How to measure an insurance call test

A practical scorecard for insurance agents and agencies: measure each stage before deciding whether to repeat, change or stop a call test.

Decide what the test needs to answer

An insurance call test should answer a specific buying question: can your team turn this source, in these states and hours, into customers at an acquisition cost your business can support? A billable call, an application and a retained customer are different outcomes. Counting them separately makes a small test more useful.

Before starting, write down the insurance product, source, agreed call model, participating agents, approved states, answering hours and maximum spend. Use the applicable dated offer for rates and billability. A 90-second threshold is a billing condition; it does not prove consumer eligibility, a sale or profitability.

Keep a simple call-level scorecard

Download the free call-test scorecard (CSV). It opens in common spreadsheet tools. Keep your completed copy private. Use a non-identifying reference to reconcile each row with your authorized call records; do not put consumer names, phone numbers, health information or recording links in a shared worksheet.

Insurance call-test stages and evidence
StageWhat to recordWhat it tells you
DeliveredA call offered to your configured destination in the authoritative logDelivery volume; it is not proof of a live answer
AnsweredA verified live agent answer, separately from voicemail or a failed connectionWhether the team could serve the opportunity
BillableThe agreement-specific billing decision and any adjustmentWhat you owe under that agreement
ApplicationA submitted application in your authorized recordsAn intermediate sales outcome
Issued and placedEach stage separately, using your carrier or agency definitionsProgress beyond application, subject to later changes
Collected and retainedCommission actually received, with reversals and chargebacks reconciled laterThe observed financial outcome for that cohort

Calculate the rates with consistent denominators

Net call spend means confirmed billable call charges less confirmed credits or adjustments. It is not the number of delivered calls multiplied by an assumed rate. Keep prepaid funds that have not been consumed separate from the cost of calls in the cohort.

  • Billable rate per delivered call: confirmed billable calls divided by delivered calls. Compare delivered, live-answered and billable counts side by side. Flag calls billed without a verified live answer for agreement-specific review; an anomaly alone does not establish a credit owed.
  • Live-answer rate: verified live answers divided by delivered calls.
  • Application rate per live answer: submitted applications divided by verified live answers.
  • Cost per placed policy: net call spend divided by placed policies. Also show a fully loaded figure including attributable labor and other acquisition costs.
  • Observed contribution: collected commission, less reversals, net call spend and attributable variable costs. This excludes fixed overhead unless you explicitly include it.

If a denominator is zero, mark the rate as not available. If a cost or outcome is missing, mark it unknown instead of entering zero. Keep the original call cohort together: do not divide this week's applications by a different week's calls.

The invented examples below are not RingFunnel results, prices or recommended spending. Keep pending outcomes, observed zeroes and missing data distinct, and use your team's named cohort-review checkpoints.

Fictional examples: pending, zero and missing data
ExampleConfirmed spendPlaced-policy recordCost per placed policyNext review
A: pending outcome$120 (arbitrary)Application pending; placed count not finalizedUnknown and provisional, not $0Review at the team's named checkpoint
B: observed zero$120 (arbitrary)0 placed observed at the team's stated checkpointNot available: denominator is zero; this does not mean zero costRetain the observed zero and review later under the team's cohort policy
C: missing costUnknown1 verified placed policyUnknown: numerator is missingReconcile the cohort's cost records

For a purely illustrative example, a test with 30 delivered calls and 24 verified live answers has an 80% live-answer rate. These are hypothetical counts, not RingFunnel results or a suggested test size. Answering rate alone says nothing about profitability.

Review operations before judging the source

If live answers are low, review destination settings, staffed hours, simultaneous-call capacity, voicemail and pause confirmations. Separate a call that never reached the intended destination from one that reached a live agent but did not produce an application. Keep supporting evidence and submit any review within the agreement's window.

If live answers are healthy but applications are low, review fit, agent handling and the original source or handoff. Do not infer a cause from one call. Where possible, keep states, hours and agent assignment comparable when testing a change. A small uncontrolled comparison can suggest the next question; it rarely proves one vendor is better.

Compare cohorts at the same age

Keep a cohort of calls together from delivery through later policy and commission updates. Record when each call arrived, the sales-stage dates, the financial observation date, and the product or source version being tested. Then compare results at similar ages, such as each cohort's agreed operational checkpoint and later financial checkpoint. Choose those checkpoints for your own product and carrier reporting cycle rather than adopting a universal waiting period.

