A call scoring rubric for insurance renewal calls

Score an insurance renewal call in the order it runs: verification, what changed, the premium, the cover, the decision, the close. Keep about a dozen yes-or-no lines, let five of them apply only to the renewal types that call for them, mark the rest not applicable instead of failed, and cap auto-fails at three. The table below is ready to copy. The call score is the share of mandatory lines passed among those that applied.

Oleg KulakovCEO, Spellit9 min read

A form written for discovery calls does badly on a renewal. It asks about budget and decision makers, and a customer who has held the policy for three years has no use for either question. The opposite mistake is a generic service form with six lines about courtesy and none about the moments where a renewal turns. This piece builds the form from the renewal itself. For which renewals deserve a conversation at all, start with what the insurance renewal call was for. For how a scorecard in general is built, there is a separate guide. This one stays with renewals.

A renewal call should be scored in the order it runs, in six blocks

The six blocks follow the real sequence of the call. The agent verifies who is on the line and says why the call is happening. Then the agent finds out what changed since last year, deals with the premium, checks the cover, hears the customer's decision and closes with what happens next. A reviewer who scores in that order can replay the recording once from start to finish instead of jumping around the form.

The order also tells you where the weight sits. Verification and the cover read-back protect the customer and the company. The "what changed" and "decision" blocks are where the call can change the outcome. A rubric that spends its lines on greeting style and closing warmth spends them in the wrong place. If you sell in the UK, the premium block has extra weight: FCA rules in force since 1 January 2022 require that a home or motor renewal price is no greater than the price a new customer would be offered, so compliance will care about that block too.

Copy this rubric: twelve lines, three auto-fails, three lines kept out of the score

Every line is a yes-or-no question about what the agent did. The "applies when" column says what has to be true about the renewal before the line counts. The last column is the sentence a reviewer should be able to quote. The lines and sentences are written for this article as a default worth starting from. They are a recommendation, not a measurement from anyone's calls.

BlockLine, as a yes-or-no questionApplies whenKindWhat a passing quote sounds like
OpeningR1. Did the agent verify the customer with the checks your firm requires before discussing the policy?AlwaysMandatory"Before we go through the policy, can I confirm a few details on the account?"
OpeningR2. Did the agent say why they were calling or what the customer asked about?AlwaysScored"I'm calling about your home policy, which renews on the fourteenth."
What changedR3. Did the agent ask what has changed since last year?AlwaysMandatory"Has anything changed at the property since last year, anyone moved in or out, any building work?"
What changedR4. Did the agent ask how a claim made during the term went?A claim was made in the termScored"You made a claim in the spring. How did that go for you?"
PremiumR5. Did the agent give the reason for the premium change together with the new figure?The premium moved materiallyMandatory"It has gone up mainly because of the claim on the policy this year, and the new figure is..."
PremiumR6. Did the agent offer a change in cover or excess as an alternative to a change in price?The customer reacted to the premiumScored"If you would rather hold the premium, we could look at a higher excess."
CoverR7. Did the agent read back what the policy covers and the excess?AlwaysMandatory"So the policy covers the building and contents, with a two-hundred excess."
CoverR8. Did the agent ask whether the cover still fits the customer's situation?AlwaysScored"Does that still match what you need, or has anything moved?"
DecisionR9. Did the customer say in their own words what they want to do?AlwaysMandatory"Yes, go ahead and renew it" or "I'll move it elsewhere"
DecisionR10. Did the agent ask what prompted the wish to leave before answering it?The customer raised leavingScored"Before I answer, what made you look at other options?"
CloseR11. Did the agent say what happens next and on what date?AlwaysMandatory"You will get the documents by email today, and the new term starts on the fourteenth."
CloseR12. Did the agent summarize the changes agreed on the call?Something changed during the callScored"So we have added the extension and raised the excess."

Three auto-fails sit above the table in the form, each checkable in the transcript: discussing policy details before verification, stating a premium or cover term that conflicts with the policy record, and skipping a disclosure your compliance team requires on this call type. Three is deliberately few. If a third of calls trip one, the list is too broad or one of its lines describes something the agent does not control.

Three lines are recorded and stay out of the score: the outcome (renewed, changed, lapsed), the reason the customer gave for leaving, and whether the call was inbound about something else. The outcome is what the other lines are meant to explain, so ranking it against them is circular. Why that matters for the whole form is covered in the general scorecard guide.

Five of the twelve lines should switch on only for the renewal types that call for them

Five of the twelve lines are conditional: R4, R5, R6, R10 and R12. Which ones switch on depends on what happened to the policy. A quiet renewal triggers almost none of them. A renewal after a claim and a premium rise triggers most.

Situation on the policyLines that switch on
Nothing changed, premium roughly flatNone of the conditional lines
A claim was made in the termR4
The premium moved materiallyR5, then R6 if the customer reacts
The customer says they are thinking of leavingR10
Anything was changed during the callR12

The consequence of skipping this is practical. If R4 is scored as a no on every call without a claim, the agents who handle the quietest renewals look like the weakest, and the review team spends its time defending numbers that measure the book rather than the agent. Leave the line out of both halves of the fraction instead, and say so on the report. Spellit Actions leaves lines that did not apply out of the calculation, and the same logic works in a spreadsheet.

