Policy data conversion that re-rates every in-force policy and proves it matches before go-live
Every converted policy proved to match on premium and coverage before go-live.
See one case, screen by screen ↓Moving in-force policies off a legacy policy administration system onto a new one, with their coverages, forms, endorsements and history, and proving each one is the same policy afterwards. Policy Data Conversion has six agents read the legacy records and the scanned forms behind them, convert every policy, re-rate it in the new system and compare it, and a wave is signed only when count, premium and coverage tie out.
How to choose AI solutions for insurance →
Updated
A new policy system is only live when the in-force book is on it. Every policy has to come out of the old system with its remarks, scanned forms and endorsements, be mapped to the new products and forms, and come out the other side as the same policy: same coverage, same premium. Conversion teams map fields in spreadsheets, load a test batch and reconcile premium at the batch level.
Batch totals hide the problem. Rating facts sit in free-text remarks and on scanned certificates, not in fields, so a field-to-field map drops them without an error. Rises and falls net out, and a dropped credit on 128 policies disappears inside a $31 million wave. The customer finds it at renewal, as an unexplained premium change — or as a jewelry schedule that never loaded and a claim that is not covered.
Estimated split for one cycle on one wave, by hand and with the solution, in working days.
Six specialist agents read, convert, re-rate and compare every policy; the conversion lead approves each fix and the program manager signs the wave.
Larkhaven Mutual is moving 214,860 in-force policies off HOMER, its in-house mainframe in service since 1994, onto Larkhaven One, one wave of renewals at a time. Wave 3 is 18,412 Homeowners policies in Iowa and Nebraska renewing in December. Offers mail Fri 30 Oct; the wave freezes Sun 25 Oct. Here is Monday 12 October, screen by screen, in the working solution.
Test load 3 ran Sunday at 22:40. The wave ribbon splits all 18,412 policies: 17,596 match to the cent, 309 within the $1.00 rounding tolerance, 472 differ, 35 can’t convert — 97.25% parity, up from 94.9% at test load 2. The first difference is Dale & Marisol Brennan, HO 418829-04: the HOMER green screen on the left, the Larkhaven One record on the right, aligned field by field.
“+$14.00 a year, the customer would see it on the renewal offer.”
Nadia presses “Re-rate in Larkhaven One”. The current term is rated again at IA-HO-2025-07, the same rate version HOMER used, and all 11 rating steps are compared. Ten are equal. Step 6, the protective device credit, is −$14.00 in HOMER and not applied in Larkhaven One. The run log says why: converted with mapping version 3.7, one value with no target — PROT DEV = Y.
Same rate version or no comparison: a rate change is never mistaken for a conversion error.
HOMER’s PROT DEV field says only Y. The alarm type sits in the remark CS ALM SOUTHGATE MON #77412 CERT REC 03/19, and the scanned certificate of alarm installation from Southgate Monitoring Co., installed 14 March 2019, shows central-station burglary and fire monitoring. Under rule 406 of the rule manual, that earns a 1% credit on the $1,400.00 base premium.
“Protective device credit dropped. HOMER’s PROT DEV says only Y; the alarm type sits in the remark field, and the map had no target for a bare Y.”
The same cause runs through 128 wave 3 policies, worth +$2,108 a year. The proposed rule shows its evidence: 128 of 128 remarks name a central station, and 119 alarm certificates and 9 monitoring invoices read all show fire and burglary monitoring. Approving it moves the map from version 3.7 to 3.8; any policy that still differs stays on the list.
“When PROT DEV = Y with no type, read the remark and the alarm certificate and set the device type.”
All 128 policies are converted again and re-rated at HOMER’s rate versions. The Brennan home now reads $1,412.00 on both sides, protective device “Central station · fire + burglary”, and wave 3’s differences fall from 472 to 344. Parity moves to 97.94%.
“128 policies fixed and re-rated · 344 differences left in wave 3.”
