Why you're getting this: You've crossed paths with CE Digital somewhere along the way. Every week we share what we're actually seeing inside high-spend ad accounts, plus how we're using AI and our internal data tool, AdSignl, to analyze performance at scale. Short, practical, no filler.
Not for you? No hard feelings. Unsubscribe anytime.
Alright, let's get into it.
This week we're getting into why blended ROAS becomes a worse decision-making metric the bigger your account gets, and why the worst thing you can do when Meta goes sideways is start changing variables to match the chaos. Two patterns we're seeing across the accounts we manage right now.
01. The bigger your account gets, the more granular your KPIs need to be
A blended ROAS of 10:1 looks clean on the dashboard. But if you pull it apart you'll often find a large percentage of the customer pool is retargeting revenue. I.e. customers who would have bought anyway.
If your spend on Meta is above $3K/day, the metrics that should actually be driving decisions are:
New-customer ROAS, measured against LTV through an MTA
New-visit percentage
First-time revenue versus retention revenue
Most large DTC operators we audit have a customer match-rate issue on Meta. On some accounts we've seen as little as 75% match when we're uploading 100% of the customer list.
Which means a chunk of what looks like prospecting is still hitting existing buyers you thought you excluded. Artificially inflating your ROAS on what you believe was a pure prospecting campaign.
Key takeaway: KPI targets should be completely different on new vs. returning customers, and delineating between the two should be a top priority.
Blended ROAS still gets reported. It just stops being the number that drives operating decisions.
Get early access to AdSignl for the customer-cohort and match-rate views we use to pull these numbers apart at the account level.
02. When Meta goes sideways, fly the plane
When Meta goes sideways, the worst thing you can do is start changing variables to match the chaos.
Oftentimes Meta becomes unstable on the backend. Bugs, UI issues, delivery anomalies. The typical Meta instability that leaves advertisers pulling apart the black box.
So when you stack changes on the operator side; new tests, new structures, panic budget pulls; all you're doing is producing mixed signals that make the account worse.
What I've learned over years of macro events like this one: stay consistent on the operator side and let the algorithm side settle.
Tighten budgets if you have to. Hold your structure. Trust your creative. Let it be ugly for a stretch because the dust will settle. Then push spend back and restart testing.
The takeaway: when you're hunting for the answer, eliminate one variable at a time. Stacking changes on top of changes is how you lose the signal entirely.
Most operators who get this wrong aren't making a skill mistake. They're making a context mistake.
If you've only managed one account, every dip looks like the worst dip you've ever seen. No baseline for what a real anomaly looks like versus a Tuesday. So you react to noise because noise is all you can see.
Most agencies have media buyers carrying 10 to 15 accounts at a time. We keep ours to a handful, and we keep the same senior buyer on an account for the full lifespan of the relationship. Not just the first 90 days. The whole arc.
The value of that buyer isn't the SOPs they ran on day one. It's the months of historical context they built watching the account behave through different conditions. That context is what lets them sit still when the dashboard panics. They've seen this movie before. They know how it ends.
Pilots don't panic when the instruments get noisy. They fly the plane.
Two different problems, same underlying mistake: reacting to surface-level signals instead of stepping back to look at what's actually driving the numbers. Stay consistent on the operator side. Get granular on the metric side.
See you next week,
Andrew