Dear [reader_firstname],

Early in my career I worked alongside a head of production who had been in the business long enough to have stories most people would not print. He had built client relationships over years, sometimes decades, and he once told me about the sharpest salesperson he had ever known.

This person was never in the office. Always out. Coffees, lunches, quick check-ins, follow-up calls. Not performative networking. Actual relationship maintenance. He knew how many kids his clients had. He knew birthdays, renovation projects, family situations. When something happened in a client’s personal life, he sent a bottle from himself, not from the company. All of it recorded in what he called his little black book.

I have enough respect for anyone reading this not to explain why that model is unsustainable. It belongs in pub stories now, and that is where it stays. But if you look past the inefficiency and focus on the knowledge it actually produced, the picture changes. For most businesses operating at any real scale, that kind of understanding had never existed in the first place. The infrastructure that replaced it was not replacing a relationship model. It was filling a space where understanding at scale had never been achieved. The problem is not that it filled it. The problem is what everyone mistook it for.

Even when some of that knowledge made it into a CRM, what arrived was the shallowest version of it. A birthday field. An automated email. “Dear [first_name],” and a stock message nobody wrote and nobody reads. The data point transferred. The understanding did not. And nobody in the organisation can tell the difference from the dashboard, because the dashboard shows “birthday campaign sent” and the open rate looks fine.

The trade was rational. What got lost in it is what this series is trying to examine.

In 2016, a study published in the Journal of the Academy of Marketing Science compared seventy thousand customer satisfaction surveys against responses from over a thousand marketing managers at the same companies. The researchers were not testing whether the data was accurate. They were testing whether managers knew what their own customers thought.

The results were striking. Managers across a wide cross-section of industries consistently overestimated how satisfied their customers were. They significantly underestimated how many customers had complained. And their understanding of what actually drove customer satisfaction was often misaligned with what the data showed.

These were not small companies without resources. Most had dedicated customer feedback systems in place. They had invested heavily in them. The infrastructure existed. The data was being collected. The surveys were being conducted at scale.

The researchers’ conclusion was blunt. Despite often representing the single largest line item in market research budgets, the customer feedback systems at most firms were not performing an effective management control role. The data was there. The understanding was not.

This is not an isolated finding. In 2019, Deloitte surveyed over a thousand executives at large companies and found that two thirds of those at senior manager level or above said they were not comfortable accessing or using data from their own tools and resources. Not unfamiliar with the tools. Not lacking access. Uncomfortable. Even at companies that described themselves as having strong data-driven cultures, more than a third of respondents still expressed that discomfort.

Around the same time, a KPMG survey of thirteen hundred CEOs across eleven economies found that two thirds said they had overlooked insights provided by data and analytics models because those insights contradicted their own experience or intuition. Not that they lacked access to the data. That they actively overrode it. And a SAS survey found that forty-two percent of data scientists said their results were not used by business decision makers.

These are not stories about bad data or broken tools. The infrastructure works. The dashboards function. The reports arrive on time. What these findings describe is something more specific and more difficult to address. A persistent, documented gap between what the measurement infrastructure produces and what the people running the business actually trust enough to act on.

There are two ways to read that gap, and most commentary has chosen the easier one.

The easier reading is that this is a skills and culture problem. Executives need more data literacy. Organisations need to invest in training. The tools need better interfaces. The data scientists need to communicate better. This is the reading that produces conference talks and transformation roadmaps and vendor solutions. It is not wrong. It is just incomplete.

The harder reading is that the executives are sensing something real.

Not that the data is inaccurate. But that the infrastructure, despite its sophistication, is not actually producing what they need to make the decisions they are responsible for. It is producing measurement. It is not producing understanding. And the gap between those two things, which is easy to ignore when things are going well, becomes impossible to ignore when something changes and the numbers stop explaining what is happening or why.

This is not a failure of analytics. It is a structural consequence of how commercial measurement infrastructure was built.

The operating model for commercial knowledge in most organisations was not designed from first principles. It accumulated.

Part one traced how this happened in detail. The short version is that a series of individually rational decisions, adopt the free analytics platform, use the tag manager that integrates with it, follow the campaign taxonomy the agency already uses, import lead-source data into the CRM using the same conventions, created an infrastructure where one ecosystem’s interpretive framework became load-bearing at every level of commercial reporting.

Anyone who has worked inside a digital team at any scale will recognise the feeling this produces. You have more dashboards than you have ever had. More data points, more segments, more attribution reports, more ways to slice performance by channel and campaign and audience. And yet in the meetings that actually matter, the ones where someone has to explain why the quarter went the way it did or what to do next, the room still relies on a handful of people who have been around long enough to know what the numbers do not say. The infrastructure provides the language. The understanding, when it exists at all, lives in the experience of individuals rather than in the systems the organisation has built. That is not a data maturity problem. That is a sign that the infrastructure was never producing what the organisation thought it was.

