Call scoring: a score out of 10, with evidence

A score without an explanation is useless. An agent who gets "6/10" without knowing why either disputes it or loses heart. A manager who can't check an automated score soon stops looking at it.

NextCall's call scoring rests on one rule: every score must be verifiable in the conversation itself.

Your scorecard, not a generic one

NextCall doesn't score your calls against a standard template. It applies your script or your evaluation scorecard, the one your managers already use or the one you build with us.

One scorecard per call type

A sales call, a sales follow-up and an after-sales call don't share the same goals. So you create a scorecard for each call type, and each call is scored with the one that matches it.

Weighted steps

Not every step carries the same weight. In a sales call, needs discovery often matters more than the closing pleasantries. You set the weight of each step, and the overall score reflects it.

Bonuses and penalties

Some behaviours deserve to be highlighted beyond the steps themselves: a firm commitment secured from the customer (bonus), a mandatory disclosure left out (penalty). You define them in the scorecard.

If you're starting from scratch, our article call evaluation scorecard: the 10 criteria that matter gives you a base to adapt.

The scale: a score out of 10 for each step

Each step is scored out of 10. The scale is fine enough to tell a decent call from a very good one, and simple enough to understand without training.

The overall score combines the steps according to their weight, then applies bonuses and penalties. Leadership then tracks the share of calls rated good or excellent.

No opportunity, no penalty

A criterion that had no reason to come up in the call doesn't lower the score. If the customer raised no objection, the agent isn't penalised on objection handling. This rule prevents unfair scores that demotivate teams.

Proof by quotation

This is the core of NextCall scoring. Every score comes with the exact sentence spoken during the call that justifies it.

Example from a sales call

A sales rep calls a prospect after a quote request. On the "Propose a next step" stage, NextCall gives 9/10 with the quote:

> "How about we speak again on Thursday at 2pm with your business partner to sign off the quote, does that work for you?"

On the "Needs discovery" stage, the score is 4/10, with the quote:

> "Right, I'll send you our standard offer and you let me know."

The agent immediately understands what worked and what was missing. The manager can verify the score in ten seconds, without listening to the whole call again.

A score without a quote is flagged

If the AI gives a score it cannot back up with a passage from the call, that score is marked "to be checked". The manager knows where to focus, and nobody is judged on an impression.

Calibrating the AI against your managers

A score is only useful if your managers agree with it. That is what calibration is for.

How calibration works

1. Your managers blind-score a sample of calls using the scorecard.
2. NextCall compares their scores with the AI's.
3. Gaps are flagged, along with inconsistencies between the managers themselves.
4. The scorecard is corrected: a criterion reworded, an example added, a weighting revised.
5. The calls are re-run with the new scorecard to check that the gap narrows.

Calibration often reveals something useful in its own right: two managers who don't score the same way. Fixing that makes evaluation fairer for the whole team.

Versioned scorecards

Every change to a scorecard creates a new version. You know which version a call was scored with, so you can compare periods without confusion.

What the score makes possible

The score isn't an end in itself. It feeds:

• advice for the next call and each agent's main focus area, covered on the team phone coaching page;
• the urgent calls list for managers, below a threshold you choose;
• tracking of team skill gaps criterion by criterion;
• customer service quality indicators, presented on the call quality assurance page;
• leadership reports: average score, trends, quality by day, time or call duration.

For an overview of the platform, see the call analysis software page. Calls arrive automatically from your phone system, as explained on the Aircall, Ringover, Diabolocom call analysis page.

Book a demo to see scoring applied to your own calls.

Frequently asked questions (FAQ)

Can I use my current evaluation scorecard?

Yes. NextCall applies your script or your scorecard. It is imported, adjusted during calibration if needed, then versioned.

What happens if I disagree with a score?

Every score comes with the quote that justifies it, so it can be checked quickly. Recurring disagreements are dealt with during calibration by correcting the scorecard.

Is an agent penalised if a criterion doesn't apply?

No. A criterion with no opportunity to arise in the call doesn't count against the score.

Does the customer's communication style change the score?

No. The DISC style hypothesis about the customer is there to help adapt the next conversation. It has no effect on the score.