The data you don't have is killing you

A retailer can tell you, to the rupee, what it sold last Tuesday. It can tell you the impressions its last campaign bought, the click-through rate, the cost per conversion. It cannot tell you why the customer bought, or why nine out of ten who saw the ad did not. This is not a gap in the data. It is the data working exactly as designed: to describe symptoms, never to diagnose.  It is “what” data. The “why” data is missing.

What-data is a lagging indicator by nature. It shows you the drop in sales, the churn, the stalled growth, only after the underlying cause has already done its damage. Why-data is the only kind capable of catching a problem while it is still asymptomatic, while it is still cheap and early to treat. An organization with excellent what-data and no why-data is not simply less informed. It is a patient who feels fine, checks a chart that only ever confirms how they felt yesterday, and has no instrument capable of detecting the disease before it becomes symptomatic. The chart looks clean for a long time. That is exactly the problem. The missing data is lethal rather than merely incomplete.

The easy patient

Transaction data is the first data any business collects, because the business cannot function without it. Goods move, sales are logged, a report is generated. It requires no persuasion and no investment in method, which is precisely the trouble. Easy data is popular data, and popular data crowds out the harder, more useful kind.

The quality of what data varies more than most managers admit. Weekly primary sales, broken down to the pin code and the SKU, will show the actual shape of demand. Monthly state wise movement of stocks aggregates dressed up as insight will show almost nothing at all. Digital-native businesses believe they have escaped this problem; in fact, they have swapped a shortage for a flood, and drowning in data is not meaningfully different from lacking it; both leave the patient undiagnosed.

A vocabulary mistaken for a diagnosis

Media has its own chart, and it is even more seductive, because it comes pre-loaded with the vocabulary of science: reach, CPMs, CTRs, VTRs. These numbers feel like insight because they are presented as intelligence. They are, in fact, closer to a thermometer reading: useful, real, and almost entirely silent on what is actually wrong.

Click-through rates across the industry hover near zero. View-through rates are a little better. Yet these are the very metrics on which media budgets are planned and defended. Every platform reports them differently and no one has built the instrument to reconcile them. It is not rare to find a platform's claimed audience larger than the demographic in the population, a fact platforms do not advertise, and marketers have quietly agreed not to ask about. A symptom this large, ignored this consistently, is not an oversight. It is a disorder.

Losing the chart of the whole patient

Zoom out from the campaign to the market, and the picture does not improve. It disappears. Syndicated audits, once the closest thing marketing had to an X-ray of the whole category, have thinned with each passing year. Share of voice, share of spend and share of market, concepts that forced marketers to look past their own results and study a competitor's, have been quietly retired. Not because they stopped mattering, but because the walled gardens made them expensive to obtain and easy to ignore. At best the platforms feed intra-platform indicators designed to spur investment rather than insights. 

This is the point at which the analogy stops being decorative and becomes literal. A doctor who reads only one patient's chart, never the ward's, will miss the epidemic. A business that reads only its own dashboard, never the market's, will miss the competitor eating its growth from a channel it does not measure. And because its own numbers still look stable in isolation, it will feel healthy for exactly as long as it takes the erosion to become undeniable, by which point the cheap interventions are no longer on the table.

Where "why" data actually pays for itself

Skeptics will ask, reasonably, whether any of this is more than a rhetorical flourish, whether "why" data changes a decision that "what" data would have gotten right anyway. It does, and the mechanism is simple. What-data tells you that sales fell in a region; why-data tells you whether the cause is price, distribution, a competitor's product improvement, or a shift in what the customer values, and each of those has a different fix. Treat a distribution problem with a pricing cut, because the sales chart was the only chart on the wall, and the business spends money curing the wrong disease. This is not a hypothetical risk. It is the default outcome of managing by what alone.

The most obscure patient of all

The consumer, in principle the whole reason any of this data exists, has become the hardest figure to see clearly. A need, an aspiration, a brand, an unmet expectation - all exist in the mind of the consumer; rarely vocalized. Market visits, exit interviews, mystery shopping, brand tracking, consumer research, focus groups: the methods that once approximated a real conversation and brought the “why” to life have given way to digital signals, because signals scale and conversations do not.

There is real promise in the newer tools: Affinity algorithms, AI-assisted reading of reviews, early synthetic-data experiments. But promise is not proof, and a large sample without statistical weighting to the population is not a bigger truth. It is a louder guess. The industry's remaining discipline lies precisely in refusing to confuse the two and treating "we have more data than ever" as a reason for confidence rather than an invitation to ask what, exactly, it is data about.

One can see the leakages in the funnel just as the reading on the thermometer but the reason for the leakages stays uncovered. The real consumer stays hidden. Intentional research is designed to uncover the why; inferring from signals still carries a lot of noise. 

The prognosis

Every organization can already answer what happened. Almost none can reliably answer why, and the two are not substitutes for each other, any more than a chart is a diagnosis. A business that only ever measures what happened will not see the disease coming. It will see the fever, after the infection has already spread, and call the fever the problem.

That is what makes the missing data dangerous rather than merely incomplete: it removes the only early warning the organization would have had; or a missed opportunity. The question worth asking is not how much data you have. It is whether any of it would catch the disease while it is still curable, or whether you would only find out once it can no longer be missed.

Premjeet Sodhi is Global Analytics Lead at WPP Media, based in New York, where he builds measurement systems connecting media to business outcomes. The article reflects his personal point of view on the subject.