With apologies to the Fun Boy Three, It ain’t data what you do, it’s the data that you do it. Why? Because far too many internal data conversations start in the wrong place. Executives demand data lakes, dashboards, and AI pipelines. They want to more data, better data, faster data. What they rarely discuss is the question that separates the useful from the decorative: What would the data have to show before we actually did something about it? Blockbuster had plenty of data, on declining store footfall, customer anger over late fees, and the steady spread of broadband. Nokia had data. Neither firm did much with it. Data does not drive reinvention. Questions do.
Most corporate data functions are not, despite what they claim, built to challenge. They are built to confirm. Dashboards are calibrated (consciously or otherwise) to measure the KPIs that make the existing business look healthy, not the signals that might require doing something about it. The result is what we might call the answer machine: a sophisticated and expensive apparatus for telling management what it already thinks.
BCG research across 850-plus companies found that only 35 per cent of digital transformation initiatives globally meet their value targets. The dominant explanation offered is usually poor technology choices or weak change management. A more plausible reading is that most companies are using data to optimise current operations rather than to question whether those operations are worth performing.
Back in 2024, PwC’s Global CEO Survey asked business leaders if they believed their organisations would be economically viable within ten years if they continue on their current course. Nearly half (45 per cent) said no, which to all intents and purposes means that the world’s most powerful business leaders can see the iceberg but are, more or less, maintaining their heading. Knowing what the data says and being willing to act on it are different things entirely.
The question that changed everything
Rolls-Royce is, ostensibly, a maker of aircraft engines with a century of engineering heritage, a famous badge, and technology of extraordinary complexity. Ask someone at Rolls-Royce what business the company is actually in today, though, and the answer is something rather different: availability. Not engines, uptime.
In the 1960s, the company pioneered a service model it called Power by the Hour, through which it charged airlines for engine performance rather than the engines themselves. The model matured into TotalCare, a subscription-based service that absorbed all maintenance risk, charged at a fixed rate per flying hour. It was built around one question: What does an airline actually want to buy? The answer was not a piece of machinery, but flying hours it can count on.
That question required data to function. Engine health monitoring, real-time telemetry, predictive maintenance algorithms. But the data came second. The question about Rolls-Royce’s own business model, about whether it was selling the wrong thing entirely, came first. Today, the company has accumulated some 70 trillion data points across its connected fleet, a repository no competitor can retroactively replicate. What looks from the outside like a data advantage is, at its root, the compounding consequence of a question asked six decades ago. Every new TotalCare contract adds to it. The gap between Rolls-Royce and its rivals is widening, not closing.
That is the pattern worth studying. Data-led reinvention does not begin with data. It begins with a question that existing incentive structures actively discourage.
The dark matter problem
There is a related failure mode, harder to spot. Companies collect enormous amounts of data on what customers do, and almost none on what they don’t, such as the products never purchased, the complaints never filed, the digital paths that end in abandonment. The most revealing data most organisations possess is the data they have never thought to gather.
MuleSoft’s 2025 Connectivity Benchmark Report found that the average large enterprise runs 897 applications, of which only 29 per cent are properly integrated. Each disconnected system is a blind spot dressed up as an asset. Companies with strong integration achieve more than ten times the return on AI investment of those without it—which tells you that the data problem most executives think they have (not enough) is usually the wrong problem (too fragmented to read).
John Deere is a useful case in point. The company collects equipment performance data not primarily to manage warranty claims but to understand whether its products actually fit how farmers work, a question that has driven a sustained pivot toward precision agriculture and subscription data services that now rival hardware revenues. What began as operational telemetry became, over time, a question about what business Deere is in. Same data, different question.
A confession, not a conclusion
None of this demands a new chief data officer, a platform migration, or a consulting engagement. It demands something considerably harder to procure, namely a willingness, at board level, to treat data as a witness rather than an alibi. When was the last time your organisation’s data made someone genuinely uncomfortable? If the honest answer is ‘rarely’ or ‘I can’t remember’, the data operation is probably an expensive confirmation machine.
The companies that have reinvented themselves with data share a single quality. They were prepared to ask questions whose answers might require dismantling something that was still working. Rolls-Royce asked what an engine was really worth to an airline. Amazon asked what its logistics infrastructure might be worth to everyone else, and built AWS. The question, in every case, came before the data strategy.
Blockbuster had 65 million customers and a five billion US dollars valuation in 2000. It had the data. What it lacked was a management team willing to sit with what the data was already answering. The numbers were not the problem. The questions were.
Photo: Dreamstime.
