What the insurance renewal call was for, and why it is disappearing before anyone found out
The renewal conversation is not being lost. It is being removed. In the United Kingdom, 70% of home policies are now sold with auto-renewal attached by default, up from 54%, and the consultancies advising insurers expect the contact center to move off simple transactions. That advice is mostly right. The part worth arguing about is the small number of renewals where a voice still changes the outcome, and how you find them. Spellit Actions is where we read those recordings.
Every article on this subject opens the same way: renewals are lost on price, the answer is to call earlier, here are scripts. The scripts are fine. The framing assumes a conversation is happening, and across most of the market that assumption stopped being true while nobody was looking.
This piece takes the other side. If you are automating renewals, and you should be, the question is what you are deleting along the way.
The setting on the policy decides retention more than the call does
The share of home policies carrying auto-renewal at inception rose from 54% across 2019 to 2021 to 70% across 2022 to the first quarter of 2024. The Financial Conduct Authority's 2025 evaluation of its general insurance price remedies covered sixteen home insurers and thirteen motor insurers, roughly 80% of home gross written premium, with a sample of over four and a half million home policies before the intervention.
Attrition splits sharply along that setting. Home policies without auto-renewal showed attrition between 32.2% and 38.5% depending on the year. With auto-renewal, between 19.6% and 21.5%. Motor showed the same shape at higher levels, roughly 49% to 54% against 33% to 35%. What a conversation adds on top of that is measured nowhere in this dataset.
The regulator draws on that difference in levels and says the data corroborates its hypothesis. What it excludes from its analysis is the change in attrition by auto-renewal status over time, because 2023 price inflation pushed people out of the market and hit customers without auto-renewal harder. It also holds no data on why anybody decided anything. It is a United Kingdom dataset. Use it to size the default, and keep it out of the business case for or against a renewal program. If your own book says the call moves renewals and you can show it, we would like to be wrong about this. Bring the counter-example.
The advice you are following says to remove the call, and it is mostly right
Bain's global study of insurance customer behavior, built on a survey of 28,765 consumers across fourteen countries, places renewing a policy among the episodes with high digital adoption and low failure rates, and expects the contact center to move away from working transactional requests by rote script toward solving complicated problems. Renewal is a simple transaction for the overwhelming majority of policies, and simple transactions belong in software.
There is no honest way to argue with that for a policy that has not changed. If nothing about the risk, the address, the vehicle or the household is different, a renewal call is a person reading out a form to somebody who has no decision to make. Removing it loses nothing worth keeping: it retires a ritual, and the hours go back to the calls that need judgment. That is the shape of what call analysis is for in insurance.
What that expectation does not settle is the exception. Bain's episode analysis is drawn from a United States auto insurance sample of fifteen thousand, and it measures how customers feel about the channel, which is a different question from what the conversation was worth when it happened. A low failure rate on a simple episode proves that automation works for simple episodes. Whether every renewal is a simple episode is a separate question, and one your own recordings answer.
People who stay are not always loyal
Staying and choosing are different acts, and a retention number cannot tell them apart. The regulator's own figures make the point without any appeal to psychology: attrition roughly halves when the policy renews itself, and nobody in that group necessarily decided anything at all.
A retention number does not carry the auto-renewal split, so the same figure means two different things and the report does not say which. A book with strong renewal rates and heavy auto-renewal penetration is reporting the strength of its defaults. A book in a market with active price comparison and manual renewal is reporting something else entirely, and nothing in the number distinguishes them.
It also sets a limit on what any conversation can achieve. If most of your retention comes from customers not looking, then improving the renewal call moves a small share of the book, and it is worth knowing that before funding a program. The case for the call has to be made on the customers who do look.
Price gets them in the door, service and the claim decide the renewal
What does a customer who is actually paying attention respond to? J.D. Power's 2025 study of United States auto insurance, covering 48,121 customers surveyed between May 2024 and April 2025, reports that good rates and low cost are the top reasons people buy, while good service and a positive claims experience are the top drivers of retention at renewal. In their model of what drives satisfaction overall, level of trust ranks above price for coverage.
