Engagement metrics for low-touch policyholders

Bind is where most insurance measurement stops. The funnelfunnelThe customer journey from awareness to purchase, typically Awareness, Interest, Consideration, Decision, Action, with prospects narrowing at each stage.View full definition → closes, the acquisition costacquisition costCustomer Acquisition Cost (CAC) is the total sales and marketing spend divided by the number of new customers gained in a period. It measures how efficiently you grow.View full definition → is booked, and the customer disappears into a twelve-month gap with one scheduled decision at the end of it. A personal auto policyholder might open the app twice in that year: once to pull an ID card before a road trip, once when the renewal notice arrives. Marketing still owns the renewal number and has almost nothing to count.

So the post-bind period gets measured by proxy. The real work is picking proxies that predict the renewal decision rather than ones that flatter a dashboard.

Why the post-bind period resists measurement

In e-commerce or media, engagement means frequent voluntary interaction: daily app opens, browsing sessions, content consumed. Insurance is structurally different:

• Low purchase frequency: most personal lines (auto, home, life) renew annually.
• Low perceived need to interact: no claim, no reason to log in.
• High-stakes but rare touchpoints: a claim, a renewal, a life event (marriage, new car, new baby).

So marketers track a narrow set of behaviours that correlate with retention and cross-sell readiness, not raw activity volume.

The three core proxies

1. App login frequency

A login without a transaction still says the brand is present. Carriers track monthly active users (MAU) among policyholders, not just prospects. A rough range from industry commentary, including J.D. Power's insurance digital experience studies, is 15-25% MAU among personal auto policyholders (estimate, varies by carrier and app maturity, 2024-2025 reporting). Beware cross-carrier comparison: Allstate runs its Drivewise telematics programme inside the main app, so the app has a reason to exist on days the customer is not thinking about insurance. That lifts the MAU floor without meaning the brand is better remembered.

2. Policy-document opens

When a policyholder opens a digital declarations page or ID card, that is a document-open event, captured in app analytics or portal logs. Spikes around renewal season or after a rate notice are expected. Spikes at random times can indicate a triggering event (a fender-bender, a new lease) worth a proactive call.

3. Renewal-email interaction

Open rates and click-through rates (CTRCTRClick-Through Rate (CTR) is the percentage of people who click a link, ad, or call to action out of those who viewed it.View full definition →) on renewal notices are the closest thing to an intent signal after bind, even though most policies auto-renew unless cancelled. Sector email benchmarks from Mailchimp, which sells the sending tooling these aggregates come from, sit around 21-24% open rate and 2-3% CTR (estimate, 2024 data, varies widely by list quality).

Building a composite engagement score

Individually each proxy is noisy. A login might be an accidental app tap. A document open might be a curious spouse. So mature teams build a composite engagement score, weighting each signal by its measured correlation with lapse.

engagement_score = (0.3 * login_freq_normalized)
                  + (0.3 * doc_open_normalized)
                  + (0.4 * renewal_email_ctr_normalized)

Each component is normalised (0 to 1, usually percentile rank within the book) before weighting. Weights are calibrated against last year's churn: whichever signal separated lapsers best gets more weight.

Worked example

A mid-size carrier with 100,000 personal auto policyholders. Over one quarter:

• 18,000 logged into the app at least once (18% MAU)
• 12,000 opened policy documents digitally (12%)
• Renewal emails to 25,000 policyholders up for renewal; 5,500 opened (22%), 600 clicked through (2.4% CTR)

Normalised against a benchmark cohort at 20% MAU, 15% doc-open, 3% CTR:

• Login score: 18/20 = 0.90
• Doc-open score: 12/15 = 0.80
• CTR score: 2.4/3 = 0.80

Composite = (0.3 × 0.90) + (0.3 × 0.80) + (0.4 × 0.80) = 0.83

Below 1.0 means this book engages under benchmark, which should trigger a look at onboarding sequences, app UX, or renewal timing.

When the same signal means the opposite thing

The dangerous property of these proxies: their sign flips with context.

A login spike in the fortnight before renewal is often a shopping signal, not loyalty. The customer received a price and is checking it against a comparison site. Read naively, that cohort looks highly engaged right up to the month it leaves. Split logins into pre-notice and post-notice windows and the two behave like different variables.

Rate change is the bigger confounder. Allstate, like most US personal auto carriers, pushed through double-digit rate increases across 2022 and 2023. In a book like that, document opens and lapses rise together, because both are downstream of the same letter. Fit a lapse model on that year without a rate-change feature and engagement will appear to *cause* churn, and the weights invert. Any composite score recalibrated annually inherits the rate environment of the year it was fitted, which is why a score tuned in a hard market misprices a soft one.

