The CFO who almost killed the data team (and what saved it)

The meeting lasted eleven minutes. That is, reportedly, how long it took Marks & Spencer's then-finance leadership to nearly axe a significant portion of its analytics budget in a 2019 planning review, after the data team presented a slide deck full of capability metrics, model accuracy scores, and data qualitydata qualityThe degree to which data is fit for purpose: accurate, complete, consistent, timely, valid and unique. Poor quality data undermines analytics, reporting and AI.View full definition → indices. Not a single line tied to revenue, margin, or cost avoidance. The CFO, by multiple accounts from people who were in the building at the time, asked one question: "What did this actually make us?" The room went quiet.

The specific internal details of that meeting are not fully public, and M&S has not confirmed the eleven-minute figure officially. But the broad shape of the story is well-documented in the UK retail press and in subsequent interviews M&S executives gave about their data transformation journey. What is confirmed: the company did undergo a significant restructuring of how it measured and communicated the value of data investment, and that shift preceded, and arguably enabled, a meaningful commercial recovery.

What actually happened

M&S had, by 2019, spent several years building genuinely sophisticated data infrastructure: personalisation engines, customer segmentationcustomer segmentationDividing a market into distinct groups of customers who share similar needs, characteristics or behaviours, so each group can be served with a tailored approach.View full definition → models, supply chain analytics. The capability was real. The problem was that the data organisation had been reporting on the capability rather than on what the capability produced.

The metrics they brought to finance were internally coherent but externally meaningless. Data freshness rates. Model lift scores. PipelinePipelineAll active sales opportunities across the stages of the sales process, together with their combined potential value and probability of closing.View full definition → uptime percentages. These are legitimate operational measures, but they answer questions no CFO is asking. The question the CFO is always asking is some version of: did we make more money, spend less money, or avoid a risk we would otherwise have hit?

After that near-miss, M&S restructured the way its data and analytics teams reported value. The emphasis shifted to attaching every major analytics initiative to a specific business outcome with a measurable financial proxy. A personalisation model was no longer reported as "achieving 15% lift in click-through rateclick-through rateClick-Through Rate (CTR) is the percentage of people who click a link, ad, or call to action out of those who viewed it.View full definition →." It became "contributing to an estimated X million pounds in incremental online revenue in Q3," with the methodology for that estimate transparent and auditable by finance. The company also began using controlled experiments more rigorously, running A/B tests at scale so that the counterfactual (what would have happened without the model) was defensible rather than assumed.

By the time M&S reported its 2022-2023 results, online clothing and home revenue had grown substantially, and the company's digital and data capabilities were cited explicitly in investor communications as a driver of that performance. Whether a direct causal line can be drawn from the 2019 budget meeting to those results involves some inference. But the directional story is coherent and the company itself frames it that way.

Why it still matters

The M&S episode is a clean illustration of a trap that catches data organisations at almost every maturity level:confusing proof of capability with proof of value.

Capability metrics are seductive inside the data function because they are things the data team can actually control and measure precisely. You built the pipeline; it runs at 99.4% uptime. You deployed the model; it achieves 0.82 AUC. These numbers feel rigorous. They are rigorous, just not for the audience you need to convince.

Boards and CFOs operate in a world of financial proxies. They are not being obtuse when they ignore model performance metrics. They are applying the same discipline they apply to every other capital allocation decision: show me the return relative to what I spent. A data team that cannot translate its work into that language is not speaking to its actual stakeholder. It is holding a conversation with itself.

The deeper issue the M&S story surfaces is organisational. When data teams are structured as shared services or cost centres without strong ties to specific P&L owners, there is no natural forcing function to produce business-outcome measurement. Nobody's job depends on proving that the churn model saved X in retention spend. The model gets built, deployed, and then quietly ignored when budget season arrives.

What changed at M&S, structurally, was that data initiatives were assigned business sponsors with skin in the game. A retail director who had signed off on a personalisation project had an incentive to track its financial contribution, because it would show up in their own reporting. That alignment, more than any particular measurement methodology, is what made the difference.

The takeaway for you

The lesson is not "build a better ROIROIReturn on Investment: the ratio of net profit to the cost of an investment. A 300% ROI means each dollar invested returns $3.View full definition → calculator." Plenty of CDOs have built elegant value-measurement frameworks that still failed to move a sceptical CFO, because the framework was assembled after the fact by the data team alone.

The durable lesson is about timing and partnership. Financial credibility for a data initiative has to be co-constructed with finance and with business unit owners before the work starts, not reverse-engineered from outputs after the model is in production. That means agreeing on the financial proxy upfront ("we will measure this as reduction in markdown rate in women's knitwear"), agreeing on the counterfactual methodology, and getting a finance business partner to sign off on the measurement approach before a single sprint begins.

When that happens, the budget conversation in eleven months looks completely different. The CFO is not being presented with a number the data team invented. They are being reminded of a number finance already agreed to track.

M&S rebuilt its data ROI story from the inside out, starting with the questions the business was already asking rather than the answers the data team was already generating. That sequencing is the whole trick. The technical capability matters, but it matters far less than whether the person holding the budget believes the story you are telling about it.