How do you measure marketing's contribution to revenue, and how do you handle attribution?

Connect marketing to revenue through the pipeline: track leads by source through to closed revenue, so you can report cost per customer and return by channel rather than cost per lead. This requires the CRM and marketing platform to be joined up, and getting that plumbing right is often the real work.

Attribution models and their biases:

• First touch — credits discovery, ignores what closed the sale.
• Last touch — credits the final interaction, systematically over-crediting branded search and retargeting and under-crediting everything that created the demand.
• Multi-touch — linear, time decay, or position-based, distributing credit across the journey. More realistic, more complex.
• Data-driven — modelled from your own conversion paths.

Be honest about the limits. No model is truth: cookie restrictions, cross-device journeys, offline touchpoints, and dark social all sit outside measurement. Long B2B cycles involving several people make it harder still.

What to do about it:

• Use attribution for directional comparison over time, not for precise credit allocation.
• Run holdout tests or geo experiments for the channels attribution cannot see. Turning a channel off in one region and measuring the difference is the only genuinely causal method.
• Ask customers how they found you — self-reported attribution is imprecise but catches what tracking misses entirely.

Note: Media mix modelling is worth naming as the approach that is returning to favour precisely because it works without user-level tracking.