Renewals, Endorsements & Financial Integrity Testing
Every mid-term change and every renewal moves money in fractions of a year and carries a policy’s data forward. A wrong pro-rata, a dropped field, or a break that does not reconcile is real money lost and a contract quietly changed. This lesson teaches you to prove the books still balance after every change.
1 The Hook
Totara Insurance, a fictional NZ general insurer, processed a mid-term cancellation for a customer who sold their car four months into a twelve-month policy. The annual premium was $1,200. The customer was owed a refund for the eight unused months.
The pro-rata refund should have been eight-twelfths of $1,200, which is $800. The system instead calculated the refund on the remaining days but used a 360-day year for the day count while charging the original premium on a 365-day basis. The mismatch produced a refund of $811 — $11 too much. On one policy, eleven dollars is nothing. Across forty thousand mid-term cancellations and adjustments that year, the same eleven-dollar drift compounded into a six-figure leakage that nobody had budgeted for.
It surfaced at the monthly reconciliation, when the premium ledger and the bank movements stopped tying out by a slowly growing amount. The break was small enough to ignore for months and large enough, by the time anyone chased it, to require unwinding thousands of adjustments to find the rule that was wrong.
Here is the lesson hidden in that story. The team had tested that a cancellation produced a refund and that the refund looked roughly right. What no test did was assert the exact pro-rata figure against an independent calculation, and no test reconciled the sum of all adjustments back to the ledger. The dangerous defects in financial integrity are not the obvious ones — they are the small, consistent ones that only a reconciliation catches, after they have already leaked at scale.
2 The Rule
Every mid-term adjustment and renewal must produce an exact pro-rata figure and reconcile to the penny, and must carry a policy’s data forward intact. A small, consistent error is more dangerous than a large one-off, because it leaks silently across the whole book until a reconciliation finds it. Assert the exact amount against an independent calculation, and prove the sum of every movement still ties to the ledger.
3 The Analogy
A dripping tap in a Kiwi bach over summer.
A burst pipe floods the floor and you fix it that day — it is obvious, urgent, and over. A tap that drips one drop a second is different. Nobody notices on any given day. But left alone all summer it wastes more water than the burst pipe ever did, and you only find out when the water bill arrives. The small, steady loss does more damage than the dramatic one, precisely because nothing flags it.
Totara Insurance’s eleven-dollar error was the dripping tap. Each refund looked fine; the leak only showed up at the reconciliation “water bill.” A financial-integrity tester is the person who checks the exact flow at the tap — the pro-rata to the penny — and reads the meter at the end — the reconciliation — so a steady drip is caught before it drains the book.
4 Pro-Rata and Mid-Term Adjustments
A mid-term adjustment charges or refunds premium for a fraction of the policy term. Getting that fraction exactly right is the heart of this lesson.
The basic shape: a refund or additional premium is the annual premium scaled by the unused (or added) portion of the term. If a $1,200 annual policy is cancelled with eight of twelve months unused, the refund is 8/12 × $1,200 = $800. Simple — until the details bite.
- Day count must be consistent: the Totara failure. If the premium is charged on a 365-day basis, the refund must use the same basis — mixing 360 and 365 produces a small, consistent drift.
- Inclusive/exclusive boundaries: does the unused period count the cancellation day or not? An off-by-one day across a whole book is a measurable leak.
- Effective date, not keyed date: the fraction runs from the effective date of the change, as in Lesson 1 — a back-dated cancellation must refund from the back-dated date.
- Short-rate versus pro-rata: some cancellations apply a short-rate (a penalty) rather than a straight pro-rata. Test that the right method is applied for the right cancellation type.
- Fees and levies: are non-refundable fees handled correctly, and are levies adjusted on the right base? A pro-rata that refunds a non-refundable fee is a defect.
- Rounding: the final figure rounds per the stated rule, to the cent — and rounding applied at the wrong step changes the answer.