This is a vendor's stated finding rather than a reproducible result: the dimensions are published in order of importance, but the weights and the method behind that ordering are not, the study is United States only, and it measures stated renewal intention, which the regulatory data above shows behaves very differently from observed behavior. It remains the clearest available answer to the question of what the conversation was ever for.
The answer has nothing to do with persuasion. A claim, a change in circumstances or a premium move creates a moment where the customer genuinely re-decides, and a form cannot tell you that moment has arrived. Somebody saying "we had a bit of trouble last winter" in the middle of an administrative call is the whole signal, and it never reaches a renewal notice. Ask what a report built for renewals rather than for sales looks like.
Four renewals are worth a call, and the renewal date cannot find them
Four renewals are worth a person: a claim during the term, a premium move, a change in circumstances, and an inbound call about something else. In each of them the policy stopped meaning what it meant last year. Automate every other renewal.
No dataset splits renewals this way. This is our read from the calls we score.
| Renewal | Why a form struggles with it | What the conversation has to establish |
|---|---|---|
| A claim was made during the term | The customer has already re-decided once, silently | Whether the experience matched what they were told at sale |
| The premium moved by more than 10% | The notice explains the number and leaves out the reason | Whether the reason is understood, and whether cover should move as well as price |
| Circumstances changed and nobody updated the policy | The form asks; a person hears the hesitation | What is now uninsured, before a claim finds it |
| The customer called in about something else | An inbound call is attention, and attention is rare | Whether the thing they called about is the thing that matters |
The fourth row costs the most. An inbound call from a customer who has not thought about the policy all year is the only moment when both parties are present and paying attention. Treating it as a ticket to close quickly is how a retention conversation gets thrown away by the department that measures handling time.
None of those four can be identified from the renewal date. Three of them leave a trace in what was said, and the fourth leaves a trace in when the customer called, the same way a viewing follow-up in property turns on a question nobody asked out loud. That is the practical reason to keep reviewing the conversations you do have after automating the ones you no longer need.
What we got wrong building this
We assumed every business wanted one measure of a good conversation, and we built accordingly. Insurance broke that assumption faster than any other market we worked in, because a qualification call, a renewal call and a claim call in the same company share almost nothing. A qualification call cannot be scored on sales technique, and a renewal call scored on the qualification rubric comes back looking excellent, since the rep followed a script that had nothing to do with the situation.
The fix was to let each conversation type carry its own rubric, and to leave the parts that do not apply out of the report entirely rather than show them as marks that were never earned.
The second thing we got wrong was subtler. We let unknown roles fall back to the sales rubric, so anybody the system did not recognize was marked down for not closing: a service agent handling a renewal, a claims handler, a receptionist. The scores were confidently wrong in a direction that made service staff look like poor salespeople.
What to do next
Pull your renewal book and split it by the four rows in the table above. If the split has never been done, do it before the retention budget and the automation roadmap are set, because it decides both.
Then take last month's inbound calls that were not about renewals and check how many were within ninety days of one. Those conversations already happened, and somebody already paid for them.
Start with the calls you already have. Give us a month of inbound service calls and we will mark the ones that were a renewal conversation wearing a service ticket, with the sentence that gives each one away. Forty-five minutes on your own recordings.
- Attrition differs sharply by whether a policy auto-renews. No dataset here measures what a conversation adds on top of that, and that gap is worth naming.
- The industry's own advice is to take renewals out of the contact center. For most policies that advice is correct.
- Customers who stay are often not loyal. They simply did not look, which is a different thing to build on.
- Where price gets somebody in the door, service and the claim decide whether they renew.
- Automate the renewal. Keep the call for the four situations where the policy changed meaning.
A live demo of our products and a real conversation about the growth problem you need to solve.
Book a demo