• Past the close"Kept selling after they agreed" is a signal of its own
• CompetitorsWho got mentioned lands as a field rather than as an anecdote
• Lead quality apartSo scoring the call never quietly scores the list

Resolution quality, the effort the customer was put to, and the operational signals that say why: repeats, transfers, holds, knowledge gaps and documentation gaps. Diagnosis deliberately outweighs resolution — a symptom treated is a ticket that comes back, and it looks identical to a real fix until it does.

• The causes of effortRepeating, transferring, holding — the things an operation can fix
• Gaps, not blameKnowledge and documentation gaps point at the process
• Expectation setWhether anybody said what happens next, and by when
• SLA riskBefore the clock runs out rather than after

Built around what regulators actually check: identity before any disclosure, third-party leakage, mandatory notices, hardship and pressure. Several items are worded so that TRUE is the bad outcome — deliberately, so that a breach can never come out as a high score.

• Auto-failA missed identity check does not cost points, it fails the conversation
• Third partiesDisclosure to somebody who is not the debtor is its own field
• HardshipIndicated and handled appropriately are two fields, not one
• Polarity flippedOn pressure and leakage, TRUE is the bad answer

Two-identifier verification, triage urgency, red-flag escalation — and whether a non-clinician stayed inside their scope. The pack separates "a red flag was mentioned" from "the red flag was escalated", because that gap is the finding.

• Two identifiersBefore any patient detail, not after
• The gap is the pointMentioned and escalated are separate fields
• Scope of practiceClinical advice from a non-clinician becomes visible
• ConsentCaptured where it has to be captured

Claim capture completeness, coverage and excess explained, fraud signals — and how somebody in the worst week of their year was handled. It scores the handling rather than the outcome: a claim nobody could pay, explained well, is still a good conversation.

• CaptureLoss date, description, documents requested, timeline promised
• ExplainedCoverage and excess as two fields, because people confuse them
• Fraud signalRecorded as an indicator, never as a verdict
• DistressWhat was noticed and what was done about it, scored apart

Suitability, risk warnings, costs and charges, vulnerability and pressure — the items a compliance review asks about first. Including the things that must never be said: an implied guaranteed return is a field of its own.

• SuitabilityAssessed rather than assumed, and understanding checked
• The forbiddenImplied guarantees and applied pressure captured on their own
• CostsDisclosed or not, as an answer with no room in the middle
• VulnerabilityIndicated and handled, again as two fields

Whether the problem was reproduced, the cause found, and the explanation pitched at the person actually on the line — plus what engineering should hear about it. Suspected bugs, missing reproduction steps and raised feature requests leave support as signals somebody can act on.

• ReproducedThe difference between a diagnosis and a guess
• Pitched rightJudged against the customer's level, not the agent's
• Handed onwardSuspected bugs and feature requests leave the ticket
• EscalationWhether it was justified, not merely whether it happened

Booking outcome, details read back, cancellation policy explained, reminder arranged — and a no-show risk read from how the customer actually sounded. Precisely what a row in an appointments table can never tell you.

• Read backName, number and slot confirmed back — or not
• PreparationInstructions and cancellation policy as fields of their own
• AlternativesWhether a second slot was ever put on the table
• No-show riskRead from the conversation, before the slot is wasted

Why they want to leave, whether anybody asked before a discount appeared, and how recoverable they are. The pack also captures whether a service failure was admitted — the most common cause that never shows up in a cancellation report.

• Ask firstRoot cause explored before an offer was put on the table
• Relevant?Offer made and offer relevant are two different fields
• Service failureAdmitted or not — the cause behind the cause
• RecoverableA read you can prioritise a win-back list with

How this customer prefers to be dealt with: communication style, what actually drives their decision, emotional state, trust, and how much detail they want — so the next person to speak to them opens the right way. Every field is a model's read of this one conversation rather than a trait somebody carries, and the confidence that ships with it is deliberately low on short calls.

• This conversationDescribes how somebody was here, not a trait they carry
• With confidenceEvery read carries its own uncertainty beside it
• Short callsThe less conversation there was, the lower it scores itself
• For the next contactA suggested opening, not a label

Composure, adaptability, confidence and script dependence as coaching signals — plus the emotional load a conversation put on the agent, so hostility absorbed and recovery needed become visible instead of invisible. Those load fields describe the call, not the person's ability: ranking people with them points the tool backwards.

• PractisableEvery criterion is a behaviour somebody can change by next week
• Load made visibleWhat a conversation cost, beside what it produced
• Script dependenceA coaching signal, never a disciplinary one
• Aggregate modeRecommended where codetermination asks for it