Northvale Insurance: allocating £40m when the threat is at the counter

Frame it outside-in. The competitive loss is not 'old technology' in the abstract; it is cycle time at two moments the customer can see - buying and claiming. A four-hour settlement implies automated validation, straight-through decisioning and clean data. Northvale's 11-day cycle and two regulator findings point to the real constraint: data quality and manual handoffs inside the claims value stream, not the age of the platform.

Test each proposal against the constraint:

• App first delivers a visible channel but no new capability; a self-service portal over an 11-day process converts a slow process into a visibly slow process and risks the regulator's findings going public. Not sufficient alone.
• Mainframe first is a two-to-three year programme whose benefits are internal (run cost, maintainability). It consumes the whole envelope before any customer outcome changes, and the CFO's 24-month payback test fails.
• Neither proposal addresses data quality, which is non-negotiable and gates both options.

Recommended shape: three tranches behind decision gates - (1) ~£6m in the first 6 months on data remediation and claims automation for the two highest-volume, simplest product lines, with a measured baseline (settlement days, rework rate, complaints); (2) ~£10m on a thin digital front end once (1) shows a stable straight-through rate, so the channel exposes a fast process rather than a slow one; (3) the balance on phased core modernisation, funded partly by released run cost and re-scoped at each gate.

What would flip this: evidence that the mainframe is a constraint rather than merely old - end-of-vendor-support dates, rising severity-1 incidents, a hard capacity ceiling, or release lead times above the market's change cadence. If resilience is genuinely at risk, sequencing (2) and (3) inverts.

IT strategy answers how technology is built, run and secured: platforms, architecture standards, service levels, sourcing, technical debt. It is largely an internal supply-side document owned by the CIO.

Digital strategy answers how the organisation uses technology to change what it sells, to whom, and how work flows to deliver it - customer propositions, business model, operating model and capabilities. It is demand-side and owned collectively by the executive, with the CIO as a major contributor rather than the author.

Why the distinction bites here: the £40m decision is not a technology decision. 'Settle claims in hours, not days' is a change to the customer proposition and to the claims operating model. Treating it as an IT strategy question produces the CIO's answer - replace the platform - which optimises the supply side while the competitive loss continues on the demand side.

Governance consequences:

1. The target customer experience and the end-to-end claims value stream must be defined before technology scope; otherwise the £40m is a budget in search of an outcome.
2. Accountability sits with a business sponsor plus the executive team, not with IT alone, because the change requires authority over claims policy, data ownership and staff capacity.
3. The portfolio needs decision gates, a benefits baseline and a review cadence - the current single capital line makes trade-offs invisible and lets each function lobby separately.
4. The six-day-week symptom matters: an operating model already at capacity cannot absorb a change programme. Capacity becomes a design constraint, not an afterthought.

Where the CFO is right: a digital strategy with no financial discipline becomes an open-ended cost. Payback tests force benefits to be articulated, and for customer-facing components - a portal, an automated claim notification - 24 months is a reasonable hurdle.

Where the test misfires: it applies a uniform instrument to a heterogeneous portfolio. Three kinds of investment behave differently:

• Run/keep-the-lights-on - payback logic is irrelevant; the test should be risk-weighted cost of failure.
• Grow - incremental improvements to existing propositions; standard payback applies.
• Transform - new capability or business model, where benefits are uncertain, arrive late, and are partly optionality (the ability to respond to the next competitor move). Discounting these at 24 months systematically kills them, which is exactly how an incumbent ends up with a modern platform and no market position.

Practical resolution: segment the envelope by intent, set a different hurdle for each segment, and make the decision gates the control rather than a single up-front financial filter. For the transformative tranche, replace payback with staged evidence: does the automated straight-through rate hit the target that justifies the next release of funding? Combine that with a risk-adjusted view - the cost of the regulator escalating the data findings is a quantifiable downside the payback model ignores.

Model answer caveat: if Northvale's capital is genuinely constrained and the CFO cannot fund all three tranches, then the payback test becomes the correct tie-breaker, and the argument shifts to whether the transformation tranche should be smaller, not whether it should be exempt.