Case Study: When Cheap Insurance Leads Cost the Most
Sector: Insurance Scope: Paid search and paid social management, conversion infrastructure, lead quality instrumentation
Client identity is confidential.
The situation
The account was hitting every marketing target and losing money.
Cost per lead was comfortably inside target and improving quarter on quarter. Lead volume was growing. The marketing function was reporting success in good faith.
The underwriting side saw something different. Bind rate on marketing-sourced leads was well below the rate on other channels, and lapse rate at first renewal on those policies was materially higher than the book average. The business was acquiring policies that did not survive to the point where insurance becomes profitable.
Nobody had connected the two views, because no system connected them.
What the audit found
No feedback loop existed between policy administration and the ad platforms. Conversions were counted at form submission. Automated bidding therefore optimised toward whoever completed forms most readily — a population that overlaps only partially with the population that binds a policy and renews it.
Intent segments were merged. Comparison-shopping traffic, compulsory-purchase traffic, and life-event traffic all ran through the same campaigns with the same bids and the same landing pages. The blended cost per lead described no actual customer.
A small set of keywords produced most of the unusable volume. A narrow group of terms accounted for a disproportionate share of leads and an even larger share of underwriting declines. They looked strong on cost per lead and were the most expensive terms in the account by any business measure.
Speed to contact was slow and inconsistent. Contact rate on leads reached quickly was materially higher than on leads reached after a delay. A substantial share of paid leads was never successfully contacted at all.
A retargeting audience had been built from a health-related enquiry page. This was a live data protection exposure that nobody had identified. It was removed immediately.
What we changed
1. Built the offline conversion loop.
Bound policies fed back to the ad platforms against the original click identifier, with premium value attached. This required cooperation between marketing and policy administration that had not previously existed, and it took several weeks to implement.
It was the highest-impact change in the engagement, and it was not a marketing change.
2. Moved bidding from lead value to expected value.
Working with the underwriting function, an expected-value figure per product and segment was supplied — reflecting expected loss ratio, not just premium. Bidding optimised to that.
3. Segmented by intent.
Compulsory-purchase, life-event, comparison, and service intent were separated into distinct campaigns with distinct bid postures, landing experiences, and expectations. Comparison traffic was retained where the product genuinely competed on price and reduced where it did not.
4. Cut and restructured the loss-making terms.
The keywords driving the highest volume of underwriting declines were paused, rebid at a level consistent with their true value, or excluded outright, depending on what the underwriting evidence supported for each. Cost per lead rose. Cost per bound policy fell.
5. Addressed speed to contact.
This was an operations change, not a media change: lead routing, alerting, and staffing were restructured around peak enquiry hours so that leads reached a person quickly and consistently rather than depending on whoever was available. It produced a larger improvement in bound policies than any change to the media plan.
6. Remediated the compliance exposure.
The health-derived retargeting audience was removed, lawful basis and retention documented across the enquiry pipeline, and audience construction rules established to prevent recurrence. Our compliance framework governs this on all managed accounts.
Results
The most visible number moved in the wrong direction, and that was expected: cost per lead rose once bidding stopped chasing form fills. What mattered was what happened beneath it. Bind rate on marketing-sourced leads improved, and the policies that did bind held up meaningfully better at first renewal than the cohort they replaced.
Cost per bound policy fell even as cost per lead climbed, because the account was no longer paying to acquire volume that underwriting would later decline or that would lapse before renewal. Speed to contact improved once routing and staffing were fixed, and that operational change moved bound-policy volume more than any adjustment to bidding or targeting.
The account had never previously calculated cost per policy surviving first renewal. Once it did, that became the number reported to the principal, because insurance profitability lives in the renewal, and an acquisition programme that is not measured to the renewal is not being measured against the business at all.
What transferred
Build the offline conversion loop before scaling spend. Without it, the platform optimises toward form fills. With it, toward policies. These are different populations and the difference compounds.
Cost per lead is not an insurance metric. It is a marketing metric that happens to be easy to collect. Cost per policy surviving first renewal is the one to report to a principal.
Speed to contact belongs in the media review. It sits with operations and it determines the return on media spend more than most media decisions do.
Further analysis of insurance acquisition is published in publications, with the wider managed media practice under business units.
To discuss an insurance acquisition programme, contact us.