5 Data Continuity Across Renewal
Renewal re-rates a policy for a new term, but it must carry the policy’s identity and history forward intact. A renewal that silently drops or alters data is a contract quietly changed — the same class of failure as Lesson 1’s endorsement overwrite, now spanning two policy terms.
- Carried-forward fields: sum insured, excess, named insured, address, no-claims history, and any endorsements made during the term must appear correctly on the renewed policy unless the customer changed them. Test a renewal that includes a mid-term endorsement and confirm the endorsed value carries, not the original.
- History and entitlements: a no-claims discount earned over the term must follow the customer into the renewal. Dropping it overcharges a loyal customer — a financial and a conduct problem.
- Linkage: the renewed policy must stay linked to the same customer, the same claims history, and the correct product version, so continuity of cover and of record is preserved.
- No gap or overlap in cover: the new term must start exactly when the old term ends — a one-day gap is a day uninsured; a one-day overlap is a day charged twice.
- Re-rating applied correctly: the new premium uses the new rate table (Lesson 3) on the carried-forward risk — new price, same risk, intact data.
6 Reconciliation and Financial Integrity
Reconciliation is the proof that the money still balances. It is the test that catches the leaks individual test cases miss, because it looks at the whole book at once.
The core idea: the sum of every premium movement — new business, endorsements, cancellations, renewals, refunds — recorded in the premium ledger must tie to what actually moved in the bank, with every difference explained. A break is any unexplained gap between the two.
- Ledger-to-bank: total premium recorded equals total premium received, less recorded refunds, with breaks flagged not absorbed.
- Movement completeness: every adjustment that should produce a financial movement does, exactly once — no endorsement that changes the premium without a matching ledger entry.
- Aggregate pro-rata check: the sum of all pro-rata refunds and additional premiums matches the sum independently calculated for the same set of adjustments — this is what would have caught Totara at scale.
- Suspense and unallocated: money received but not matched to a policy sits in suspense and is investigated, not lost.
- Period close: at month-end the book balances, opening plus movements equals closing, and any carried break is documented.
This is where insurance financial-integrity testing meets the ledger and reconciliation testing of the Banking QA track: the discipline is the same — debits and credits must tie out, and an unexplained break is a defect, not a rounding nuisance to wave away.
7 What to Test
The practical checklist for a renewals-and-integrity tester:
- Exact pro-rata: every mid-term refund and additional premium matches an independent calculation to the cent, on the correct day-count basis and from the effective date.
- Boundary days: cancellation on the first day, the last day, and a single day before expiry — the edges where day-count and inclusive/exclusive errors surface.
- Renewal data continuity: endorsed values, no-claims history, and linkage carry forward; no cover gap or overlap at the term boundary.
- Re-rate on renewal: the new rate table is applied to the carried-forward risk and the premium is exactly correct.
- Movement-to-ledger: every premium-changing action writes exactly one matching ledger entry.
- Aggregate reconciliation: the sum of movements ties to the bank and to an independent recomputation, with breaks flagged.
- Rounding discipline: rounding is applied once, at the right step, per the stated rule, to the cent.
8 Building Integrity Test Cases
A strong integrity test case fixes the policy and the change, asserts the exact figure against an independent calculation, and ties the movement back to the ledger — not just “a refund was paid.”
Here is a worked test case written to catch the exact Totara day-count bug:
Scenario: Mid-term cancellation, pro-rata refund
Risk category: Inconsistent day-count basis (systematic leakage)
Pre-conditions: 12-month policy; annual premium $1,200; charged on a 365-day
basis; cancelled with exactly 8 of 12 months (243 days) unused.
Independent calc: Refund = 243/365 × $1,200 = $798.90 (same 365-day basis,
refund from the effective date, less any non-refundable fee).
Expected result: 1) Refund equals the independent calc EXACTLY, to the cent.
2) Day-count basis matches the basis the premium was charged on.
