A view from Owen Greenwood, Client Partner – Insurance
Policy administration system (PAS) transformations have a habit of starting with optimism. There is a target platform, a programme plan and a date when the old estate is meant to disappear.
Then the data gets involved, and that is often where the programme becomes something very different.
My view is simple: PAS transformations rarely become difficult because organisations don’t know how to move data. They become difficult because the complexity and history contained within that data has been underestimated.
The challenge is therefore not simply transferring information from one database to another. It is preserving the meaning and context of that information as it moves.
That is why PAS migration is not just a technology project. It is an insurance data transformation problem.
Your PAS has been carrying more than data
Most insurers already know their data is complicated. What is often underestimated is just how much complexity becomes visible when you try to move it.
A PAS may have supported a book of business for years. During that time, policies have changed, products have evolved and the way information is captured may have changed with them. Much of that complexity can be absorbed by the existing environment because the systems, processes and people around it have learnt how to interpret it.
Migration removes that familiarity. The organisation suddenly has to be explicit about what the data means, how different elements relate to one another and how those relationships should be represented in a different target model.
That is where apparently simple questions become difficult. Which version of a policy matters? How should its history be represented? How do endorsements, cancellations or renewals relate back to the policy? What happens when information exists in a source system but does not fit neatly into the future-state structure?
The problem is therefore not only data quality in the traditional sense of whether a field is complete or correctly formatted. It is whether the organisation understands the data well enough to move it without losing its context.
You can technically move a piece of data and still get the migration wrong.
The more complex the product, risk or policy history, the more important that understanding becomes. This is why, when an insurer tells us its data is complicated, our response is usually that it is probably more complicated than you think.
That isn’t a criticism of the organisation or its data teams. It is simply what experience of insurance migration tells us. Complexity can be absorbed by an established estate for years. Migration is often the point at which that complexity finally becomes visible.
The later it is discovered, the harder it becomes to control.
Your first migration question shouldn’t be ‘how?’
It should be “what?”
What actually needs to move?
That sounds obvious, but it can fundamentally change the shape of a PAS programme. An insurer with multiple legacy systems does not automatically need to migrate every policy, every piece of history and every dataset into the new environment.
Some books may be suitable to run down. Some data may need attention before it is fit to move. Some migrations may be relatively straightforward, while others may involve source data where the quality and complexity mean the organisation first needs to decide whether migration is the right answer at all.
When several systems are involved, another question quickly follows: what should move first?
These are not simply technical decisions for the migration team. They shape scope, sequencing, effort and ultimately how much complexity is carried into the future state.
Getting these decisions right early can fundamentally change the shape of a PAS transformation.
That is why we believe migration strategy needs to happen before migration execution.
Instead of asking, How do we move everything we have?, insurers should be asking, What does our future estate genuinely need, and what is the most sensible way of getting there?
That means establishing clarity around what should migrate, what may remain where it is for a period, where quality needs to be addressed and how migrations should be sequenced across the wider programme.
There is a significant difference between making those decisions deliberately at the beginning and discovering halfway through delivery that some of the data being moved should never have been in scope.
One PAS is rarely the whole story
The strategic question becomes even more important where an insurer is managing several policy administration systems.
That may be the result of different business lines, acquisitions or simply years of change. Whatever the reason, the programme objective is often not simply to replace one PAS. It is to move from a fragmented estate towards a more consolidated future state.
That introduces another risk: treating every migration as though it is an entirely separate problem.
Every source will have differences. There will be different structures, data quality issues, products and nuances. But when several systems are being migrated towards the same target environment, not everything starts again from zero.
The target has not changed. The team has already learned how the target model works. It has already discovered where difficult mappings tend to appear and developed a deeper understanding of the organisation’s own data and its particular complexities.
That knowledge has value.
After the first migration, the question should therefore not simply be How do we deliver the next one? It should be What have we already solved that we shouldn’t have to solve again?
A programme of PAS migrations should become more informed and more repeatable as it progresses, rather than repeatedly returning to the starting line.
The first migration should make the second better. The second should make the third better again.
That is how PAS transformation starts becoming a capability rather than a collection of disconnected projects.
The target system does not remove the legacy problem
Transformation programmes naturally spend a lot of time focused on the destination. What platform are we moving to? What will the future architecture look like? What new capabilities will it give the business?
Those questions matter, but a new target platform cannot make difficult source data simple.
The target still needs to receive data that has been understood, mapped and transformed correctly. If there is ambiguity in the existing estate, simply moving that ambiguity into a newer platform does not resolve it.
The critical work therefore happens in the space between source and target. The source model and quality need to be understood, as does what the target expects. Then, at a detailed level, the team needs to determine how each relevant element of data moves between the two and what transformation is required along the way.
Sometimes that transformation can appear straightforward — a date may simply need to be represented in a different format. But across an insurance book containing multiple policy events, relationships and different forms of risk information, those individual transformations quickly become numerous and interconnected.
This is where insurance expertise matters.
PAS migration is not simply a mapping exercise. It requires people who understand what the data represents in an insurance context, not just where it sits technically. Knowing that two fields can be mapped is very different from knowing whether they should be mapped, and what meaning needs to be preserved when they are.
A good cutover SHOULD be boring
There is another belief we hold quite strongly: a good PAS cutover should be boring.
The final migration should not feel like the first time the organisation is discovering whether its approach works. Long before production cutover, the source and target should have been understood, the mappings built, the required transformations applied and the inevitable quality issues surfaced.
Crucially, the migration itself should also have been run repeatedly through testing.
The important point is not simply how many times it has been executed. Every run creates another opportunity to expose an issue while there is still time to understand and resolve it.
By cutover, moving the data should therefore feel familiar. Not because insurance data has suddenly become simple, but because its complexity has already been encountered, understood and dealt with.
That puts an insurer in a very different position from relying on the final migration event to prove that the programme has got the data right.
PAS transformation is bigger than migration
The technology matters and the migration engineering matters, but neither is enough on its own.
PAS transformation sits across the data, the processes surrounding that data, the people involved and the decisions the organisation makes about what its future estate should look like.
In our experience, programmes can begin to struggle not because moving data is conceptually difficult, but because of everything the migration uncovers along the way.
The answer is not to pretend that complexity can be removed. It is to confront it earlier and manage it deliberately.
Understand the estate before assuming everything needs to move. Understand what the policy data actually represents. Make deliberate choices about migration, run-off and sequencing. And when there are several migrations ahead, build on what has already been learned rather than treating each one as a new beginning.
Policy administration systems do not have to be where transformation programmes go to die.
But avoiding that outcome starts by recognising what PAS migration really is: not simply a data movement exercise, but the transformation of a complex insurance data estate.
How Dufrain can help
Whether you’re still deciding which PAS is right for your business, working through what should migrate and what shouldn’t, or looking to reduce the risk and effort of moving multiple legacy systems, we can help.
From early-stage PAS strategy and platform selection through to data migration and legacy retirement, Dufrain helps insurers make better decisions, reduce delivery risk and create a more repeatable route to PAS transformation.
Where multiple migrations are moving into the same target platform, that repeatable approach can create significant efficiencies — in the right circumstances, we believe around 70% efficiency across subsequent migrations is achievable.
Talk to us about your PAS transformation plans, or ask us how a repeatable migration approach could help reduce risk, accelerate legacy retirement and improve delivery efficiency.
View some of our client stories here.