The dormant loyal

Zero engagement is not uniformly bad news. A policyholder on auto-pay and paperless billing, or a homeowners policy paid through a mortgage escrow account, can go three years without a login and renew every time. Inertia is doing the retention work, and a re-engagement campaign aimed at that segment mostly reminds people they have a bill.

Practical split for the silent tail: separate dormant-loyal from dormant-at-risk using payment method, tenure and the size of the last rate change. A five-year customer on auto-pay facing a 2% increase needs nothing. A first-term customer paying manually who faces a 12% increase and has not opened an email in 90 days is the one worth a phone call.

Products that engineer engagement rather than infer it

The proxy problem partly disappears if the product itself demands interaction. Vitality built its life and health proposition around verified activity, with rewards tied to gym visits and tracked exercise, so participation and reward redemption are directly observable rather than inferred from app taps. Ping An works the same idea at scale through its health and services platforms, with hundreds of millions of registered users feeding a book where contracts per retail customer sit around three, so an engagement decline shows up as fewer services touched rather than as silence.

The second-order effect is a selection confound. People who join a rewards programme were already more likely to renew, so the engagement score partly measures the customer's disposition rather than the marketing's effect. Anyone reporting an engagement lift against retention should hold the persistency figures from the benchmarking lesson beside it and check whether the engaged cohort was simply a better cohort to begin with.

Why these proxies earn their place

Two things marketing gets measured on depend on them:

• Lapse risk: no logins and no renewal-email opens in the 90 days before renewal is a real predictor, and it is cheap to compute. It sits in lapse-propensity models as three or four features alongside price change and tenure.
• Cross-sell readiness: an auto document open shortly after a homeowners portal session is a bundling candidate. Engagement spikes often trail a life event by weeks.

Both feed the multi-year value model the CLVCLVLifetime Value: the total revenue (or profit) a customer generates throughout their entire relationship with your business.View full definition → lesson builds, which is where these signals get priced.

Benchmarks: US and Europe snapshots

Insurance engagement data is fragmented and rarely published in standard form. Treat these as directional estimates:

• US personal lines app MAU: roughly 15-25% of policyholders (J.D. Power digital studies, 2024-2025 cycle).
• US renewal email open rates: 21-24%, CTR 2-3% (Mailchimp aggregates, 2024).
• UK and EU motor and home: broadly comparable, often cited in the 15-20% MAU range for direct digital carriers, with thinner cross-market data. Price comparison culture means renewal-window engagement runs hotter, because policyholders are primed to shop.

The Insurance Information Institute (III) publishes US digital adoption trend data useful as a cross-check; retention and persistency comparatives belong to the benchmarking lesson in this module.

A regulatory note on tracking

In the EU and UK, tracking logins and email opens for marketing purposes falls under GDPRGDPREU regulation governing how organizations collect, store and use personal data, with fines tied to global revenue for breaches.View full definition → (General Data Protection Regulation) and, in the UK, UK GDPR plus PECR (Privacy and Electronic Communications Regulations). Consent for behavioural tracking and email analytics must be explicit and documented, which caps how granular a score can legally get. Separately, Apple Mail Privacy Protection (from 2021) and Gmail image proxying pre-fetch tracking pixels, inflating open rates and breaking any weight calibrated before 2021. Treat opens as directional and lean on CTR and login data.

🎬 [VIDEO: "How Insurance Companies Use Customer Data" - youtube.com - search for recent explainers from insurance industry channels covering digital engagement and CRMCRMCustomer Relationship Management: software and strategy to manage and analyse customer interactions throughout their lifecycle.View full definition → tracking practices]

Key takeaways

• Engagement after bind is proxy-based: logins, document opens and renewal-email interaction stand in for attention the product never demands.
• Composite scoring beats single metrics, but only with a rate-change feature in the model; without one, engagement can appear to cause the churn it is merely reporting.
• The sign flips with timing: pre-renewal login spikes often mean shopping, so split pre-notice and post-notice behaviour.
• Silent is not always at risk: auto-pay, escrow-billed and long-tenure policyholders renew without engaging, so segment the dormant tail before campaigning at it.
• Benchmarks are estimates: US MAU roughly 15-25%, renewal opens 21-24% with 2-3% CTR, and open rates have been unreliable everywhere since 2021.