3) Refund runs from the effective date, not the keyed date.
Reconciliation: The refund writes exactly one ledger entry; ledger movement for
this cancellation ties to the bank refund; 0 unexplained break.
Boundary cases: Cancel on day 1 (near-full refund), day 364 (one-day refund).
Evidence required: Refund calculation with the day-count basis; ledger entry;
bank reconciliation line; the independent calculation.
Traceability: Risk R-01 (day-count basis mismatch in pro-rata).
Result: [Pass / Fail]
Notice what makes this catch the Hook bug: it computes the refund independently on the same day-count basis the premium was charged, asserts an EXACT cent match, and explicitly checks the basis is consistent — so the 360-versus-365 mismatch fails. It then ties the movement to the ledger and bank. That is the difference between a real integrity test and “a refund was paid.”
9 Common Mistakes
🚫 Accepting “the refund looks about right” instead of an exact figure
Why it happens: A refund close to the expected ballpark feels correct, and small errors look like rounding.
The fix: The Totara trap is a small, consistent error that looks plausible on every policy. Recalculate the pro-rata independently on the documented day-count basis and assert the system matches to the cent. A dollars-off drift that passes a ballpark check leaks across the whole book.
🚫 Testing pro-rata without checking the day-count basis is consistent
Why it happens: The formula looks right, so the basis underneath it goes unexamined.
The fix: If the premium is charged on 365 days and the refund computed on 360, every adjustment drifts by a small, consistent amount. Assert the refund uses the same basis the premium was charged on, and test the boundary days where the mismatch is largest.
🚫 Testing renewal on a clean policy instead of an endorsed one
Why it happens: A fresh policy is the easiest to set up and renew.
The fix: A renewal most often drops the endorsed value — carrying forward the original sum insured instead of the mid-term change — and a clean policy hides that entirely. Renew a policy that was endorsed during the term and confirm the endorsed values and no-claims history carry forward.
🚫 Treating an unexplained reconciliation break as a rounding nuisance
Why it happens: A small break is easy to wave away as noise and absorb.
The fix: A break is an unexplained gap between the ledger and the bank, and a small one that grows is exactly how systematic leakage shows up. Flag and explain every break; the aggregate of movements must tie to the bank and to an independent recomputation, with nothing absorbed.
10 Now You Try
Three graded exercises across renewals, pro-rata, and reconciliation. Write your answer, run it for AI feedback, then compare to the model answer.
Read the description of a fictional NZ insurer’s mid-term adjustment and renewal flow below. Identify 3 financial-integrity risks that could leak money, drop data, or break reconciliation, and name the area each touches (pro-rata, day-count, data continuity, reconciliation).
Mid-term refunds are calculated on a 360-day year while premiums are charged on 365 days. When a policy that was endorsed mid-term is renewed, the renewal carries forward the original sum insured from the start of the term, not the endorsed value. A no-claims discount earned during the term is not applied at renewal. An endorsement that raises the premium updates the policy but does not always write a matching premium-ledger entry, so some adjustments never appear in reconciliation.
List 3 integrity risks and the area each touches:
Show model answer
There are at least four real risks here; any three well-explained earns full marks. 1. Day-count basis mismatch — refunds on 360 days, premiums on 365. Area: day-count / pro-rata. Impact: a small, consistent drift on every adjustment that leaks across the whole book — the Totara trap. Fix: refund on the same basis the premium was charged; assert the exact cent. 2. Renewal drops the endorsed value — the renewal carries the original sum insured, not the mid-term endorsed one. Area: data continuity. Impact: the customer is renewed on the wrong cover; a contract quietly changed across the term boundary. Fix: renew an endorsed policy and confirm endorsed values carry forward. 3. No-claims discount not applied at renewal. Area: data continuity / entitlements. Impact: a loyal customer is overcharged — a financial and a conduct problem. Fix: confirm earned entitlements follow the customer into the renewal. Bonus: premium-changing endorsement with no matching ledger entry. Area: reconciliation / movement completeness. Impact: adjustments never appear in reconciliation, so the book silently fails to tie out. Fix: assert every premium-changing action writes exactly one ledger entry. The trap: each refund and renewal "works" individually — the damage is small, consistent, and only visible at reconciliation, after it has already leaked at scale.
