Multi-Touch Attribution: Building a Model That Connects Marketing to Revenue

Why Last-Click Attribution Persists Despite Being Wrong

Last-click attribution assigns 100% of the revenue credit to the final touchpoint before conversion. It persists because it is simple to implement, easy to understand, and produces clear, defensible numbers. It also systematically overvalues bottom-of-funnel tactics (retargeting, branded search) and undervalues awareness and consideration channels that generated the lead in the first place.

The result: marketing teams optimise for the last-click channels because that's what the attribution model credits. Awareness channels that are genuinely valuable lose budget. Retargeting and branded paid search capture disproportionate credit for conversions they didn't cause. The distortion compounds over time as budget allocation follows attribution rather than leading it.

Attribution Model Comparison

• Last-click: simple, biases toward bottom-funnel, undervalues awareness
• First-touch: gives full credit to acquisition channel, ignores nurture and conversion stages
• Linear: equal credit to all touchpoints, ignores the varying influence of different stages
• Time-decay: more credit to recent touchpoints, appropriate for short sales cycles
• Position-based (U-shaped): 40% first touch, 40% last touch, 20% distributed — appropriate when acquisition and conversion are both high-priority
• Data-driven: uses statistical models to assign credit based on actual conversion path patterns — highest accuracy, requires significant data volume

Data Completeness as the Foundation

Attribution models produce precise answers from whatever data they're given. If your tracking is incomplete — if offline touchpoints aren't captured, if cross-device journeys are broken, if sales touches aren't logged — the attribution model will produce precise but wrong answers, confidently.

Audit your touchpoint capture completeness before building attribution models. Common gaps: phone and in-person sales interactions (usually in CRM but not linked to marketing touchpoints), offline events and trade shows, dark social (sharing that happens outside trackable channels), and cross-device journeys where cookies don't persist.

Using Attribution to Drive Budget Decisions

Attribution models have value only insofar as they inform decisions. The purpose of a multi-touch attribution model is to enable more accurate budget allocation — to identify the channels and tactics that genuinely contribute to revenue, and to shift investment toward them.

Challenge the model's conclusions. If the attribution model suggests that a channel you believed was high-performing is actually contributing less than previously thought, design a holdout experiment to validate: pause the channel for a defined period with a matched control group and measure the impact on conversion rates. This is the only reliable way to distinguish correlation from causation in attribution.