Measurement Myths That Refuse to Die

Every February, planning season ends and audit season begins. Clients bring us last year's numbers and ask what they mean, and every year I am struck less by what the dashboards say than by what people believe about them. The tooling has improved enormously. The mythology has not budged. Here are the five myths we spent the most breath arguing against this quarter, written down so I can send a link next time.

Myth one: attribution is a truth machine

Attribution models do not observe reality. They allocate credit according to assumptions someone chose, and the choice of model can swing channel-level ROI by multiples without a single customer behaving differently. Last-touch flatters the bottom of the funnel. Data-driven models flatter whatever the algorithm can see, which excludes most of what brand work does.

We treat attribution as bookkeeping, useful for operational comparisons within a channel, and nearly useless for the question executives actually ask, which is "what would happen if we stopped?" Only incrementality testing answers that. Geo holdouts, audience splits, go-dark experiments. They are slower and less flattering than a dashboard, which is exactly why you should trust them more.

Myth two: zero-click means zero effect

The panic of the moment. Search behavior has migrated into answer engines and assistants, referral traffic to informational pages keeps sliding, and the reflexive conclusion is that the investment is wasted. But being the brand the assistant names is not a traffic event. It is a preference event. We are seeing clients with collapsing informational click-through and rising branded demand, direct traffic, and share of citation. The click was always a proxy. The proxy broke. The thing it was a proxy for did not.

The practical shift: measure presence and preference, not just sessions. Share of citation across major assistants, branded search volume, and incrementality on demand. It is fuzzier. It is also honest.

Myth three: the dashboard is the measurement strategy

A dashboard is a display surface. A measurement strategy is a short list of decisions you intend to make and the evidence that would change them. Most reporting stacks we audit contain hundreds of metrics and support approximately zero decisions. Our rule with clients is blunt: every number we report must be attached to an action someone would take if it moved. If nobody can name the action, the metric is decoration, and decoration has a maintenance cost.

Myth four: brand and performance are different budgets

This one is organizational, not analytical, and it does the most damage. The split made sense when the channels were different. It makes no sense when a single assistant answer is simultaneously your brand impression, your comparison shopping moment, and your conversion path. The brands winning right now run one demand system with one set of questions:

• What do we want to be known for, and is that knowledge spreading?
• Where does preference actually form for our category?
• What spend is incremental, and what is just harvesting demand we already created?

Myth five: more precision means better decisions

The quiet one. Teams delay good decisions for months waiting for measurement certainty that will never arrive, while competitors act on directionally solid evidence. Marketing measurement is closer to navigation than accounting. You want to know you are pointed roughly right and moving. You do not need the position to six decimals, and anyone selling you six decimals is selling you the myth from section one with extra steps.

The meta-point under all five: measurement is a craft discipline like everything else we do. It requires taste about what matters, honesty about uncertainty, and the discipline to design tests instead of collecting exhaust. The tools will keep improving. Whether the thinking does is up to us.