Outbound Calling in Insurance: Staying Compliant Without Losing Reach

Outbound calling usually sits with the contact centre manager or the compliance team. In my view, it belongs on the leadership agenda.

It sits right between two numbers every insurance board watches: renewal premium and regulatory risk. And both are under pressure at the same time.

Why outbound calling matters to the business

Policies are still lapsing at a rate most boards would rather not think about. In FY25, the best 13th-month persistency in life insurance, measured by number of policies, was 83%. That means even the industry leader lost roughly one in six policies after the first year. At LIC, 13th-month persistency was 67.23% and 61st-month persistency was 48.92% in the first half of FY25.

The phone is still where many of those renewals are saved. SMS and email reminders are cheap, but they are easy to ignore. A conversation lets you deal with the actual reason a customer is drifting: they plan to pay next week, their bank account changed, or the premium went up and nobody explained why.

The problem is that customers have stopped answering unknown numbers. Truecaller’s 2025 India report shows how bad the noise has become:

• 4,168 crore spam calls were identified in India in 2025
• 66% of calls from unknown numbers were spam
• Financial services made up 18% of spam calls
• Spam complaints rose from 13.62 lakh in 2023 to 31.09 lakh in 2025

Every genuine renewal call now competes with that. And mistakes cost more than they used to. With TRAI’s 2025 amendment, IRDAI’s recording rules, the DPDP Act and the FCC’s ruling on AI voices in the US, a careless campaign is no longer just inefficient. It is a regulatory event.

What customers want from an insurer’s call

The research on this is surprisingly consistent. A 2026 survey of US insurance customers by the Independent Insurance Agents & Brokers of America found that 87% still consider a human agent important, and only 6% would rely on AI alone after an accident, storm or major claim.

Yet the same survey found 61% prefer an agent who uses AI. Customers said AI adds the most value for claim status updates, comparing quotes and answering general questions.

The line is clear once you see it. Updates are factual and expected, so customers are happy to hear from an AI that their premium is due on 15 March or that their claim documents have been received. Buying insurance is a financial decision, so customers want to ask questions, push back and trust the answer. An automated voice trying to sell a policy feels like spam, even when it comes from a genuine insurer.

This is not really a technology problem. It is a design decision, and most insurers haven’t made it deliberately.

Which insurance companies are using AI for calls, and with which technology

AI is no longer an experiment in insurance contact centres. Here are the deployments that have been made public, along with the technology behind each one.

The reasons insurers are turning to AI are practical. An AI voice agent can make thousands of reminder calls a day at the right point in each customer’s renewal window. It can speak to policyholders in Hindi and regional languages without separate teams. It makes routine calls cheaper, so human agents can focus on conversations that need judgement, and it follows the approved script every time.

Two patterns stand out. Most deployments are about service, not selling: renewals, reminders, claim status and servicing. And where AI does touch sales, as at Star Health, it guides a human agent rather than replacing one. One caution: most of these numbers come from the vendors themselves, so treat them as best-case results and measure your own baseline.

The split that works: AI for updates, humans for sales

Putting the customer research and the deployments together, this is how I would divide outbound work.

The split also lines up neatly with the rules. Service calls belong on the 1600 series, where selling is not allowed. If your AI only ever makes service calls, it can never slip a sales pitch into one.

The outbound calling rules in one table

Executives don’t need every clause, but they do need to know what has changed.

Customers also have a much easier way to complain. Between April and June 2026, TRAI logged about 1.09 million spam complaints, and 89% of them came through the TRAI DND app. A bad call can now be reported in one tap.

Where AI outbound calling goes wrong

AI repeats whatever it is given, at scale. A bad list or a bad script doesn’t produce one bad call. It produces thousands before anyone notices.

The most common failure is using AI to sell. An automated voice pitching a policy leads to hang-ups, spam reports and eventually blocked numbers, which then hurts the genuine service calls made from the same numbers.

The others are just as avoidable. A voice bot dialling a list nobody consented to invites regulatory action under TRAI or the TCPA. A bot that misstates a waiting period or a benefit creates mis-selling, with the recording as evidence. A renewal reminder that adds an offer breaks the 1600 series rule. And automation that calls the same customer again and again turns a helpful reminder into harassment.

There is also a newer risk that insurers don’t control: fraudsters cloning voices to impersonate insurers. Every scam call like that makes customers a little less likely to trust the genuine one.

How Venturesathi runs compliant outbound calling

Our approach follows the same split, and the discipline is in doing it on every campaign, every day.

We classify every campaign as service or sales before a single call is made. That one decision sets the number series, the script and whether AI can be used at all.

AI handles service calls only. Reminders, claim updates and confirmations go out on 1600 numbers. The AI says upfront that it is an automated assistant calling on the insurer’s behalf, and hands over to a person the moment the customer asks or the conversation needs judgement.

Humans sell, with AI supporting them. Our sales and lapse revival agents get real-time prompts for disclosures, product details and common objections, so they sound informed rather than scripted.

Compliance is built into the system, not left to memory. Lists are scrubbed for consent and DND before every campaign, and opt-outs apply immediately. Calling hours, number series and attempt limits are enforced by the dialler, so a busy day can’t lead to a missed rule.

Every call is reviewed. Speech analytics checks all recordings for missing disclosures, risky phrases and complaints. Our quality team reviews what gets flagged, and the findings feed into coaching and script updates every week. We also track answer and complaint rates per number, so we can fix problems before a number gets flagged as spam.

Recordings are protected with encryption, restricted access, logging and deletion on schedule. Our piece on data privacy in insurance outsourcing explains why the insurer stays accountable even when a partner makes the calls.

Three questions for insurance leadership

If I were sitting with an insurer’s leadership team, I would ask three questions.

Is AI making any of our sales calls? If nobody can answer that quickly, the risk is already in the system.

How many renewals are we losing to unanswered calls rather than to price? Most insurers track lapse rates. Very few track customers who were simply never reached on a number they trusted.

Could we show the consent behind last month’s calls within minutes? In 2026, that is what a regulator will expect.

The insurers who keep their reach won’t be the ones who make the most calls. They’ll be the ones whose calls customers still choose to answer.

Venturesathi runs compliant outbound calling for insurers in India and the US, with AI where it helps and people where it matters. If that’s the balance you want, we can help.