Walk into any DTC brand's finance meeting and the question is the same: "what's the ROAS on Meta?" Someone pulls up the dashboard. The dashboard pulls last-click attribution. The number looks good, or it doesn't, and someone makes a budget decision.
If the last-click number is what you're optimizing toward, you're optimizing toward the wrong thing.
What last-click actually measures
Last-click gives 100% of the credit to the final ad before purchase. In practice, that means:
- Branded search always wins. Of course it does — by the time someone searches your brand name, every other channel already did the work of putting your brand in their head.
- Retargeting always wins. By definition retargeting reaches people who already engaged with your brand. The last-click number for retargeting is enormous because everyone retargeted is already deep in the funnel.
- Top-of-funnel awareness always loses. The TikTok creative that introduced your brand to a stranger, the YouTube pre-roll that built recognition over six weeks — last-click sees none of that work.
Why CFOs love it anyway
Last-click is defensible. It's the number you can point at and say "this ad → this click → this purchase." There's no modeling, no probabilistic guessing, no "trust me." For a CFO who's just been burned twice by attribution-vendor handwaving, that simplicity is the point.
The problem is that defensible isn't the same as accurate. Last-click systematically overcredits the bottom of the funnel and starves the top. Brands that scale on last-click usually realize, six months later, that their CAC went up and their organic + branded search traffic went down — because they cut the campaigns that were creating demand in the first place.
What to do instead
Don't pick a "real" attribution model. Instead, watch multiple models in parallel and look at where they disagree. The disagreement is the signal:
- Campaign strong on last-click, weak on first-touch? It's converting demand other channels created. Don't scale; protect the channels feeding it.
- Strong on first-touch, weak on last-click? It's doing real awareness work. Don't kill it because retargeting isn't crediting it back.
- Weak across all models? Now you have a defensible kill.
- Strong across all models? Scale.
What we tell brands using Persequor
The dashboard shows last-click ROAS, first-touch ROAS, linear ROAS, time-decay ROAS, and position-based ROAS for every campaign — same data, five lenses. The right move on most campaigns isn't visible in any one of those numbers. It's visible in the gap between them.
Last-click is defensible. Multi-touch is correct. The CFO meeting goes better when you bring both.
One question we ask every demo customer: "what's a campaign you killed last quarter?" Half the time, when we run their data through multi-touch, that campaign was the unsung hero of two others that look fine on the dashboard.