Stop Chasing One-Day-Click ROAS: Your Best Static Might Be Your Least Incremental Ad

"Just kill the videos and put everything behind the static. It's doing a 15 and the videos are barely breaking even."

A founder said that to me on a call recently, looking at his own dashboard, and on the numbers in front of him he was completely right. One static was reporting a 15-something ROAS. His video ads were sitting near 3. If those were the only two numbers you ever saw, you'd move the budget without a second thought.

We didn't. And the reason why is the most useful thing I can teach you about creative testing this year.

The ad that looked like a hero

Let me walk you through the exact shape of the problem, because it's sneaky.

Take one ad account, prospecting campaign, a spread of creative running. You open your columns and the headline return on ad spend on your best static reads about 15.77. Genuinely. That's the kind of number that makes you want to pour every dollar you have into it.

Now change one thing. Instead of looking at the standard 7-day-click, 1-day-view number, you look at the same ad through an incremental lens, the conversions the platform actually believes happened because of the ad rather than ones that would have happened anyway. On that same static, the number drops from 15.77 to 2.85.

Read that again. Same ad, same spend, same period. One way of counting says 15.77. The honest way of counting says 2.85.

So what happened to the other roughly 13 points of ROAS? They were never really yours. That static was getting the credit for sales that were going to land regardless. It was sliding in front of people who'd already decided, an impression a second before a purchase that was always coming. On a last-click-ish view it looks like a machine. On an incremental view it's barely pulling its weight.

This is the trap with one-day-click ROAS. It rewards the ad that shows up at the finish line, not the ad that got the runner moving. And your "best" creative, the one the dashboard keeps crowning, is very often the one doing the least actual work to bring you a new customer.

Why your statics keep winning the wrong contest

There's a reason this pattern shows up on statics so often.

A clean product static, the one with the offer and the hero shot, tends to convert the people who were already most of the way there. It's a closer. It catches the click from someone deep in consideration. So it racks up a gorgeous in-platform ROAS while mostly harvesting demand that already existed.

Your upper-funnel video, the one introducing the product to someone who'd never heard of you, looks worse on the same scoreboard. Longer path to purchase, more people who watch and don't buy today, a softer one-day-click number. But that video might be the only thing in the account actually creating new demand.

If you judge both on headline ROAS, you defund the ad that grows the business and double down on the ad that just collects the harvest. You'd feel like you were optimising. You'd be quietly shrinking.

The holdout that settled it

I'm not asking you to take my word for the videos-versus-statics thing, because someone ran the clean version of this test and the result is worth sitting with.

There's a well-known case where a brand ran two near-identical campaigns. Same targeting, same optimisation, same settings. The only difference was the creative: one campaign held upper-funnel video, the other held lower-funnel statics. Then, instead of trusting in-platform numbers, they ran a proper geo holdout, the kind where you withhold ads from whole regions and measure the lift in actual sales against regions that saw them.

On Facebook-reported ROAS, the statics looked like the obvious winner. You'd have cut the video.

On incrementality, the videos were dramatically more incremental than the statics. The statics had been taking credit; the videos had been doing the causing.

That's the whole argument in one experiment. The metric inside the platform and the truth about what drove sales pointed in opposite directions. If creative diversity decisions get made on the in-platform number, you will systematically pick the less incremental ad and feel good about it.

In-platform numbers can't validate creative diversity

Here's my actual take, and it's the thing I'd put on the wall.

You cannot validate a creative-mix decision with in-platform metrics. Full stop. Whether videos earn their place against statics, whether that wild new angle is worth keeping, whether you should spread budget across ten formats or pour it into one, none of that can be settled by ROAS columns, because those columns can't tell the difference between an ad that caused a sale and an ad that stood near one.

The only things that can answer it are holdouts. A geo holdout where regions with the creative get measured against regions without it. A conversion lift test if you've got the scale and the access. Those measure causality. Columns measure correlation, and correlation is exactly what's lying to you when a 15.77 turns into a 2.85.

This is why I think incrementality-led creative is becoming the real dividing line between accounts. Anyone can read a ROAS column. Knowing which of your ads actually move the business, and being willing to keep funding a "worse-looking" video because the holdout says it's the one creating demand, that's the harder and far more valuable skill.

A net-new-visit warning while we're here

One more place the dashboard fools you, because it's related.

You'll get an ad reporting something like 90% new visitors and think you've cracked top-of-funnel. Fresh audience, loads of people who've never seen you, brilliant.

Then you pull the rolling reach report and find the ad barely reached anyone new at all. It was earning first-time clicks from people who'd already been served your ads plenty of times. High net-new-visit rate, almost no net-new-reach. You hadn't found new people. You'd just finally got familiar people to tap through.

Which is fine, that's a real job, but it's not the job you thought you were measuring. Same lesson as the ROAS trap: the surface metric described one thing and the underlying reality was another. You don't always need to reach new people. Sometimes you need to reach the same people in a new way. You just have to know which one is actually happening before you make a budget call on it.

How I'd actually pick creative when the metrics lie

So if headline ROAS can't be trusted to crown your winners, how do you choose? Roughly how I'd run it:

• Use incremental attribution as your reading lens, not just your optimiser. When you do creative analysis, look at which ads drove the most incremental conversions, not the highest raw ROAS. The ad with the best incremental contribution is the one earning its keep.
• Sanity-check any big creative-mix call with a holdout. Before you axe a whole format because its columns look soft, run a geo holdout. Let it tell you whether that format is incremental or invisible. Don't let a 15.77 talk you into a decision a 2.85 would veto.
• Keep funding the demand-creators even when they look worse. Your upper-funnel video may always read uglier than your closing static. That's expected. If the lift test says it's incremental, it stays, because something has to bring new people in or the statics eventually run out of warm demand to harvest.
• Pair the formats on purpose. Videos to create demand, statics to convert it. They're not competing for one slot. They're doing different jobs, and the in-platform scoreboard just happens to flatter one of them.

The shift in mindset is small but it changes everything: stop asking "which ad has the best ROAS" and start asking "which ad is actually causing sales that wouldn't have happened otherwise." Those are different questions, and only the second one grows a business.

Try it on your own account

You don't need a project to see this for yourself. Open your best-performing ad, the one you'd protect with your life, and look at it through an incremental view next to its standard ROAS. If the two numbers are close, lovely, that ad is genuinely pulling. If the incremental number falls off a cliff the way that 15.77 fell to 2.85, you've just found out your hero is mostly a harvester.

Do that across your top five ads before your next creative review. Then ask yourself how many budget decisions you've made this year on the strength of a column that couldn't tell causation from coincidence. That's the exercise. It tends to change how you look at the whole account.