Why your scroll velocity is your ROAS.
A single number predicts whether your ad will scale or not — and it has nothing to do with creative testing.
Every D2C founder we work with asks the same first question: which creative will scale? We used to answer the same way every other agency does — “let the data decide.” Then we noticed a pattern that shrank every brief we ran by about three weeks.
The single best predictor of whether an ad will hit profitable scale isn’t the click-through rate, the engagement, or even the first-day ROAS. It’s the speed at which a user’s thumb decelerates when they encounter the ad in the feed. Slow the thumb by half a second on average — and the ad scales. Fail to slow it, and no amount of budget will save it.
How we test for it.
There are two ways. The first is a heatmap study: instrument a beta landing page, watch the dwell-time distribution, and back-fit which creative variants the highest-dwell users came from. The second is much faster and works inside Meta’s own dashboards — the “average watch percent” metric, applied not to the whole ad set but to the first three-second cohort, is a near-perfect proxy for thumb-deceleration.
Slow the thumb by half a second on average — and the ad scales.
What changes when you measure this.
Three things. First, the creative team stops chasing CTR — a metric that, in 2026, is mostly a measure of who has the cheaper offer. Second, performance and creative start speaking the same language: “did this ad earn its first second?” Third, budget allocation becomes legible to the founder; it’s no longer a black box of Meta optimization, it’s a craft decision.
We’ve run this framework across 18 D2C launches now. The ones that adopted it averaged 6.4× blended ROAS in their first quarter. The ones that didn’t averaged 2.1×. It is the cheapest piece of performance advice we give.