Something is broken in how brands measure creative performance on digital media platforms, and most marketing leaders have not noticed yet.
Over the last several years, every major advertising platform has undergone the same structural shift. Meta’s Andromeda made it obvious. Google’s Performance Max accelerated it. TikTok, Amazon, retail media networks, and emerging AI-native platforms are moving in the same direction where targeting, bidding, and placement optimization are increasingly automated and creative selection is algorithmic.
Across platforms, the machine now decides who sees your ad, when they see it, and how often. The primary lever brands still control at scale is the creative inputs fed into those systems.
The gap nobody is addressing is that the metrics most brands use to evaluate creative were built for a completely different era, focused on audience targeting and not algorithmic creative selection.
Using the wrong measurement framework does not just produce incomplete data. It leads to flawed decisions. Teams kill creative that should scale, protect creative that should be retired, and under-invest in the engine that now drives growth.
It is time to focus on creative learning velocity.
The Metrics We Inherited and Why They’re Failing Us
ROAS, CTR, CPC, and frequency caps were designed to answer one question.
“Is this ad reaching the right people efficiently?”
That was the right question when targeting was the lever. When marketers manually built audiences, layered interests, excluded segments, and adjusted bids by cohort, performance metrics were diagnostic tools for media precision.
In an AI-driven world, that level of precision is largely automated, with platforms now handling much of the targeting and optimization work that marketers once managed manually. As a result, the more important question has shifted.
Instead of asking whether an ad is reaching the right people efficiently, we should be asking whether the creative itself is giving the algorithm enough meaningful signal to discover and expand into new pockets of demand.
ROAS alone cannot answer that.
A high ROAS ad might be optimized for a narrow, already saturated segment with little room to scale. A lower ROAS ad might be exploring new territory, generating broader behavioral signals, and laying the groundwork for efficiency gains the system has not unlocked yet.
Traditional measurement frameworks treat the first as a winner and the second as a failure. In an AI environment, that logic can cap growth.
What You Actually Need to Measure: Creative Learning Velocity
The brands outperforming in today’s algorithmic landscape are not just producing more creative. They’re generating signals faster, iterating on winners more quickly, and systematically expanding their creative territory over time.
That’s creative learning velocity. And it’s a fundamentally different thing to measure.
It has three components:
- Signal generation rate: How quickly is your creative portfolio producing actionable data? This isn’t just about volume, it’s about meaningful variation. Ads that look and feel the same collapse into the same delivery pool and generate redundant signals. True signal generation requires creative diversity across hooks, formats, visual styles, emotional motivators, and product angles.
- Iteration cycle speed: How fast can you move from a learning to an action? The brands with structural advantages aren’t necessarily making better creative at the start. They’re learning faster. They’re identifying a winning hook on Tuesday, producing three variations by Thursday, and in-market by the following week. Slow creative cycles are now a meaningful competitive disadvantage.
- Creative territory coverage: This is the least-discussed metric and arguably the most important. It asks: how much of the potential creative space for your brand have you actually explored? Most brands have deep coverage in one or two zones–their hero product, their core visual identity, their established messaging–and almost no coverage elsewhere. Andromeda rewards breadth. Brands with broad creative territory consistently outperform brands with deep but narrow portfolios.
A New Measurement Framework
Translating these concepts into operational practice requires rethinking how creative is evaluated at every level of the organization.
At the asset level, the question shifts from “did this ad perform?” to “what did this ad teach us?” Underperformers that generate clear, actionable signals have real value. Measurement should capture learning outcomes, not just conversion outcomes.
At the portfolio level, the question shifts from “what is our best ad?” to “how well does our creative portfolio cover the space?” A simple coverage audit, mapping your active creative across format, tone, message, and motivator dimensions, will immediately reveal where you’re over-indexed and where you have unexplored territory.
At the organizational level, the question shifts from “what is our ROAS?” to “how fast are we learning?” Learning velocity should be a KPI that leadership tracks alongside efficiency metrics.
The Compounding Advantage
The reason this shift matters so much is that creative learning compounds over time.
Brands that build fast iteration cycles and broad creative coverage today are not just winning in Q2. They are building a structural advantage that becomes harder to close over time. Their systems get smarter. Their signal libraries grow richer. Their teams move faster with each cycle.
Meanwhile, brands still optimizing creative through a targeting-era lens are not holding steady. They are losing ground as competitors accelerate. The gap does not stay flat. It widens.
Andromeda changed what drives growth on Meta. The brands that adapt how they measure and manage creative will be the ones positioned to capture that growth.