The pro-rata test case below only checks a refund was paid. Rewrite it to assert the exact figure on a consistent day-count basis and tie it to the ledger, with these fields: Test ID, Scenario, Risk category, Pre-conditions, Independent calc, Expected result, Reconciliation, Boundary cases, Evidence required, Traceability. Use a fictional NZ vehicle-policy mid-term cancellation as the context.
“Cancel the policy mid-term. Check a refund is paid. Pass if the refund looks about right.”
Rewrite as a financial-integrity test case:
Show model answer
Test ID: FIN-PRO-015 Scenario: Vehicle-policy mid-term cancellation, pro-rata refund Risk category: Inconsistent day-count basis / movement not reconciled Pre-conditions: 12-month vehicle policy; annual premium $960 charged on a 365-day basis; cancelled with exactly 90 days unused; effective date of cancellation recorded. Independent calc: Refund = 90/365 × $960 = $236.71 (same 365-day basis, from the effective date, less any non-refundable fee). Expected result: 1) Refund equals the independent calc EXACTLY to the cent. 2) The day-count basis matches the basis the premium was charged on (365). 3) The refund runs from the effective date, not the keyed date. Reconciliation: The refund writes exactly one premium-ledger entry; the ledger movement ties to the bank refund; 0 unexplained break. Boundary cases: Cancel on day 1 (near-full refund), day 364 (one-day refund); a back-dated cancellation refunding from the back-dated effective date. Evidence required: Refund calculation showing the day-count basis; the ledger entry; the bank reconciliation line; the independent calculation it was checked against. Traceability: Risk register R-01 (day-count mismatch) and R-04 (movement not reconciled). What makes it strong: it recomputes the refund independently on the same basis and asserts the exact cent, checks the basis is consistent, runs from the effective date, and ties the movement to the ledger and bank. The original would pass on any plausible refund.
Design a test plan of 5 test cases for a fictional NZ house-and-contents renewal of a policy that was endorsed mid-term. Each case needs at least: an ID, what it verifies, an acceptance criterion, and the evidence required. Cover carried-forward endorsed values, no-claims continuity, no cover gap/overlap at the term boundary, correct re-rate on the new table, and aggregate reconciliation of the renewal movements.
Show model answer
REN-01 | Verifies: endorsed values carry forward into the renewal | Acceptance criteria: the renewed policy shows the mid-term endorsed sum insured and excess, not the start-of-term originals; before/after comparison confirms it | Evidence required: pre-renewal endorsed record vs renewed record diff REN-02 | Verifies: no-claims and earned entitlements continue at renewal | Acceptance criteria: the no-claims discount earned over the term is applied to the renewal premium; 0 lost entitlement | Evidence required: entitlement record across the term; renewal premium breakdown showing the discount REN-03 | Verifies: no cover gap or overlap at the term boundary | Acceptance criteria: the new term starts exactly when the old term ends — 0 uninsured days and 0 double-charged days | Evidence required: old term end date and new term start date; cover-period continuity check REN-04 | Verifies: re-rating uses the new rate table on the carried-forward risk | Acceptance criteria: the renewal premium equals an independent calculation using the new table on the intact risk, to the cent | Evidence required: rate-table version stamp; factor-by-factor breakdown; independent calc REN-05 | Verifies: renewal movements reconcile in aggregate | Acceptance criteria: every renewal premium movement writes exactly one ledger entry; the sum of movements ties to the bank and to an independent recomputation; breaks flagged | Evidence required: ledger entries; bank reconciliation; aggregate independent recomputation Strong plans: each case is specific, has a measurable criterion, names concrete evidence, and together they cover carried-forward endorsed values (REN-01), entitlement continuity (REN-02), no gap/overlap (REN-03), correct re-rate (REN-04), and aggregate reconciliation (REN-05). Weak plans say "check the renewal works" five times — that is the difference being marked.
