Clevrr Capsule #007: The 0.84x ROAS campaign you should NOT kill

Hey there, different email today. We shipped something BIG.

Introducing the Clevrr Pixel.

Until now, Clevrr AI worked with what the platforms gave us. We pulled official data from Meta, Google and Shopify, decomposed it, found the leaks, and told you why your numbers moved. That worked well and it still does.

But there was always a ceiling, and it wasn’t ours. It was the platforms’.

Meta sees Meta. Google sees Google. Shopify sees the last UTM and nothing before it. GA4 sees sessions but not spend. Every tool you already have looks at one slice of the journey and grades its own homework.

So we stopped asking them and started recording it ourselves.

What it does

The Clevrr Pixel records every session on your site and stitches it to the campaign, channel, product and order behind it. Not modelled attribution. Not an estimate. The actual path, visitor by visitor, from first touch to checkout.

Then everything Clevrr already did sits on top of that: RCA traces the cause, the agent prioritises what to fix, and monitoring watches it daily.

What it found in the first seven days

We ran it on a D2C shapewear brand’s account. 88,227 sessions, 1,775 orders, 27 Meta campaigns and 5 on Google, over one week.

Five budget decisions came out of it. Every one was the opposite of what Ads Manager said.

Their worst campaign was holding the account together. Biggest spender, 0.84x ROAS, the kind of campaign anyone would pause on Monday morning. It turned out to be the single most assisting campaign they had, sitting in 261 buyer paths. Pausing it would have quietly starved every higher-ROAS campaign downstream.

Their best campaign wasn’t creating anything. Brand search at 4.24x looked like the hero. But nobody searches for a brand they’ve never heard of. It was assisting in 51% of paths with an 8.8 day average lag. Those buyers met them on Meta first and came back through Google to finish. Google collected the demand. Meta created it.

Their biggest leak wasn’t in the ads. 59,558 people viewed a product and never carted it. 91% of everyone who looked. Lifting that one step from 8.8% to 10% adds roughly 780 carts a week at the same spend.

One order, four campaigns, 19 minutes. A single ₹1,780 order touched four different campaigns in under twenty minutes. The last one took 100% of the credit. Three others did the work of getting the shopper there, and no platform could see it.

Why this is the unlock

Rebuilding just the first of those findings by hand would mean exporting four platforms, matching UTMs to orders, reconstructing every buyer’s path and classifying every campaign’s role. Days of analyst time for one week of data, and the budget is spent before the answer arrives.

The brand had all of it seven days after installing.

That’s the shift. We were already good at telling you what happened and why. Now we can see the part no platform shows you: which campaigns genuinely create demand, which ones just collect it, and where buyers actually fall out.

Why now matters

Diwali is two weeks away.

In that account, average days to convert ran between 5.9 and 9.4 days, with a tail past 40. Campaigns that start discovery in the week before the peak are mostly paying for shoppers who buy after it’s over. The demand for your festive window has to be built now, and building it means knowing which of your campaigns build versus which collect.

Setup takes about ten minutes. The first real read comes at day seven.

Code CLEVRR3M gets you three months free, which covers the rest of the season.

Or book 20 minutes with me and I’ll walk through what it finds in your account, live.

Normal Capsule back next week.

Talk soon,
Yuvraj
Founder, Clevrr AI

P.S. If you want the full nine-page breakdown of that account, including the order-level journeys and the action plan that came out of it, it’s here. It’s the most detailed thing we’ve published.