After a call ends, a score appears on the record. The team looks at it and says "positive call". Then they work that list and get nowhere. The score was not wrong; it was read wrong.
Two different questions, one number
There are two separate questions you can ask about a call:
- Tone: How did it go? Was the customer polite, tense, did they raise their voice?
- Interest: How close are they? How ready for a purchase, an appointment, the next step?
These two are frequently independent. A perfectly polite customer can say "thank you, we are not considering it right now": tone excellent, interest zero. Conversely, a customer who gets irritated on the phone may hold the highest purchase intent — because they care.
Reading the score as interest is what serves a sales operation. "How pleasant it was" is valuable for quality assurance, not for prioritisation.
When there is no score
A call without a score does not mean a low score. An unanalysed call — too short, no recording, or an analysis backlog — sits with an empty score.
The classic trap here is treating empty as zero. A list that does so pushes never-analysed calls to the top as "most negative", and the team spends days on the wrong records.
The right behaviour: show an unanalysed call as its own state — "not evaluated yet".
Score staleness
If you have had four calls with a customer, which one is the person's score? In most setups it should be the most recent, because interest changes over time.
A common failure here: the person's score is left over from an old call and new calls do not update it. The symptom is easy to spot — the last call on the record is yesterday, while the score matches a date months ago. Sorted by interest, those records sit in the wrong place.
What to use the score for
- Prioritisation. High interest means call first.
- Campaign targeting. The middle band is the most productive target for a campaign; the high ones are already with the team, the low ones are a waste of time.
- Trend over time. If one person's score is declining, the customer is drifting away; that alone is a warning.
What not to use it for
- Measuring rep performance. A rep who handles difficult customers will naturally show lower scores.
- Judging from a single call. A verdict on a person comes from contact history, not one conversation.
How to explain it to the team
This distinction is not a technical detail; it is a daily usage matter. Labelling the score in the panel as "interest" rather than "sentiment" does more than training. When the name is right, misreading declines by itself.
Tomorrow: the moment a conversation passes to a human — handing over without losing context.