11 Self-Check
Click each question to reveal the answer.
Q1: Why is a small, consistent error more dangerous than a large one-off?
Because it leaks silently across the whole book. A large one-off is obvious and gets fixed; a small per-policy drift — like Totara’s eleven dollars from a day-count mismatch — looks plausible on every policy and only surfaces at reconciliation, after it has already compounded across thousands of adjustments. The dripping tap wastes more than the burst pipe.
Q2: What is the one check that would have caught the Totara bug?
An independent recalculation of the pro-rata on the same day-count basis the premium was charged, asserted to the cent — plus an aggregate reconciliation of all adjustments back to the ledger and bank. “The refund looks about right” let the drift through; an exact independent figure and a tying reconciliation catch it.
Q3: Why test renewal on an endorsed policy rather than a clean one?
Because a renewal most often drops the endorsed value — carrying forward the original sum insured instead of the mid-term change — and a clean policy hides that defect entirely. Renewing a policy that was endorsed during the term, and confirming the endorsed values and earned no-claims history carry forward, is what exposes the continuity failure.
Q4: What is a reconciliation break, and how should it be treated?
An unexplained gap between the premium ledger and the bank movements. It must be flagged and explained, never absorbed as rounding noise — a small break that grows is exactly how systematic leakage shows up. The aggregate of all movements must tie to the bank and to an independent recomputation, with every difference accounted for.
Q5: Why does the day-count basis matter in a pro-rata calculation?
Because charging premium on one basis (say 365 days) and refunding on another (360) produces a small, consistent drift on every adjustment. The fraction of the term must be computed on the same basis the premium was charged. Test the boundary days, where the mismatch is largest, and assert the refund to the cent against an independent calculation on the correct basis.
12 Interview Prep
Real questions asked in NZ QA interviews for insurance roles. Read the model answers, then practise your own version.
“How would you test a pro-rata refund on a mid-term cancellation?”
I never accept ‘the refund looks about right.’ I recalculate it independently — the unused fraction of the term times the premium, on the same day-count basis the premium was charged, from the effective date — and assert the system matches to the cent. I explicitly check the basis is consistent, because charging on 365 and refunding on 360 produces a small, consistent drift that leaks across the whole book. I test the boundary days, day one and the day before expiry, where the mismatch is largest, and a back-dated cancellation. Then I tie the refund to exactly one ledger entry and to the bank, so the movement reconciles.
“Reconciliation is off by a small amount that keeps growing. How do you investigate?”
A small, growing break is the signature of a systematic per-transaction error, so I look for a rule applied consistently wrong rather than a one-off. My first hypotheses are a day-count basis mismatch in pro-rata, an inclusive/exclusive boundary off by a day, or a class of premium-changing adjustments that updates the policy but never writes a ledger entry. I’d take a sample of adjustments, recalculate each independently, and diff against what the system recorded to find which type drifts. Then I’d confirm every premium-changing action writes exactly one matching ledger entry, and add an aggregate reconciliation so the drip is caught early next time.
“What is the riskiest renewal to test, and why?”
A renewal of a policy that was endorsed mid-term. A clean policy renews fine and hides the real defect — the renewal carrying forward the original sum insured instead of the endorsed value, or dropping a no-claims discount earned over the term. So I deliberately renew an endorsed policy and diff the renewed record against the pre-renewal endorsed record, confirming the endorsed values and earned entitlements carry forward. I also check there is no cover gap or overlap at the term boundary — a one-day gap is a day uninsured, a one-day overlap is a day charged twice — and that the re-rate uses the new table on the intact risk.