Table of Contents
Key takeaways
- Saturn Group grew from £100M to £600M since 2023 under private-equity ownership - growth ambitions, not internal readiness, set the pace.
- Before the transformation: 20-40 separate Excel files, some over 100 sheets, with a single solvency model or P&L taking two to three weeks to produce.
- A build that would traditionally take years was compressed into three months using a GenAI-accelerated delivery approach.
- Getting the data design right upfront is the single factor Saturn's finance director credits most.
- Phase two extends into scenario and stress testing - roughly 6,000 actuarial assumptions sitting behind roughly 300-400 planning drivers.
Growth that doesn't wait for finance to be ready
In a fireside chat at Pigment Catalyst London, Deloitte partner Yolaine Kermarrec spoke with Steve Smith, Commercial Finance Director at Saturn Group, about building an FP&A function from scratch under real deadline pressure.
Saturn Group owns two insurers and an MGA that services them. Since 2023, backed by private equity with explicit growth ambitions, the business has grown from roughly £100 million to £600 million. Smith joined in June to build FP&A essentially from zero:
- ~90 staff in the insurer when he arrived, plus a thousand more staff moving across from the MGA side
- No FP&A function on the MGA side at all
- "Everything was Excel all the way through" - 20 to 40 separate files, some running 100+ sheets
- A single solvency model or P&L took two to three weeks to produce
In an insurance market where pricing moves in real time daily, that cycle time is a structural risk to how fast the business can respond to its own market.
Why speed of delivery mattered as much as speed of decisions
Smith had evaluated EPM platforms before at a sister company under the same PE ownership, and had been impressed by Pigment even then. This time, Deloitte brought a delivery approach built for acceleration:
- A GenAI-driven method using Deloitte's library of EPM accelerators across the full implementation lifecycle
- Pigment's modeler agent amplifying the build phase itself
The result: a build Smith estimates would traditionally take the better part of a year happened in roughly three months. The team is now three-plus months in, nearly live, running underwriting, P&L, and cash flow modules - sourcing everything from Pigment instead of reconciling 20 separate data sets.
In a highly acquisitive, PE-backed business, every month rebuilding financial infrastructure on old tools is a month finance can't yet support the deal activity the growth strategy demands.
The one thing Smith says mattered most
Get the data design right in the design docs, before build starts. Saturn didn't get everything right on the first pass - Smith is candid it was roughly 90%, not 100% - but getting the foundation mostly right up front is why a multi-year build compressed into a single quarter without hidden rework surfacing later.
His advice: "Get your data sets up front in your design docs. If you get that wrong, you're going to have loads of hurdles later on to try and solve."
What Saturn is actually optimizing for
The target isn't "replace Excel with something faster." It's:
- A full set of financial statements, including solvency capital requirements, produced within five to six working days of month-end
- Fast enough that leadership can make real-time decisions in a market where pricing itself moves daily
There's a talent thesis underneath the tech story too. Smith is direct that he doesn't want his team spending time as "number crunchers" - he wants finance to be "a big driver" for the business, and is already recruiting toward that, including bringing in data science graduates who'd applied elsewhere in the organization.
Phase two: where the real complexity lives
Phase one - underwriting, P&L, cash flow, one sourced data model - is nearly complete. Phase two is harder and more insurance-specific:
- Insurers run stress tests roughly every six months; Saturn currently manages ~20 distinct stress scenarios (market shifts, economic shocks, competitor failure, with particular focus on inflation).
- Saturn's planning model runs on roughly 300-400 assumption drivers - but the actuarial layer behind those drivers involves roughly 6,000 individual assumptions.
- The ambition: bring the actuarial team, who already run millions of scenario calculations, into building that complexity directly inside the platform - not feeding the model from external assumption sets.
That's a bigger ambition than most planning transformations attempt: not just planning the business, but absorbing actuarial-grade scenario modeling into the same governed environment finance already runs on.
Why this matters beyond insurance
Saturn's situation is specific - insurance, PE ownership, a three-month build - but the tension is universal: the business is scaling faster than the finance infrastructure supporting it, with no comfortable multi-year runway to fix that gap.
The takeaway isn't "move fast and fix data quality later." It's closer to the opposite: teams that move fast under real deadline pressure are the ones that get the unglamorous data design work right early. Speed and rigor aren't in tension - under enough pressure, rigor is what makes speed possible at all.
