Every few months, the marketing measurement conversation resets. Attribution is broken. The cookieless future has made tracking impossible. Media mix modelling is too expensive for most budgets. The tools are not sophisticated enough. These are real challenges, and they are worth taking seriously. But they are also frequently invoked as explanations for measurement problems that originate somewhere far simpler: the team has not agreed on what it is trying to measure, and why.
Before the tool problem, there is a question problem. And no attribution model, however sophisticated, can compensate for the absence of clear, commercially grounded questions about what the marketing activity is supposed to achieve.
The definition problem that precedes the measurement problem
Ask ten people on a marketing team what success looks like for the current campaign and you will typically get ten different answers. Some will describe volume — impressions, reach, clicks. Others will describe engagement — shares, comments, time on site. A few will describe commercial outcomes — leads, opportunities, revenue influenced. If the team has not agreed on which of these constitutes success before the campaign launches, measurement becomes a post-hoc exercise in finding evidence that the campaign did something positive, rather than an honest assessment of whether it did the specific thing it was supposed to do.
Vanity metrics and why teams reach for them
Vanity metrics — impressions, follower counts, total page views — are popular in marketing reports not because they are the most useful measures of commercial performance, but because they tend to go in the right direction most of the time. They create the appearance of progress even when commercial performance is flat. Teams reach for them under pressure because they are easier to produce and easier to defend than the metrics that actually connect to revenue — qualified leads generated, pipeline influenced, customer acquisition cost, retention rate by cohort. The discomfort of measuring what matters is more productive than the comfort of measuring what moves.
The most common measurement problem is not that the numbers are wrong. It is that the team is measuring the wrong numbers with great precision.
Building measurement around questions, not tools
The right way to build a measurement framework is to start with the commercial questions your business needs marketing to answer — not with the metrics your analytics platform makes it easy to produce. For a B2B marketing team, the most important questions are usually variants of: which activities are generating qualified pipeline? Which channels are producing the highest-quality leads at the lowest cost? At what points in the buyer journey are prospects most likely to disengage? Which content is directly associated with shortened sales cycles?
Each of these questions has a specific metric or set of metrics that would answer it. The discipline is to identify those metrics, ensure you have the tracking infrastructure to produce them reliably, and commit to reviewing them on a cadence that allows decisions to be made with them — not simply reported on.
Attribution is a model, not a fact
Every attribution model is an approximation. Last-click attribution overstates the contribution of the final touchpoint and understates the role of earlier brand interactions. Multi-touch models distribute credit more fairly but depend on complete data, which most organisations do not have. Media mix modelling is more statistically robust but operates at a level of aggregation that obscures channel-level decisions. The solution is not to find the perfect attribution model — it does not exist. The solution is to understand what your current model misses, calibrate your expectations accordingly, and supplement it with other evidence sources: controlled experiments, customer surveys, sales team intelligence, and holdout tests.
The practical discipline of useful measurement
There are three things high-performing marketing teams do around measurement that lower-performing teams do not. First, they define their metrics before the campaign launches, not after. Second, they distinguish between operational metrics — ones used by the team to manage day-to-day execution — and strategic metrics — ones used to demonstrate commercial value to leadership. They manage the business on the former and report to leadership on the latter, with different reporting formats for each. Third, they run experiments — controlled tests with proper holdout groups — at least quarterly, so they have a source of causal evidence rather than purely correlational data.
None of this requires a sophisticated martech stack or a dedicated data science team. It requires clarity about what you are trying to achieve, the discipline to measure it consistently, and the willingness to report honestly on what the numbers say — including when they say the activity is not working.
Fix the questions first. The measurement will follow.