How a $2M/Month Ecommerce Brand Lifted ROAS 54% and Cut CPMs 28% With Hyper

This brand was spending about $2 million a month on Meta ads. At that scale, the numbers below matter in a way they wouldn't for a small account. A 54% lift in ROAS might not sound dramatic on its own, but on $2 million a month in spend, it's an enormous swing in absolute dollars. They shared their figures with us for this case study; everything else here is anonymized.

The problem was one most large DTC operators will recognize. CPMs kept climbing, Advantage+ had taken away the levers they used to pull, and the one thing that still moved the needle, fresh creative, was the thing they could never make enough of. Every winning ad burned out in a couple of weeks, and the team couldn't produce replacements fast enough to stay ahead of it. At $2M a month, falling behind on creative isn't a minor inefficiency. It's six figures of wasted spend.

So they handed their Meta and Google ads to Hyper and let it run.

What changed

The fix wasn't a clever setting. It was throughput and judgment, applied without stopping.

Constant cross-platform creative intelligence. This is the part people underestimate. Hyper wasn't just scraping the Meta Ad Library. The agents scanned the whole internet continuously: Instagram and TikTok for what was trending in the category, Reddit for what the brand's customers were actually talking about, and the Meta Ad Library for the competitor ads running right now. Then they turned that into briefs: the references that were working, the competitors worth studying, and start frames and concepts the team could generate from. Fresh, on-trend creative, sourced from everywhere people actually were, not just one ad library.

A testing loop that never stopped. The agents kept variations flowing and swapped in fresh creative before performance dipped, testing roughly 20x faster than the team had managed. More tests, found faster, meant more winners live at any given time.

Segmentation against the black box. With Advantage+ deciding most of the targeting, the agents segmented customers and shaped the signals to keep spend off low-intent traffic. On a $2M monthly budget, trimming that waste is where a big chunk of the 28% CPM reduction came from.

Execution, not recommendations. None of it sat in a report waiting to be actioned. The agent made the calls across Meta and Google and executed them.

Inside the creative engine

The engine ran continuously. The agents scanned Instagram and TikTok for trends in the brand's category, read the brand's subreddit for what customers actually complained about and asked for, and watched the Meta Ad Library for competitor moves. Each signal became a brief: the trend, the angle, and start frames the team could generate from. New creative launched into broad targeting in small isolated tests, so the agent could tell which specific ad moved the result. Fatiguing ads were swapped before they dragged the account down. The brand went from making a handful of new ads a week to never running out.

The numbers

Over twelve months:

• ROAS came up about 54% versus the brand's prior baseline.
• CPMs came down about 28%.
• Creative was tested roughly 20x faster.
• On $2M/month in Meta spend, that combination is a major swing in revenue and a large reduction in wasted impressions, month after month.

The mechanism was simple in hindsight. More tests meant more winners. Better cross-platform briefs meant stronger creative. Better segmentation meant less wasted spend. The brand stopped losing the race against creative fatigue because it was no longer running that race by hand.

An unexpected finding

The best-performing creative angle of the quarter didn't come from the Meta Ad Library. It came from Reddit. The agents caught a recurring complaint in the brand's subreddit, customers frustrated by a problem the product happened to solve, weeks before any competitor turned it into an ad. The creative built around that exact phrasing became the brand's top performer. Watching where customers actually talk, not just where competitors advertise, is the part most tools miss.

This is one of the accounts behind the numbers in how we built Hyper, the best AI for Meta and Google ads. For the head-to-head version, where an agency ran half its accounts on Hyper and half by hand, see the agency benchmark, and for the operational side, the original case study.