Most insurers are good at two moments in a customer's life: the sale and, if they are lucky, the renewal. Between them lie months or years of silence, and in our view that silence is the most expensive habit in retail insurance.
There is a running joke in the Polish market that insurers only remember a customer when the policy is about to end. The joke is generous. Often the customer gets an email saying the policy is expiring, and nothing else: no agent calls, no automated call, no conversation about whether the cover still fits. We have worked with insurers on receivables, renewals and benefits for years, and the pattern is the same. The contact strategy is built around the contract, not around the person who holds it.
What the silence costs#
Silence looks free. It does not appear on any budget line. Its cost shows up elsewhere, in places that are rarely connected to the contact strategy:
- Renewals decided elsewhere. By the time the expiry email arrives, the customer has often already received a quote from a comparison site or another insurer. The renewal conversation starts after the decision has been made.
- Cross-sell that never happens. A customer with motor insurance may well need home cover, and the other way round. Nobody asks, because nobody calls a customer who is not about to leave.
- Receivables that drift. A missed premium payment is a different conversation in week one than in month three. Without early contact, a small arrear becomes a lapse.
- Benefits nobody uses well. In health and assistance products, customers who do not know how to use their cover experience it as worthless, and that experience is what they remember at renewal.
Each of these is a separate problem with a separate owner: sales, retention, finance, claims or benefits. Together they describe one thing. The insurer has a relationship it only activates under pressure.
The industry knows it has a transformation gap#
Insurers are not short of ambition on AI. KPMG's study of AI value in insurance, published on 29 September 2026 and based on insurance leaders in 20 countries surveyed in May 2026, found that 44% of respondents placed their organisations in the top quartile for AI transformation. The same study found that full redesign of processes around AI was rare: none had done it for sales and distribution or underwriting, and only a small minority for policy servicing and claims. Only 11% said they had the strong data foundations and governance needed to scale AI.
| Finding | Share of respondents |
|---|---|
| Place themselves in the top quartile for AI transformation | 44% |
| Have fully redesigned sales and distribution or underwriting around AI | None |
| Have the strong data foundations and governance needed to scale AI | 11% |
Self-assessment by insurance leaders from organisations with 500+ employees, 20 countries, fieldwork 20–29 May 2026. Total sample size not stated in the release; not specific to customer contact.
We read this as a gap between activity and redesign. Insurers are using AI for content and routine tasks, but the processes that touch the customer through the life of a policy have mostly not been rebuilt. That matches what we see. The technology to call every policyholder at the right moment exists. The process that decides who to call, when, about what and with which answers approved usually does not.
Contact through the life of the policy#
Our position is simple: a policy is a multi-year relationship, and the contact strategy should follow it. That does not mean calling customers for the sake of it. It means a small number of well-chosen moments where a short conversation changes something:
- Shortly after the sale: confirming the customer understands what they bought and how to use it, especially for health and assistance products.
- At a missed payment: an early, polite contact that establishes whether it was an oversight, a problem with the payment method or a sign of difficulty, and agrees a next step.
- At a life event the insurer already knows about: a new car on a motor policy, a new address, a second product enquiry. These are natural moments to ask about related cover.
- Well before renewal: a conversation, not an email, at a point when the customer has not yet looked elsewhere, with the option of a call from an adviser if they want one.
Each of these is short and has a clear purpose. Each happens inside agreed hours, with a limited number of attempts. The customer is never surprised by who is calling: the conversation opens with the insurer's name and makes clear that it is an AI assistant speaking on its behalf.
Why a voice, not another email#
Insurers default to email because it is cheap and leaves a record. Both are true, and neither makes it effective. An expiry notice competes with every other message in the inbox and asks nothing of the customer. It cannot hear that the customer has sold the car, that the premium went up and they are annoyed, or that they would happily add home cover if someone explained it in two sentences.
A short conversation does all of that. It also produces an answer the insurer can act on: will renew, wants an adviser, has changed circumstances, is shopping around, does not want to be called again. Those are business statuses that sales and retention teams can work with the same day. An unopened email produces nothing.
Cost has historically been the reason insurers did not call. Putting an adviser on the phone to every policyholder at four moments in the policy's life was never going to pay. That constraint has changed. When the first contact is handled by an AI call centre and advisers only speak to customers who asked for them, the economics of the relationship look different, and the insurer's sales team spends its time on people who want to talk.
Handing over to an adviser without flooding the team#
A cross-sell or renewal conversation often ends with a customer who wants to talk to a person. That transfer has to respect the capacity and working hours of the insurer's own team. Sending more live transfers than advisers can take is worse than sending none: the customer waits, the adviser is rushed and the interest cools. So transfers are paced to the team, and outside its hours the customer is offered a callback at a time that suits them.
Receivables are where the case is easiest to prove#
Of the four moments, overdue premiums are the easiest to measure, which is why insurers often start there. The goal of an early receivables call is not pressure. It is a specific answer, at the first contact, about what happens next: paid already, will pay by a date, needs a different payment method, wants to cancel, is in difficulty and should speak to someone.
A result like this matters beyond receivables. It suggests that policyholders will take a short, direct call from their insurer's AI assistant and give a clear answer, which is the same willingness a renewal or cross-sell conversation depends on. Each product line still has to be tested on its own base before anyone assumes the effect carries over.
What contact will not fix#
We should be clear about the limits. The biggest economic problem for many insurers is the claims ratio, and customer contact has little influence on it. A well-run conversation may improve how claims and settlements are handled at the margins, but we do not promise an effect on claims costs and would be suspicious of anyone who did.
Contact also cannot rescue a product that does not fit, or a price that is out of line with the market. What it can do is make sure the customer hears from the insurer before they hear from a competitor, that small payment problems are caught while they are small, and that people who would buy more cover are actually asked.
Why this is a process question, not a tool question#
An insurer could buy a voice platform tomorrow and still not solve this. The hard work lies in deciding which policyholders to contact at which moment, which answers the conversation may give, how the data moves between the policy system and the calling process, what happens to a customer who wants a person, and how the result is measured against a comparable group that was not contacted. Those decisions cut across sales, retention, finance and compliance, and they need someone to own them for years, not for a pilot.
There is a compliance layer too, and it differs by purpose. A call about an overdue premium rests on the existing contract. A call offering a new product needs a legal basis for marketing contact, which the insurer holds and documents. Data used to choose the moment, such as a known life event, must be data the insurer may use for that purpose. We act as a processor within the insurer's own framework, under a data processing agreement, and every change to what the conversation says is versioned and approved by the insurer before it goes live. Getting this right at the design stage is far cheaper than discovering it after the first complaint.
That is why we run these processes ourselves rather than handing over software. We design the conversations with the insurer, run them, listen to them every day and meet the insurer regularly to decide the next base, the next product line or the next moment in the policy's life to cover.
We play on the client's side of the table. – Dawid Wójcicki, Founder & CEO of Vee
What to ask your own organisation#
- Apart from the sale and the renewal notice, when does a typical policyholder hear from us by voice?
- How many days after a missed premium does the first conversation happen, if at all?
- Which life events do we already know about and never act on?
- Who owns the policyholder relationship between sale and renewal, and how is that person measured?
- Can we measure the effect of contact against a comparable group of policyholders who were not contacted?
If the honest answer to the first question is 'never', the opportunity is not a new channel. It is remembering the customer you already have.
Follow The First Second
Occasional issues on the mechanics and economics of AI-powered customer contact. No noise, no vendor news.
Follow via RSS