What matters here is the consequence. The infrastructure was built to measure activity inside a digital environment. Sessions, clicks, conversions, source attribution, channel performance. It was built to answer one question. What happened? And it answers that question well enough that entire organisations have built their reporting, their budgets, and their strategic planning around the answer.

But the question most senior leaders are actually trying to answer is different. They are trying to answer a different question. What should we do? And the distance between “what happened” and “what should we do” is where the gap lives.

Knowing that conversion rate dropped by eight percent tells you something happened. It does not tell you whether the cause is pricing, competition, a shift in customer expectations, a deterioration in product experience, a change in the broader market, or simply that the traffic mix shifted in ways the attribution model cannot distinguish. The infrastructure can show you the what. It was never designed to show you the why. And most organisations have been stretching “what” to cover “why” for so long that the stretch has become invisible.

This is what the executives in those surveys are sensing. Not that the numbers are wrong. That the numbers, however precise, do not bridge the gap to the decisions they need to make. The discomfort is not with the tools. It is with the distance between what the tools provide and what the role demands.

Some organisations invest separately in qualitative research, customer panels, ethnography, the kind of direct understanding that sits outside the measurement stack. But for the majority, the measurement infrastructure became the primary input for commercial decisions, and the gap between what it provides and what leaders need was absorbed rather than addressed.

If the gap were simply a matter of incomplete information, it would be manageable. Every organisation has blind spots. You identify them, you commission research, you fill them in.

But this is not a gap in information. It is a gap in infrastructure. And the difference matters because it determines what happens when the gap is tested.

The test arrived with GDPR. The expectation was that the compliance burden would force businesses to re-examine what they actually knew about their customers, where that knowledge came from, and whether the infrastructure supporting it was fit for purpose.

That is not what happened in most cases. The regulation created complexity that favoured scale and operational convenience. Organisations with less engineering capacity and less internal specialist knowledge reached for solutions that could handle the new requirements without a fundamental rebuild. The ecosystem that was already in place offered exactly that. Consent frameworks, updated analytics, server-side implementations, all designed to maintain compatibility with the dominant measurement infrastructure while meeting stricter requirements.

The first-party data strategies that the industry recommended were, in practice, often implemented through the same ecosystem’s tooling. The strategic imperative to own your data frequently ended up meaning collect your data and then process it through the same infrastructure so it remains commercially useful.

From what I have seen, the companies that entered the regulatory era with the least independent knowledge of their customers generally emerged with less independence, not more. The regulation raised the floor for data practices, which was valuable. But it also raised the barrier to building anything outside the dominant framework.

The regulation did not create the dependency. It exposed the absence beneath it.

That absence reframes the trust problem described earlier.

When two thirds of senior executives say they are not comfortable with their own data, and two thirds of CEOs actively override analytical insights in favour of experience, the standard response is to treat this as a maturity gap. More training, better tools, stronger culture. And those things may help at the margins.

But there is another possibility, and the research supports it. I believe the discomfort is not irrational. The executives are right to sense that something is missing, because I do.

What is missing is not better data. It is independent understanding. The kind that does not come from a dashboard, that cannot be configured during an implementation, that does not arrive pre-shaped by someone else’s framework.

The kind that would let you answer the question “what should we do?” without first translating it into “what do the metrics say?”

Most organisations have spent fifteen to twenty years building increasingly sophisticated measurement infrastructure while the capacity for independent commercial judgement has quietly atrophied. Not because the people got worse. Because the infrastructure made it feel unnecessary. The dashboard answered enough questions, in enough detail, with enough apparent precision, that the slower, harder, more ambiguous work of building genuine customer understanding felt like a luxury. Something for brand consultancies and design firms. Not a core operational capability.

And now the infrastructure that provided that false sufficiency is facing its most significant disruption in two decades. The discovery layer is shifting. The measurement conventions are under pressure. The identity frameworks are fragmenting.

What I keep arriving at is not a technical question. It is a commercial one. If the infrastructure that has been providing your commercial understanding is disrupted, and you never built anything independent underneath it, what do you actually know about your customers that is yours?

Not a field in a CRM. Not an automated workflow. Not an open rate that tells you a message was delivered but nothing about whether it mattered.

Yours. The way a little black book was yours.

That is not a rhetorical question. It is an operational one. And the speed at which it is arriving is faster than most organisations have prepared for.