The money already spent
What an agency has already spent on its current system is gone whether it stays or moves, so it should play no part in the decision. The decision is a comparison of future costs and future benefits only, and sometimes that comparison favours staying. Two things have changed it. Configured platforms remove most of the up-front build cost. And moving an agency book, while not effortless, is a bounded task: the old system can run off the in-force book while the new one takes over at each renewal.
Sunk costs and forward costs
The pull of past spending is well documented. In Arkes and Blumer’s 1985 study, theatregoers who paid full price for a season ticket attended more plays than those given a discount. The plays were the same; only the money already paid differed. Staw (1976) found that people responsible for a failing investment commit more to it, not less. The person who chose the current system is often the person deciding whether to replace it.
The remedy is to sort every cost before comparing anything.
| Cost | If you move | If you stay |
|---|---|---|
| Licence, build and customisation already paid | Sunk: excluded | Sunk: excluded |
| Remaining contract term | Payable either way; only the overlap with a new system counts | Payable |
| Exit fees, data extraction, testing, training | Payable | Nil |
| Ongoing fees, upgrades, new product builds | New platform | Current vendor, plus increases |
| Rekeying, corrections, bordereau effort | Expected to fall | Continues |
| Execution risk | Real; price it | Not zero for an ageing system |
Two points are often missed. A remaining contract is not a reason to stay; timing the move to its end reduces the overlap. And staying is not free: in a 2025 West Monroe survey of 300 US insurance executives, 54% said more than half their IT budget goes on maintaining existing systems.
Staying is the right choice when the current system does the job at an acceptable cost, when exit costs would not be recovered in a reasonable period, or when the new platform has not yet built what the agency needs. That last test applies to Cuttleflow as much as anyone.
Configured, not built
The largest forward cost of switching has been the build. McKinsey (2025) puts custom core builds for insurers at five to ten years, against three to five for configured off-the-shelf platforms. Those are carrier figures, not agency ones, but the direction holds: the less that is built for one customer, the less there is to pay for, test and maintain. Perspective 14 explains why the cost of a large estate grows faster than the estate itself.
Cuttleflow is designed on that basis. Products are configured, not built per client. Cuttleflow standard products are built and maintained at Cuttleflow’s cost, and fees are a percentage of premium. Cuttleflow Go, in development, will let an agency set up a financial lines product on the Cuttleflow standard wording with its own rates and rules. If the agency does not pay for a build, what remains of the switching cost is data, people and time.
Migration: not effortless, but bounded
Data migration has a poor record. Bloor Research’s 2007 survey of large enterprise projects found most overran or were abandoned, usually because scope was not understood and data quality was worse than expected. An agency book is a smaller problem: one source system, a defined set of products, and records in the thousands rather than millions. The same risks exist; they are easier to control.
Cuttleflow’s Book Importer specification sets out how a book moves. It works from file exports (CSV or Excel) of the kind most systems already produce, and offers two modes:
- Renewal-forward (the default). Clients and in-force policies are loaded for reference. The old system keeps reporting its run-off; each policy moves to Cuttleflow at its first renewal. Nothing is converted in bulk, so this is the cheapest and lowest-risk route.
- Full cutover. The whole book reports through Cuttleflow from an agreed month, with the managing agent’s written agreement. Totals must reconcile to the old system to the cent.
Three principles keep it contained:
- It records history; it does not re-decide it. Premium, commission and tax move unchanged. Retroactive dates and original inception dates are kept, so continuity of cover is not lost.
- Every row is checked and nothing is silently corrected. Each row is marked ready, query or reject. A query goes to a person.
- People sign off, and it can be undone. A batch needs two named sign-offs before it is committed, and can be reversed as a whole.
What remains hard is honest work: poor data in the old system, staff time in busy weeks, and reconciling outstanding premium at cut-over. Migration is the most variable part of any implementation, so it should be scoped from a sample export before anyone commits to a date.
Weighing the long-term benefit
The benefit of a better system accumulates, so it should be weighed over three to five years, not against the first month of disruption. The gains that matter are less rekeying and fewer corrections, faster bordereaux and reconciliation, quicker product launches, data an agency can answer questions from on the day they are asked, and a book that is easier for a capacity provider to oversee and an acquirer to value.
Worked example (all figures illustrative). An agency writes $15m of premium. It has spent $600,000 on its current system; that is sunk and excluded. The system costs $120,000 a year with 18 months left to run. Staff spend 25 hours a week on rekeying and bordereaux, about $91,000 a year. A new product would cost $80,000 to build on the current system.
| Five-year cost | Stay | Move at contract end |
|---|---|---|
| Current system | $600,000 | $180,000 (18 months) |
| New product build | $80,000 | Nil (configured) |
| Switching (overlap, staff time, risk allowance) | Nil | $113,600 |
| Operational effort (halved after the move) | $455,000 | $295,750 |
| New platform fee | Nil | F |
| Total | $1,135,000 | $589,350 + F |
Moving is cheaper if the new platform’s fee over 42 months is under about $545,650. If the saving is only a quarter, that threshold falls to about $466,000; if switching costs double, to about $432,000. An agency can compare those figures with a real quote. The $600,000 already spent appears nowhere.
To apply this: set sunk costs aside in writing; time the move to the contract end; measure one month of current effort before estimating any saving; test the result with the benefit halved and the switching cost doubled; and test the mapping on a real export before committing.
A note on status
The Book Importer described above is specified for Australian professional indemnity books first; other classes follow. Cuttleflow Go is in development. The figures in the worked example are illustrative, not Cuttleflow prices. This paper describes a method for comparing costs; it is not financial advice.
Sources
- Arkes, H. R. & Blumer, C. (1985). The psychology of sunk cost. Organizational Behavior and Human Decision Processes, 35, 124–140. Study details via Navarro & Fantino.
- Staw, B. M. (1976). Knee-deep in the big muddy. Organizational Behavior and Human Performance, 16, 27–44.
- McKinsey & Company (2025): How P&C insurers can successfully modernize core systems
- Insurance Business (20 October 2025): West Monroe survey of insurance executives
- Bloor Research (2007). Data migration survey, as reported in trade press.
Cuttleflow Systems · Perspective 25 · 33°53′S · 151°16′E · Sydney