Wording lines are a different case. If your compliance team requires exact words at verification or disclosure, that is script adherence, and it belongs on the mandatory side. The conversational lines, such as R3 and R8, should be scored on whether the question was asked and answered, not on matching a phrase. Script adherence is worth reading before deciding which of your lines are which.

Two calls scored with the rubric: a quiet renewal and a premium rise

Both examples below are constructed to show the arithmetic. They are not recordings and not client data.

The first is a home renewal where nothing changed and the premium is flat. R1, R2, R3, R7, R8, R9 and R11 apply, which is seven lines. The agent verified the customer, asked what changed, read back the cover and closed with a date, but never asked whether the cover still fit, so R8 fails. All five mandatory lines pass. The call is clean, short, and mostly not applicable: five of twelve lines were never in play.

The second is a renewal after a claim, with a premium rise. The customer says, "That's a big jump from last year." The agent answers, "That's what the system has come back with," and moves on to the cover read-back. R5 fails, because no reason was given with the figure, and R6 fails, because no alternative was offered. The customer ends with "I'll have a look around", so R9 fails as well. R10 is not applicable, since leaving was never named. Six mandatory lines applied, and four passed. The review shows the quote, with its timestamp, next to R5, and the plan for the agent is one sentence to try: "It has gone up mainly because of the claim this year. If you would rather hold the figure, we can look at a higher excess."

That second case is the one to compare against your current form. A form that tells you the call scored 67% has said very little. A form that points to the moment the customer said "big jump" and the sentence that followed it has told the coach what to work on.

Before the form goes live, have two reviewers score the same five calls and compare them line by line. Percent agreement alone can overstate how well they agree, because reviewers sometimes agree by guessing; McHugh's explainer on Cohen's kappa covers the measure that corrects for chance.

Keep the outcome and the tone out of the score, and read them next to it

Outcome and tone are useful, and they stay on the report beside the score. The outcome tells you where the renewals are lost. The tone of the customer, which the glossary entry on call sentiment treats as a weak signal on its own, helps a coach find the calls worth listening to first. Neither should move the number, because the number is there to say what the agent did.

Read them side by side. A high score with a lapsed renewal points at the rubric: the lines may be checking the wrong things. A low score with a renewal that went through points at the agent: the customer stayed in spite of the call. Both readings need the outcome kept apart from the score, so that the comparison can be made at all.

What to do next

Pick five recordings from last month's renewals, one from each situation in the second table if you have them. Score each by hand against the twelve lines and write down how many lines applied. If most calls leave fewer than half the lines in play, the form is built for a different kind of call.

Which renewal calls your agency records, and which of them can be reviewed, is for you and your compliance team to decide. The insurance page shows where Spellit draws its line: first enquiries, service and administrative calls. In Spellit Actions the scoring is built from your own checklist, and every verdict arrives with the quote and timestamp from the recording. If your team already has a renewal checklist, that is the starting point. If you want to see it on your own calls, bring ten real recordings to a demo.

Key points
  • A renewal call is mostly conditional. Whether the agent should ask about a claim, explain a premium rise or handle a request to leave depends on what happened to the policy, so a fixed form fails agents for things the call never needed.
  • Order the form the way the call runs. Six blocks are enough: opening, what changed, premium, cover, decision, close.
  • Mark a conditional line not applicable when its trigger is absent. A blank is not a zero.
  • Keep outcomes and the stated reason for leaving next to the score, not inside it.
  • Put a quote with a timestamp behind every verdict. A line you cannot picture a passing sentence for is not ready.
See the demo now. Then run it on yours.

A live demo of our products and a real conversation about the growth problem you need to solve.

Book a demo

FAQ

How many lines should an insurance renewal scorecard have?

A default worth starting from is about a dozen yes-or-no lines, with no more than six or seven of them mandatory, and three auto-fails at most. That is a recommendation, not a measurement. Five of the lines should be conditional, so a quiet renewal is judged on seven or so and a complicated one on ten or eleven. The test of length is whether a manager can say the mandatory lines from memory. If nobody can list them without opening the form, there are too many, and the coaching conversation will drift to whichever line the reviewer happens to remember.

Should the rubric score whether the customer renewed?

No. Record the outcome next to the score and keep it out of the calculation. Renewal is the result the other lines are supposed to explain, so putting it inside the score makes the score partly circular: agents with an easy book look strong and agents with a hard one look weak. The outcome is still worth reading. A pattern of high scores with lapsed renewals tells you the lines check the wrong things, and a pattern of low scores with renewals that went through tells you the customers are staying despite the call, which is a different problem and a different conversation.

How do you score a renewal call where nothing changed?

Score only the lines that applied and mark the rest not applicable. In the quiet example above, seven of twelve lines were in play, and the five that were not stay out of both halves of the fraction. A short, uneventful call with all mandatory lines passed is a good call, and the report should not mark it down for questions the situation never called for. The common error is the opposite one, where every conditional line is scored as a no, and the agents on the calmest part of the book end up with the lowest numbers.

How do you check that two reviewers score a renewal call the same way?

Have both reviewers score the same five recordings independently, then compare line by line, not by total. Disagreement usually sits in two places: lines whose passing sentence nobody wrote down, and lines whose trigger is unclear, such as what counts as a material premium move. Fix those by adding the passing quote to the form and by writing the trigger as a threshold your team agrees on. Repeat when you change the form. Two reviewers who agree on the total but not on the lines are agreeing by accident.