YouDon’tHaveanAnalyticsProblem

You Don’t Have an Analytics Problem

The analytics industry is enormous. Tools, certifications, dashboards, attribution models, tracking stacks, data warehouses, visualisation layers. The collective investment is staggering.

And yet most organisations produce data that nobody acts on.

Isn’t it strange that we keep looking for the problem inside the tool?

The real problem is further back. Further back than the dashboard. Further back than the KPI. Further back than the channel or the campaign or the creative. The failure is structural, and it happens before anything is built or measured or reported. It happens when organisations treat measurement as a data problem rather than a thinking problem.

There is a framework that exists to correct this. It is called the Digital Marketing & Measurement Model. Notice the word marketing sits before measurement. That ordering is not accidental.

What a dashboard cannot tell you

A dashboard is an answer machine. It answers questions. But it cannot tell you which questions matter. It cannot tell you whether a 1.2 million view YouTube campaign was a success or a failure. It cannot tell you whether a 4% conversion rate is something to celebrate or something to investigate. It cannot tell you, after six weeks of optimisation, whether you are closer to what you set out to achieve.

Numbers without intent are just numbers.

This is the version of the Measurement Trap that analytics professionals rarely name. The trap is not that organisations measure the wrong things. It is that they measure without first agreeing on what right would look like. The presence of data creates the sensation of understanding. A populated dashboard performs the work of comprehension without delivering it.

Being data-driven has become vocabulary. The behaviour has not always followed.

The sequencing problem

The Digital Marketing & Measurement Model is, at its core, an argument about sequence.

It holds that five things must be defined in a specific order before any measurement work begins. Business objectives first. Goals for each objective second. Key performance indicators third. Numerical targets fourth. Analytical segments fifth.

Each step is impossible to complete honestly without the prior one. You cannot identify the right KPI before you know the goal it is serving. You cannot set a meaningful target before you know what you are measuring. You cannot define useful segments before you know what question you are trying to answer.

The model is not a planning template. It is a forcing function. Its value is not that it produces a document. It is that it requires a specific type of thinking before the tools are switched on.

Start with the hardest question. Why does this website exist? Why is this campaign running? What would have to be true for this digital initiative to be considered a success? These are not analytics questions. They are business questions. And they require the people responsible for the business to answer them.

Senior executives play a central role in step one precisely because the answers they give constrain everything that follows. If the executive cannot answer the question, the measurement work built beneath it will be built on sand. More data will not rescue it.

Three areas. Every time.

A complete measurement model covers three distinct territories. Acquisition: how people arrive. Behaviour: what they do when they get there. Outcomes: what they do that matters to the business bottom line.

A model that only covers one of these is incomplete. A model that covers acquisition and ignores outcomes has confused activity for purpose. A model that defines outcomes without thinking about the behaviour required to reach them has skipped the middle of its own argument.

The three territories matter because they map to different questions with different owners. Marketing typically owns acquisition. Product and UX own behaviour. Finance and commercial leadership own outcomes. A measurement model that spans all three forces a conversation across those three functions. That conversation is often the most valuable thing the model produces. Not the spreadsheet. The conversation.

When metrics become the objective

There is a specific failure mode that the DMMM is designed to prevent, and it is worth naming precisely.

A metric begins as a proxy for something that matters. Page views once tracked genuine audience interest. Email open rates once tracked genuine engagement. Click-through rates once tracked something approaching intent. These are reasonable proxies. They are never the thing itself.

Without a chain of accountability from business objective to goal to KPI, the proxy becomes the objective. The original purpose disappears. The metric survives. People begin optimising for the metric because the metric is what gets reported, and what gets reported is what gets managed, and what gets managed is what the system believes matters.

Proxy Collapse is not a measurement error. It is a structural outcome of measuring without prior thinking. The DMMM breaks the chain before it forms by anchoring metrics to goals and goals to objectives before the measurement infrastructure is built. The anchor cannot be installed retrospectively. It must exist before the first number is collected.

Setting targets for each KPI before seeing the data performs the same function. It looks trivial. It is not. A target set before the campaign runs reflects what the business actually believed was achievable. A target set after the data arrives reflects something else entirely.

The strongest objection

The most serious challenge to this model is not about its logic. The logic is correct. The challenge is about its conditions of possibility.

Business objectives change faster than planning cycles. Campaigns run for weeks. Markets shift. A measurement model defined in week one may be obsolete by week three. Modern marketing practice explicitly values iteration and adaptation over upfront definition. The DMMM looks, from a certain angle, like waterfall planning applied to measurement. And waterfall planning has a well-documented track record.

This is a real objection. It should not be waved away.

But it confuses the frequency of planning with the necessity of thinking. The model does not prevent iteration. It makes iteration legible. Without agreed objectives and targets, you cannot tell whether your iteration is moving toward something or simply moving. Undirected activity looks exactly like agility in motion. The difference only becomes visible in outcomes.

The subtler version of the objection is political. Who actually has the standing to pull senior executives into a room and extract honest answers about what a campaign is for? In many organisations, the analyst asking this question does not have that standing. This is true. The practical answer is that an incomplete DMMM still beats an absent one. Even partial anchoring to business intent changes the quality of every analysis produced beneath it. And a documented model, however partial, is a negotiating instrument. It creates a shared reference point where none existed before.

The contract

The model has a final property that is easy to miss. When it is complete and agreed, it functions as a contract.

Not a legal document. Something more durable than that. A shared definition of what success means, co-signed by the people responsible for achieving it. The analyst knows what to measure. The executive knows what success looks like. The marketer knows what behaviour they are trying to drive. Everyone is solving the same problem.

This contract has a specific consequence. When you crack open the dashboard and see the numbers, you know immediately whether they are good or bad. Not because the tool told you. Because you decided in advance.

That is the difference. That has always been the difference.

The analytics industry will continue producing better tools. Faster dashboards. Smarter attribution. More granular tracking. None of it changes the prior question. And the prior question is always the same.

What are you actually trying to achieve?

-----

Before you open the next report, three questions worth sitting with.

If you stripped your analytics reports of all the numbers, would what remained describe what the business is actually trying to achieve?

Could your most senior stakeholder look at your KPI dashboard and know, without being told, whether it was a good week or a bad week?

Are your segments defined by what matters to the business, or by what your tool makes easy to segment?