More than double the monthly revenue in 60 days.

A 3.5x ROAS — and shrinking revenue.

When this DTC pet brand came to us, their previous agency was reporting a 3.5x blended return on ad spend across Meta and Google. On paper, it looked like a functioning paid media program.

But one number told a different story. In their first month running ads, revenue sat at $49,437. Six months later — still running ads, still paying an agency — it had dropped to $42,014. A 15% decline while actively investing in paid media.

How does a business with a 3.5x ROAS see a decline in monthly revenue after 6 months? The answer starts with understanding what that number was actually measuring — and what it wasn't.

Three problems hiding inside one metric.

When we audited the account, the inflated ROAS wasn't a fabrication — it was a metric that didn't mean what the agency was using it to mean. The attribution model had fundamental problems stacked on top of each other.

1. The platforms were counting revenue they didn't earn.

Without existing customer exclusions in place, a majority of the revenue being attributed to paid was coming from customers that already existed inside Shopify. The algorithm wasn't finding new buyers — it was converting the easiest ones. Turning paid acquisition spend into a retention vehicle, without anyone realising it.

To make matters worse, this brand wasn't running ads in a vacuum. They had active distributor relationships, wholesale accounts, retail store placements, trade show revenue, and direct B2B sales. When someone bought because of a rep at a trade show or picked it up off a shelf, tried it, liked it, and then bought again online — Meta or Google claimed that order.

That customer isn't a new customer the ad created. They already knew the product. The ad just happened to remind them. You can't scale these campaigns. There was no way to distinguish between a sale the campaign generated and a sale that was going to happen regardless. The CAC being reported was essentially impossible to trust — even being generous and agreeing with platform attribution, counting offline-discovered customers as paid acquisition wins, it was twice as high as the brand thought.

2. Both platforms were claiming the same orders.

Meta and Google don't coordinate. This means any customer who touched both channels could be counted twice, once per platform. The 3.5x blended ROAS was built on double-counted revenue sitting across two separate dashboards with no deduplication. One of our first decisions was consolidating spend entirely onto Meta, eliminating the cross-platform attribution problem and giving one platform algorithm a single, clean signal to optimize against.

When you strip the double-counting out and look at actual new customer revenue against real ad spend, the number collapses. Even under the most generous interpretation — full platform attribution, no incrementality adjustments — the true new customer ROAS was 1.02x. Which meant the business was losing money on every new customer it brought in.

3. Branded search was capturing demand, not creating it.

A significant portion of Google spend was going toward branded search campaigns — ads that showed up when someone was already typing the brand name. These campaigns reported strong ROAS numbers, which pulled the blended average up. But high conversion rate on branded search doesn't mean the ad caused the conversion. Someone searching your brand name was already going to buy.

4. Bottom of funnel ads were narrowing their reach.

The creative strategy leaned heavily on basic catalogue ads — dynamic product ads retargeting site visitors and past customers. These may drive repeat purchases, but they do nothing for new customer acquisition. With no existing customer exclusions and the account structured around catalogue ads, the algorithm spent most of its budget optimizing for the easiest conversions — people who already knew the brand — rather than finding net-new buyers.

Low reported CAC on paper. Shrinking customer base in reality. You can't grow a business by only selling to people who already know you.

Rebuilding the system around real growth.

Before touching creative or budget, we changed the metrics we were managing to. If your north star is disconnected from your P&L, every optimization compounds the problem. The first step was establishing a measurement framework tied directly to business outcomes.

MER gives you a clean, platform-agnostic view of marketing efficiency — revenue divided by spend, no attribution games, no double counting. Acquisition MER narrows that to what actually matters for growth: are we profitably bringing in people who have never bought before? These metrics connect directly to the P&L and can't be inflated by branded search or cross-platform duplication.

From there, we rebuilt the account. Existing customer exclusions went in immediately. Catalogue ads were pulled back and replaced with prospecting creative built to reach cold audiences — people who had never heard of this brand. We built a proper acquisition funnel: creative that speaks to someone with no prior awareness, not someone who browsed the site last week.

You can't scale a system you can't trust.

The previous agency wasn't just running bad ads. They were running the wrong system — one that reported strong numbers back to the client while the business quietly declined. When your north star doesn't connect to your P&L, every optimization compounds the problem.

Paid media has one job: profitable new customer acquisition. Not retention. Not reactivation. Not taking credit for revenue that was already going to happen through other channels. The goal is to find people who don't know you exist, show them why they should care, and bring them in at a cost that makes the business grow.

That's the system we built. It's the only system worth building.