Most marketing dashboards answer the wrong question. Impressions, followers, and open rates describe activity; the business wants to know contribution: which spend created customers who would not have arrived otherwise. Closing that gap is less about tooling and more about deciding, in advance, what evidence you will trust.
Pick one north star per channel
Every channel needs a single success metric tied to money: cost per qualified lead for paid search, revenue per recipient for email, assisted pipeline for content. Supporting metrics can diagnose problems, but only the north star decides whether budget grows or shrinks. When everything is a KPI, nothing is.
Attribution is a model, not the truth
Last-click flatters whatever touches the customer last, usually branded search. Platform-reported conversions flatter the platform reporting them, and privacy changes have made click-level tracking blurrier every year. Use attribution to compare campaigns within a channel, and use stronger evidence to compare channels:
- Incrementality tests: pause or reduce a channel in some regions and watch what actually changes.
- Holdout groups: the difference between exposed and unexposed audiences is the real effect.
- Self-reported "how did you hear about us": crude, but it catches dark-social influence nothing else sees.
Build the boring foundation
Consistent UTM conventions, conversions tracked in one agreed system, and a weekly one-page scorecard beat any expensive dashboard nobody trusts. Review the scorecard on a fixed schedule and record the decision each number prompted. A metric that never changes a decision is a candidate for deletion.