I recently wrote an AMC query that exposed some really cool details around DSP A
By Dustin Wassner · June 24, 2026 · Curated by George's Blog
I recently wrote an AMC query that exposed some really cool details around DSP Audience Impression Share and Attributed Sales, and like always, I want to share it with you 😁
I call it the segment index - If a segment is 5% of your impressions and 5% of your sales, it indexes at 1.0, then it's pulling its weight. Above 1.0, it's pulling its weight. Below 1.0, you're overspending.
Here's what it found on one of our accounts:
- The single biggest audience, 46% of all Impressions, drove only 22% of Ad Sales. It indexes at 0.49. We were allocating half the budget into our least efficient segment.
- Meanwhile a tiny retargeting pocket, 1.3% of Impressions, drove 9.2% of Ad Sales.
A 7.3x index and a 36x ROAS, quietly carrying the account.
Approaching it like this helps us go from "I think this audience is working" to "here's exactly how hard each one is pulling, ranked".
Why this matters: yes, the console's audience reports hint at some of this: Impressions by segment, even Conversions if you dig.
But it can’t provide you with a normalized index across every segment, deduped at the user level, with the attribution model in your control.
Why AMC really shines with this: A single shopper sits in a dozen segments at once, so "sales by segment" is meaningless until you decide how to split the credit across all that overlap. That where AMC comes in.
I want to give this query away, but I’m going to try something a little different on this post - if it gets more than 60 likes, then I’ll share the query below in the comments.
The fun part (to me at least): I built and ran this entire query with the Claude workflow that I describe here: https://lnkd.in/epxrsZZP
For some of my other AI related content, check this out: https://lnkd.in/eqkuV2Yn
Happy to talk shop on AMC anytime - feel free to DM me. 🙂
#btrmedia #adtech #retailmedia #amazonads Amazon Ads