A channel reported 8x return and 0% lift. How can both be true?

A channel reported 8x return and 0% lift. How can both be true?

A direct-to-consumer brand's paid-social reported an 8:1 return on ad spend. A simultaneous conversion-lift study, run by the platform, found the incremental lift was statistically zero. The question that defines modern attribution: how does a channel "earn" 8x while causing nothing?

The two measurements. Reported ROAS is attribution arithmetic — conversions the model assigns to the channel, divided by spend. Incremental lift is a randomized experiment — conversion rate of an exposed group minus a withheld control group. They answer different questions: "what did the model credit?" versus "what did the ads cause?"

What the case showed. In documented DTC conversion-lift studies, channels with strong reported ROAS frequently showed lift far below their attributed return; several prospecting-versus-retargeting comparisons found retargeting at 8-10x ROAS but near-zero lift, while prospecting at 2-3x ROAS carried most of the true incremental conversions. The model was crediting the channel that closed warm users it didn't warm up.

The nuance. High ROAS and zero lift coexist whenever a channel sits late in journeys it didn't initiate. The arithmetic is correct; the causal claim it implies is false. This is correlation and causation wearing the same number.

Bottom line for practitioners: ROAS ranks channels by where conversions are attributed, not by what they cause — and the two rankings can be opposites. Run platform lift studies on your highest-ROAS channels specifically, because high reported return is exactly where non-incremental credit hides. The 8x channel may be your least valuable.