We’ve watched it happen more than once: a store scaling ad spend on a dashboard ROAS that looks great, while the bank account quietly disagrees. Revenue up, profit down, founder confused. The dashboard wasn’t broken. It was just answering a different question than the one that matters.
Here’s why platform ROAS misleads, and the four numbers that should actually run your store. No finance degree required; we promise to use small words and one worked example.
Why the dashboard flatters you
Two reasons, both structural. First, ad platforms grade their own homework. Meta and Google each claim credit for conversions they touched, and a customer who saw your Instagram ad, googled you, and clicked a Search ad gets counted by both. Add the platforms’ reported revenue together and it routinely exceeds what your store actually took in. Agency audits, ours included, find this constantly.
Second, ROAS measures revenue, not money. A 3x ROAS sounds triumphant until you subtract the product cost, shipping, payment fees, packaging, and returns, at which point plenty of 3x campaigns are funding a very expensive hobby. Fashion brands with high return rates know this pain intimately: a returned order shows up in ROAS and leaves through the back door of the P&L.
The four numbers that don’t lie
1. Blended CAC. Total marketing spend, all channels, divided by total new customers. No platform gets to grade itself; the denominator is reality. If blended CAC is climbing while platform ROAS looks stable, believe the blended number.
2. Contribution margin. What one order actually leaves behind after product cost, shipping, transaction fees, and expected returns, before ad spend. This is the number that tells you what a customer is worth to acquire. Most founders can get to it in a spreadsheet in an afternoon, and the afternoon regularly changes the whole strategy.
3. LTV, honestly windowed. Lifetime value gets abused to justify spending: a hopeful “lifetime” makes any CAC look fine. Use a 60-day LTV for this quarter’s spending decisions and a 12-month LTV for planning, both measured from your own order history rather than optimism. The classic health benchmark is LTV at least 3x CAC.
4. Payback period. How many days until a new customer’s cumulative margin covers what you paid to get them. Under about three months and your cash recycles fast enough to grow; much past six and growth eats cash even while the dashboard celebrates.
The worked example
Store sells a $60 product. Platform reports 3x ROAS, so $20 of ads per order. Landed product cost $22, shipping $8, fees and packaging $4, returns average another $3. Contribution before ads: $23. After ads: $3 per order. Now the whole business turns on the second purchase, which arrives with no ad cost and drops nearly the full $23 to the bottom line. Same store, same ads, and the difference between struggling and thriving is entirely in whether customers come back. This is why we keep saying retention math and ad math are the same math.
What to do with this on Monday
Build the boring spreadsheet: the four numbers, updated monthly. Judge channels by blended results and incrementality (pause a channel for two weeks in one region and watch what actually changes; it’s crude and it’s revelatory). Set a CAC ceiling from your contribution margin rather than from what the auction charges. And treat platform ROAS as what it is: a directional signal for comparing creatives inside a platform, not a verdict on the business.
If you want a second pair of eyes on the math before scaling spend, that conversation is exactly where Clavis Ads engagements start, because ads built on honest numbers are the only kind we’re interested in running.