1 min read
Every Platform Grades Its Own Homework : Scott on the ROAS Trap
Jul 27, 2026 by Scott Desgrosseilliers
9 min read
The Paid Traffic Truth : The ROAS Trap Held a Second Week — and Got Wider
Jul 22, 2026 by Scott Desgrosseilliers
10 min read
The Paid Traffic Truth : Your best ROAS channel costs the most per new customer
Jul 14, 2026 by Scott Desgrosseilliers
8 min read
The Paid Traffic Truth : Meta's New Customer Conversion Just Jumped 66%
Jul 8, 2026 by Scott Desgrosseilliers
8 min read
The Paid Traffic Truth : The New Customer Markup Your Dashboard Hides
Jul 6, 2026 by Scott Desgrosseilliers
6 min read
Your Data Is Being Estimated — And Nobody Told You
Jun 15, 2026 by Scott Desgrosseilliers
6 min read
Three Attribution Shifts That Changed This Week — And What They Mean for Your Revenue
Jun 8, 2026 by Scott Desgrosseilliers
10 min read
How Meta, Google, and TikTok Changed Your Numbers Without Telling You
Jun 5, 2026 by Scott Desgrosseilliers
9 min read
How Link Necklaces Cut Meta nCAC 22% and Nearly Doubled New Customers
May 28, 2026 by Scott Desgrosseilliers
2 min read
The Monday Routine That Stops Wasted Ad Spend : The Five Forces Playbook
Apr 27, 2026 by Scott Desgrosseilliers
2 min read
How the Optimization Force Improves Campaign Performance
Apr 20, 2026 by Scott Desgrosseilliers
5 min read
What is a Good ROAS for Ecommerce? Benchmarks & ROI Tracking
By Scott Desgrosseilliers on May 10, 2022, 12:00:00 PM
WHAT IS A GOOD ROAS FOR ECOMMERCE IN 2026? THE HONEST ANSWER (AND WHY THE NUMBER LIES)
Everyone wants a single number - what's a good ROAS? Here's the honest answer up front — the commonly quoted benchmark is 3:1 to 4:1, but the actual 2026 ecommerce average has slid to around 2.87:1, and the median is closer to 2:1. That means roughly half of all ecommerce brands are running below a 2:1 return. So if you're benchmarking against "4:1," you're comparing yourself to a number most stores never hit.
But the bigger problem isn't the benchmark. It's that ROAS itself — the way most brands measure it — is being quietly mis-reported by the very platforms you're using to calculate it. Let's fix both.

