For most of the last decade, the story marketers told themselves about a struggling ad account was a targeting story: wrong audience, wrong exclusions, wrong lookalike percentage, budget not reaching the right people. That story is mostly out of date now, and accounts that keep diagnosing performance problems as targeting problems keep missing the actual lever that's stuck.
What actually changed
Since Apple's iOS 14.5 App Tracking Transparency rollout, the amount of granular, cross-app user-level data available for manual targeting has shrunk dramatically. In response, every major platform — Meta with Advantage+, Google with Performance Max — has shifted toward algorithmically-driven, broad-targeting delivery. The platform's machine learning model finds the audience now. Manual interest stacking and narrow custom audiences, the targeting toolkit that used to be the differentiator, matter far less than they did five years ago.
That doesn't mean targeting is irrelevant — audience exclusions, retargeting windows, and top-of-funnel vs. bottom-of-funnel structure still matter. But the fine-grained "we found an audience nobody else knows about" edge that used to separate good media buyers from mediocre ones has largely been automated away by the platforms themselves.
So what's left to differentiate on?
Creative. When the algorithm is finding the audience, the variable that determines whether that person actually converts is what they see when the ad shows up in their feed. Two accounts running identical targeting, identical budgets, and identical landing pages will post wildly different ROAS if one is running fresh, native-feeling creative and the other is running the same three static ads it launched with three months ago.
This is the uncomfortable part for a lot of owners: it means the fix for a stalling account often isn't "let's adjust the audience" — it's "let's completely rebuild the creative." That's a harder conversation, but it's the honest one.
The testing framework that actually works
Treating creative like a real testing discipline — not a monthly refresh — is what separates accounts that keep improving from accounts that plateau. The framework:
- Hypothesis first. Every new creative concept should test something specific — a new hook, a different pacing, a UGC-style approach vs. a polished static. "Let's just try something new" isn't a hypothesis; it doesn't tell you what you learned when the test ends.
- Enough volume to read a result. A concept that got $40 of spend and no conversions didn't fail — it never got tested. Set a minimum spend threshold per concept before you call a winner or a loser.
- Kill fast, scale faster. Concepts that clear the bar get more budget immediately. Concepts that don't get killed without sentimentality — creative fatigue sets in fast, and a "maybe it just needs more time" ad set is usually just burning budget.
- Systematize what wins. When a hook format wins, don't treat it as a one-off — build variations of it, apply the pattern to other products or offers, and add it to a living hook library so the next test starts from a proven foundation instead of a blank page.
The thumb-stop is the new targeting
If there's one metric to obsess over now, it's hook rate — the percentage of people who stop scrolling in the first two to three seconds. A high hook rate with a mediocre offer will usually still outperform a weak hook with a great offer, because the algorithm can't convert someone who never stopped scrolling long enough to see the offer at all.
That's the real shift: the skill that used to live in audience research has moved into the first three seconds of the video, or the first glance at the static. Accounts that haven't made that shift are still optimizing the lever that matters less, while the lever that matters more sits untested.
If your account has been running the same creative for more than a month, that's very likely where your ROAS problem actually lives — see Creative & Content for how we run a real testing cadence, not a quarterly refresh. Book a growth audit and we'll tell you honestly whether it's a targeting problem or a creative one.