Insights · Insurance AI Weekly

Your retention problem starts in year one: what new auto shopping data means for insurer analytics

·

,

Most auto insurers treat retention as a renewal problem. New data says the real risk comes much earlier. First-year customers account for a large share of lost policies and an even larger share of people leaving the market, and the analytics most carriers run are not built to catch them.

What happened

On November 12, LexisNexis Risk Solutions released its U.S. Insurance Demand Meter for the third quarter of 2026. Auto shopping grew 0.7% year over year, down from 1.4% in Q2, and new business grew 3.2%. Over the previous 12 months, 47.2% of policies in force were shopped at least once, just below the record 47.3% set in Q1. Non-standard shopping grew 5.4%, its first positive quarter of the year, while standard shopping slowed to 1.3%. Only 28 states showed shopping growth, down from 42 in Q2. LexisNexis says the data covers nearly 90% of U.S. insurance shopping activity. (LexisNexis Risk Solutions)

The report also added a new analysis of policy losses. Policies lost over a 60-day window equal about 18% of active policies on an annual basis. About 70% of those lost policies stay in the market (switching carriers, returning to the original carrier or dropping a second policy), and about 30% leave the market entirely. First-year policyholders make up 40% of lost policies and more than half of market exits. LexisNexis also said rate increases may no longer be the only main reason people shop, with broader affordability pressure playing a bigger role. (Insurance Business)

Our take

The first-year figure is the one to act on. If 40% of lost policies come from customers in their first year, then retention models that score the whole book at renewal are looking in the wrong place at the wrong time. By the time a new customer reaches their first renewal, many of the ones likely to leave have already shopped, and some have already gone.

This is a data and analytics problem more than a pricing problem. Most carriers already hold the signals that flag a shaky new customer: how they were acquired, payment behavior in the first few months, coverage changes, service contacts, a quote that only just won on price. What is usually missing is a model and a workflow that look at those signals in the first 60 to 90 days, when outreach can still make a difference.

The affordability point matters too. If customers shop because their overall budget is under pressure rather than because of a single rate increase, then a churn model built mainly on rate-change size will miss a growing share of the risk.

What to build into your retention analytics

  • A separate early-tenure model. Score first-year customers on their own, using signals from the first months instead of renewal-time features.
  • Acquisition context. Feed in channel, quote competitiveness and discount use at purchase. Customers won on price alone are the likeliest to leave on price.
  • Affordability signals. Add payment-plan changes, missed or late payments and coverage reductions alongside rate-change size.
  • Exit vs. switch. Track whether lost customers moved to a competitor or left the market. The fixes are different.
  • A 60-day action trigger. Connect the model to a defined outreach step for agents or service teams, and measure whether it changes outcomes.

Shopping levels are holding near record highs, so the market is not going to calm down on its own. The carriers that keep more of their new business will be the ones whose analytics pay attention in the first year, not just at the first renewal.

← All Insurance AI Weekly posts

Get the weekly brief

Get our short weekly roundup of insurance software and AI news in your inbox.

Putting AI to work in insurance?

We help insurers, MGAs and agencies modernize core systems, get their data ready and adopt AI responsibly.