A call with an application still pending is not a failed policy, and it is not retained revenue. Report it as pending. If one cohort has had longer to issue, pay and experience reversals, the raw totals are not a fair comparison. If follow-up coverage is incomplete or costs have not been reconciled, leave the result provisional.

Retained must have an explicit meaning: for example, still in force at a named observation date in the records you use. That does not mean the policy can never lapse later. Reconcile subsequent changes back to the original cohort instead of making a new cohort appear worse or better.

Worked example: applications are not net commission

Every number below is hypothetical. This example illustrates the method. It is not a RingFunnel result, price quote, promised conversion rate or recommended test size.

A hypothetical call cohort at one financial checkpoint
RecordIllustrative value
Delivered calls / verified live answers40 / 36
Confirmed billable calls30
Applications / initially placed policies10 / 8
Policies still in force at the stated checkpoint6
Commission received / recorded reversals$3,000 / $600
Net commission received$2,400
Net call spend after confirmed adjustments$1,800
Attributable variable handling costs$300

The observed contribution is $3,000 − $600 − $1,800 − $300 = $300, before fixed overhead, taxes and any later changes. Record a reversal only once. Do not subtract it again if the commission figure you started with was already net of reversals.

Call-only cost per application is $1,800 ÷ 10 = $180. Call-only cost per initially placed policy is $1,800 ÷ 8 = $225. Including the variable handling cost, acquisition cost per policy still in force at this checkpoint is ($1,800 + $300) ÷ 6 = $350. These measures answer different questions; substituting applications for retained policies would change the result.

The $2,400 net commission is observed cash after the recorded reversals. It is not automatically final profit or a forecast of lifetime commission. Any unearned advance, future reversal exposure, unpaid cost or other unresolved item should be identified separately under your own accounting and carrier terms.

Calculate a break-even threshold before the next test

Using the same hypothetical cohort, suppose the planning assumption is $400 of net commission per retained policy: $2,400 ÷ 6. If call spend and variable handling costs total $2,100, the simplified break-even count is $2,100 ÷ $400 = 5.25, so at least six comparable retained policies would be needed to cover those included costs.

Holding that assumed six-policy outcome and $300 handling cost fixed, the call-spend ceiling would be (6 × $400) − $300 = $2,100. Across 30 billable calls, that is $70 per billable call. This is a sensitivity calculation, not a vendor price, safe bid or recommended budget. Fewer retained policies, lower net commission, additional servicing costs or later reversals reduce the ceiling.

Use a conservative, evidence-based net-commission assumption, state which costs are included, and decide how much uncertainty your cash position can support. If net commission per policy is unknown or non-positive, this model does not provide a usable spending ceiling. For the next order, connect the review to our guide on when to reorder inbound insurance calls.

Turn the results into an agency call plan

If you own an insurance agency or lead an IMO, use the agency and IMO planning brief to connect the test results to your team's answering capacity.

Record the insurance products and each agent's licensed states, staffed hours and time zone, agents available to answer, intended call volume, and the person who owns the buying decision. Bring that brief and the results you want to evaluate when you book an agency call-program review.

Confirm fit, current availability and the applicable terms before starting or expanding the program.

Set review and stop rules before spending

Choose an operational checkpoint and a later financial checkpoint. At the first, confirm delivery and answering work and reconcile billing. At the later checkpoint, allow time for policy status and commission records to mature. The right timing depends on your product and records; there is no universal number of days that makes the result final.

Write down when to pause for a routing failure, when to ask for a billing review, and what evidence would justify another capped test. Do not keep increasing spend simply to make an unfavorable sample look better. A positive early result still needs confirmation as more outcomes mature.

Make the next decision specific

End the review with one decision, an owner and a date: repair an answering issue, clarify a disputed definition, test one source change, repeat under comparable conditions, or stop. Your scorecard supports a review; it does not rewrite your agreement.

For preparation, use our call-vendor evaluation guide, compare inbound calls and live transfers, and review pricing and billability questions. When you are ready to discuss program fit, book a conversation with RingFunnel. Availability and commercial terms require confirmation.