KPIHelm — Your KPIs. Our insight. Your future.
All articles

Your blended ROAS looks fine. One of your channels is losing money.

A blended ROAS is an average, and an average is the one number that cannot show a split. It takes everything you spent across every platform, divides it into everything those platforms say they earned, and returns a single figure that describes no channel you actually run. When one channel is strong enough, that figure sits comfortably above your break-even while another quietly burns cash underneath it. Nothing looks wrong — because the winning channel is paying for the losing one, and the average has no way to say so. This guide is about finding the channel the average is hiding. If your ROAS has visibly fallen, that is a different diagnosis: start with Your ROAS dropped on Shopify. If it looks healthy and the month still leaves no money, the cause may be margin rather than channels — that is Your ROAS looks healthy. So why is there no money at the end of the month?.

Step 1: What the average actually hides

Put numbers on it, because the size of the gap is the surprising part. A month with 10,000 of ad spend:

  • Meta: 6,000 spent, 24,000 in sales → 4.0x
  • TikTok: 4,000 spent, 4,800 in sales → 1.2x
  • Blended: 10,000 spent, 28,800 in sales → 2.88x

Now bring in what an order actually leaves you. At 40 kept in every 100 of sales, break-even sits at 2.5x — the arithmetic behind that number is in the break-even guide. Blended 2.88x clears it. The month is profitable. Everything reads green.

Read the two channels separately and the picture changes:

  • Meta brought back 9,600 in margin on 6,000 of spend — it made 3,600.
  • TikTok brought back 1,920 in margin on 4,000 of spend — it lost 2,080.

The store still finished ahead, by 1,520. But 2,080 was burned to get there, and the one number anyone looked at never moved. Worse, it moves the wrong way: a blended figure improves whenever the winning channel grows, whether or not the losing one was ever fixed. The month that looks like progress can be the month the leak got bigger and the winner grew faster than it.

Step 2: The sum underneath the average isn't real

There is a second problem, and it is the one that decides the whole question. Each platform reports the conversions it believes it caused, by its own rules: how long after a click it still claims credit, whether a view counts at all, what happens when someone sees an ad on one platform and clicks on another. Those rules differ between platforms, and they change over time.

The consequence for a blended number is arithmetic, not philosophy: adding platform-reported revenue together counts twice every order that more than one platform claims. Check it on your own store. Total what the platforms report for a period, then compare it with what Shopify recorded from those channels over the same period. If the platform total is higher — it usually is — the difference is credit claimed twice, and nothing in either dashboard tells you which platform earned it.

So a blended ratio is not merely uninformative. Its numerator is a number your store never received. It is an average of measurements taken with different instruments, some of them counting the same event twice, presented as a single truth.

The practical rule that follows: revenue is real where the money arrived, which is the store. Platform figures are useful for comparing a platform against its own history, and unusable for adding up. This is the same reason you cannot judge one platform's ROAS against another's, covered in step 4 of the ROAS-dropped guide — here it just has a bigger consequence, because blending doesn't compare the platforms, it merges them.

Step 3: Break-even belongs to the channel, not to the store

A store-wide break-even answers a store-wide question: did the month leave money. It cannot tell you whether a specific channel paid for itself, and a channel that doesn't is a loss even inside a profitable month.

It also isn't one number. Break-even depends on what an order leaves you, and channels rarely sell the same mix — if one channel moves cheap, thin-margin items while another sells the expensive ones, their break-even points genuinely differ. Applying a single store-wide figure to every channel is the same mistake one level down: an average standing in for a split.

One legitimate exception, and it only counts if it was decided in advance: buying the first order at a loss, because you know your repeat rate and the customer comes back. That is a strategy. Discovering it at month end, without that number, is a leak with a story attached.

Step 4: The ten-minute check that finds the channel

Do this per channel, not per campaign — campaign-level noise hides the pattern at this stage.

  1. List spend by channel for a closed period, long enough that returns have mostly landed.
  2. Take revenue from the store, split by channel, not from the ad platforms — for the reason in step 2.
  3. Multiply each channel's revenue by what an order leaves you, then subtract that channel's spend. A negative number is the channel the average was hiding.
  4. Compare each channel with its own previous period, never with its neighbour. A 1.2x channel may be normal for that platform and that audience, or may have been 3x last month — those are opposite situations and only its own history separates them.
  5. Check the spend split, not just the ratios. A blended figure can hold steady for months while spend quietly shifts from the channel that works to the one that doesn't. If the blend is flat and the composition moved, the blend is flattering you.

Step 5: Turn one channel down, then judge it by data

The reflex is to switch the losing channel off. Reduce it first instead, for a reason worth knowing: channels are not always independent. A channel that looks weak on last-click numbers can be introducing customers who convert somewhere else, and killing it outright can pull down a channel that looked healthy. Cutting spend in half and watching what happens to store-wide revenue — not just that channel's — tells you which of the two you have, and costs one period to find out.

Then do what any honest diagnosis ends with: change one thing, and judge the result by how much new data has accumulated rather than by how many days have passed. If the losing channel shrinks and the store total falls with it, it was doing work the attribution never showed. If the store total holds, you just recovered the money the average was hiding.

Next in this series: reading revenue per channel from the store rather than from the ad platforms raises the obvious question — why do the two disagree in the first place, and by how much. That is the subject of Meta says 40 sales. Shopify says 24. Neither one is lying..

KPIHelm never merges your platforms into one number. Connect your TikTok, Meta and Google Ads accounts to your Shopify store: each platform is read in its own context and against its own history, next to what your store actually recorded — so a winning channel cannot cover for a losing one, and the channel that needs attention is named instead of averaged away. Install it free on the Shopify App Store or see a sample report.