344 policies still differ, net +$9,099 a year, across five causes — each with a count, an average, an example policy and a drafted rule. Roof age taken from the year built: 61 policies, about +$195 each; Ruth and Arne Halvorsen’s roof was replaced in 2021, but the age came from the 1962 build year, so $1,644.00 became $1,826.40. A 1.5% wind and hail deductible rounded to 2%. ZIP 50023 split into territories 14 and 14A, moving 96 Ankeny homes.
A rule fixes the conversion, not the rate, and a person approves every rule.
Teresa Villanueva’s HO 04 61 lists four jewelry items worth $18,400. The extract carried the schedule’s $202.40 premium but not its items, so Larkhaven One has no items — the coverage is missing. The same is true on 39 policies, $333,900 of scheduled property in all. The rule reads each scanned HO 04 61 and loads every item; a person checks any item over $10,000.
“Coverage differs — cannot be accepted; approve the rule.”
Mapping version 3.8 covers 9 products, 64 forms, 212 fields and 1,318 code values. Values at 95% confidence or more are approved on their own; the rest wait for a named person. ENDT SPP to Scheduled personal property (HO 04 61) waits for Tomás; the roof age class, the wind and hail deductible, the split territory and the loss-free years wait for Nadia. Every decision is versioned.
21 policies have no construction type, 9 have a ZIP that does not match the county, and on 5 the named insured has died with the estate pending. They stay on HOMER until resolved, each with its agency and an action: request from the agent, correct the address, or convert by hand at renewal.
The Wave reconciler ties wave 3 measure by measure. Coverage A ($5,237,418,000), forms, endorsements and mortgagees tie. Written premium is +$9,140 against a $3,123 tolerance, and scheduled property is $333,900 short. Six blockers are listed, each with its owner, and the signatures go in order: Nadia Kerrigan, then Lorraine Abbott, then Grant Whitfield.
“Wave 3 signs when every policy ties out, or each open item is accepted with a reason. Coverage differences cannot be accepted.”
The same wave, seen by the four people who answer for it — what their test loads looked like, and what they look like now.
Six specialist agents read, convert, re-rate and compare every policy; the conversion lead approves each fix and the program manager signs the wave.
Reads the legacy extracts — policy, coverage, rating, remarks and claims — profiles every field and links each policy’s scanned documents.
Reads scanned declarations, endorsements, alarm certificates, inspection reports and appraisals, and cites the page for every value.
Maps legacy products, forms, fields and code values to the new system, with a confidence and sample policies.
Converts each policy, re-rates its current term in the new system at the legacy rate version, and compares field by field and rating step by step.
Groups differences by cause, traces each cause to a legacy field, remark or scanned form, and drafts one rule per cause.
Ties counts, written premium, every coverage limit, forms, endorsements and mortgagees per wave, lists each blocker with its owner and drafts the sign-off report.
Approves every rule. The agents propose; a named person decides.
The conversion team can ask about any policy, difference, rule or wave in plain words — or say what to decide. Every answer cites the record or page it rests on.
Why is the Brennan renewal $14 higher?
Because the protective device credit was dropped. HOMER says only PROT DEV = Y; the alarm type is in the remark CS ALM SOUTHGATE MON #77412. The scanned certificate shows central-station fire and burglary monitoring, which earns 1% of the $1,400.00 base premium — $14.00. 128 wave 3 policies share this cause.
Which differences would a customer notice at renewal?
359 policies would see a higher premium: 128 lost the alarm credit, 61 have a roof aged from the build year (about +$195 each), 96 Ankeny homes moved territory and 74 lost loss-free years. Worse, 39 would lose scheduled jewelry and art coverage — $333,900 in all.
Can wave 3 sign off before the freeze on 25 Oct?
Yes, if the six rules are decided this week. Offers mail Fri 30 Oct and the wave freezes Sun 25 Oct. Today 472 policies still differ and scheduled property is short by $333,900, which cannot be accepted; the 35 that can’t convert need Lorraine Abbott’s acceptance.
