ROAS is the metric that wins the deck and loses the partnership. Not because the number is wrong, but because by the time it moves, the cause is already weeks behind.
A 27x ROAS reads like a current event. It is actually a 4-week-old event. The campaigns producing that ROAS were briefed 6 weeks ago, shot 5 weeks ago, edited 4 weeks ago, launched 3 weeks ago, optimised 2 weeks ago, and started returning the multiple now. If next month's ROAS halves, the cause already happened. You are reporting on history.
The leading metric we run inside Ignis is save rate. Save rate is the percentage of viewers who saved the post for later. It tells you, in real time, whether the audience found the piece valuable enough to revisit. Save rate predicts CTR which predicts conversion which predicts ROAS, in that order, with about a 3-week lead.
Why save rate works as a leading indicator. The save action is high-friction relative to a like or a share. Liking is a reflex. Sharing is a statement. Saving is a personal commitment to come back. The platform algorithms know this and weight save rate heavily in distribution. A piece with a 4 percent save rate will outrun a piece with a 0.4 percent save rate by a factor of 5 to 10 in 30-day reach.
We watch save rate at the post level inside the first 24 hours. Pieces tracking above the rolling 90-day median get amplified. Pieces tracking below get killed quickly. The decision is mechanical, not creative.
The result is that Ignis ad accounts almost never have a "bad month" because we caught the leading indicator three weeks before the lagging one moved. The clients who have stayed with us through three reporting cycles have all seen the ROAS chart get smoother over time, not because the work got safer but because the kill-and-amplify decisions got faster.
If your current dashboard leads with ROAS and that is the metric you optimise on, you are flying with a 4-week-delayed altimeter.