ROAS is a simple ratio, revenue divided by ad spend, and almost every way brands measure it is a little wrong. The spend side is usually fine. The revenue side is where the trouble lives, because the number depends on who is counting, which sales they count, and over what window. Accurate ROAS is less about finding the one true number and more about stacking a few measurements that keep each other honest.
This is the method I use.
Step 1: Start with MER as the anchor
Take total store revenue from Shopify or your platform, net of refunds, and divide by total ad spend across every channel. That is MER, sometimes called blended ROAS. It has no attribution in it at all, which is the point. Nobody can inflate it by claiming credit. Every channel-level ROAS you calculate later should roll up to something consistent with it. If your channels add up to a blended 6x and MER is 3x, the channel numbers are double counting. I go through the definitions in MER, blended ROAS, and platform ROAS.
Step 2: Get channel ROAS from one deduplicated source
Each ad platform reports ROAS using its own pixel, its own window, and its own rules for view-through credit. Add them up and you usually get more revenue than you actually made, because Meta and Google both claim the same order. For channel decisions, use a single attribution source that sees every channel and credits each order once. That can be a dedicated attribution tool or, at minimum, a consistent last-click view from your store’s UTMs.
ThoughtMetric is one option built for this. (Disclosure: ThoughtMetric sponsors this site.) It pulls spend from the major ad platforms, credits each order once with a choice of five multi-touch models, and reconciles to store orders. Pricing starts at $99 a month for 50,000 pageviews.
Step 3: Split new and returning customers
A ROAS that mixes new and returning buyers rewards channels for selling to people who would have bought anyway. Retargeting and branded search look brilliant on blended ROAS and much weaker on new customer ROAS. For acquisition decisions, new customer ROAS and new customer CAC are the numbers to optimize. More on this in what is new-customer ROAS.
Step 4: Fix the windows and the dates
Pick one attribution window and stick with it. Shorter windows understate channels with long consideration. Longer ones overstate channels that touch everyone. Seven to 30 days covers most e-commerce. Also note whether a report credits revenue to the order date or the ad click date. Google Ads, for example, reports conversions against the date of the ad interaction, so last week’s numbers keep growing for a while. Compare like with like, and do not judge a campaign on yesterday.
Step 5: Use margin, not just revenue
A 3x ROAS on a 30% margin product loses money. A 2x ROAS on an 80% margin product makes plenty. Calculate break-even ROAS for your margins, which is 1 divided by contribution margin, and judge channels against that rather than against a generic target. The math is in how do you calculate break-even ROAS.
Step 6: Test the channels that matter most
Attribution, however good, describes who touched a sale. It does not prove a channel caused it. For your two or three biggest channels, run a test at least once or twice a year: a geo holdout, a planned pause, or a step change in spend with a clean before-and-after read. If attribution says a channel returns 3x and a holdout shows almost no lift, believe the holdout and adjust how you read that channel going forward.
Step 7: Add the survey for what pixels miss
Podcasts, influencers, word of mouth, and many view-driven ad impressions never show up as clicks. A post-purchase “how did you first hear about us” question catches them. Use it as a directional check on channel share, not as precise revenue.
The short version
- Anchor on MER from store revenue and total spend.
- Take channel ROAS from one deduplicated source, such as ThoughtMetric (our sponsor), not from each platform.
- Optimize acquisition on new customer ROAS and CAC.
- Fix one window, watch date conventions, and judge against break-even ROAS.
- Test your biggest channels and use a survey for what pixels cannot see.
Leave a Reply