Show me the policies that lose coverage
39 policies, all with a scanned HO 04 61 whose items never reached the extract — Teresa Villanueva’s $18,400 jewelry schedule is the first. The Document reader has read all 39 schedules; approving the rule loads 212 items.
13 screens from the working solution, on its sample data. Pick one to see it large.
What Larkhaven Mutual decided for this conversion. Every change is a new version on the record.
| Setting | Default | Choose from |
|---|---|---|
| Premium tolerance per policy | $1.00 | Any amount counted as rounding |
| Wave tolerance at sign-off | 0.01% of premium | Any share of written premium |
| Coverage tolerance | None | Any coverage difference blocks sign-off |
| Rate version must equal the legacy system’s | On | On · Off |
| Auto-approve mappings at | ≥ 95% confidence | Below it, the forms specialist decides |
| Scheduled items needing a person | Above $10,000 | Any item value |
| Renewal offers mail | 32 days before the effective date | Any number of days |
| Wave freeze | 5 days before offers mail | No rule changes after it |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample wave of policies · “target” = the design goal, measured in the live solution · “estimated” = our estimate · “typical” = published figures (Roland Berger, How insurers can maximize value from core insurance system transformation — about half of core-system replacements miss time or budget, with underestimated data migration among the causes). People, companies, policies and addresses named on this page — Larkhaven Mutual, HOMER, Larkhaven One, Southgate Monitoring and every insured — are fictional: characters and sample data in the working solution.
Moving in-force policies off a legacy policy administration system onto a new one, with their coverages, forms, endorsements and history, and proving each one is the same policy afterwards. Policy Data Conversion has six agents read the legacy records and the scanned forms behind them, convert every policy, re-rate it in the new system and compare it, and a wave is signed only when count, premium and coverage tie out.
The Parity checker re-rates each policy’s current term in the new system at the same rate version the legacy system used, then compares every field and every rating step, not just the total premium. A difference points to one rating step — in the demo, step 6, the protective device credit — and from there to the field, remark or document behind it.
Yes. The Document reader reads scanned declarations, endorsements, alarm certificates, inspection reports and appraisals, and cites the page for every value; the Legacy record reader keeps remark lines whole. Pages it cannot read go to a person — in the demo wave, 12 of 41,206 scanned documents.
No. The agents are read-only on the legacy system, and they never change a premium, a rate or a coverage to force a match. A fix is a rule that changes how the policy is converted, and a person approves every one.
They stay on the legacy system until resolved, listed by reason — such as a missing construction type, a ZIP that does not match the county, or a deceased named insured with the estate pending — with an owner and an action: request the data from the agent, correct it, or convert by hand at renewal.
Yes. The solution tracks each wave on the renewal calendar, from read to converted, re-rated, matching and signed, with parity across test loads. The renewal-offer lead time and the wave freeze are settings, and changes in the legacy system are re-converted and re-rated every night until the wave goes live.
Yes. The conversion lead approves every rule, the forms specialist decides every mapping below the confidence threshold, and the wave is signed in order by the conversion lead, the policy services lead and the program manager. Coverage differences cannot be accepted, and other open items only with a written reason.
The Agentic Solution Engine builds and deploys it from your requirements — your legacy extracts and record layouts, a sample of scanned forms, your product and form map and your rule manual — and it goes live once every quality gate has passed. We will walk you through it on a sample of your own policies first.
Changes keyed and checked once, so keying errors don't surface at claim time.
1-min demoCertificate of Insurance IssuanceNo certificate promises more coverage than the policy gives.
1-min demoCancellation & Nonrenewal ComplianceEvery notice checked against state rules and moratoriums before it is sent.
1-min demoWe’ll run Policy Data Conversion on a sample of your own legacy policies and scanned forms.