Fractional Chief AI Officer engagement to build a cross-media attribution product — correctly measuring the contribution of TV, radio, and billboard alongside digital channels for the first time.
A marketing analytics company building a cross-media attribution product for retail and e-commerce clients engaged us in a fractional Chief AI Officer capacity. Their clients were spending across a broad mix of channels — paid search, SEO, display, email, and significant investment in offline media: TV, radio, and billboards. The challenge was that their existing attribution methodology could not meaningfully account for what offline channels were actually contributing.
Traditional attribution is built on clickstream data. A channel gets credit when a user clicks a link — which means every channel that doesn't generate a click is invisible to the model. For digital channels this is imperfect but workable. For TV, radio, and billboard it fails entirely: no one clicks a television ad. These channels were showing zero contribution in standard attribution reports regardless of how much spend went through them or how much awareness they were building.
The downstream effect was predictable. Clients saw their paid search and retargeting channels consistently taking credit for conversions — because those are the last touchpoints before purchase — while offline channels appeared to do nothing. Budget decisions were being made on a picture that was systematically wrong, consistently rewarding channels that close deals and starving the channels that create the demand in the first place.
Our role was to define the attribution methodology at the product's core and oversee its implementation. The fundamental shift was moving away from click-based attribution entirely and towards a model built on conditional probability — using all customer journeys, both converted and non-converted, to measure each channel's genuine incremental contribution to purchase decisions.
For offline channels specifically, the model infers contribution from the relationship between channel activity and conversion patterns — using temporal and geographic signals to establish where TV, radio, and billboard sit in the real customer journey, even without individual-level touch data. TV, which had previously been invisible in attribution reports, was correctly positioned as an early-journey awareness channel that initiates purchase intent rather than closes it.
Budget optimisation was layered on top — enabling clients to model the effect of shifting spend across all channels simultaneously, including offline, before committing to a plan. For the first time, clients had a complete picture of what each campaign was actually delivering across the full media mix: not just the digital channels they could already track, but the TV spots, radio placements, and billboard activity that had previously been invisible in their reporting.
Let's talk about what's achievable for your specific context.
Start the Conversation →