Auditing Amazon DSP Attribution to Find Your True MER

Why DSP and Sponsored Ads Double-Count Sales

The primary source of attribution inflation is the structural difference in how Amazon's advertising networks claim credit for a purchase.

Sponsored Products, Sponsored Brands, and Sponsored Display operate under one attribution umbrella. If a shopper clicks a Sponsored Brands ad, then clicks a Sponsored Products ad, and buys, Amazon de-duplicates the click. Only the final click (Sponsored Products) gets the credit.

DSP operates on an entirely separate ledger. It uses a 14-day lookback window that includes both Click-Through Attribution (CTA) and View-Through Attribution (VTA).

Consider this sequence of events:

[Day 1: DSP Impression (VTA)] ──> [Day 3: Sponsored Products Click (CTA)] ──> [Day 4: Purchase]

In your reporting, you see $100 in ad-attributed sales. In reality, you only collected $50 in revenue. If you evaluate each channel in a silo, you will over-allocate budget to DSP under the false impression that it is generating net-new demand.

The Formula: Calculating True MER

To protect your margins, you must look past individual campaign ROAS and focus on account-level efficiency. Your main metric for this is Marketing Efficiency Ratio (MER), sometimes referred to as Total ROAS.

Evaluating MER alongside TACoS (Total Advertising Cost of Sales) reveals the true health of your marketing engine:

How to Audit DSP Incrementality

To find out how much of your DSP revenue is actually incremental, you need to isolate the overlapping touchpoints. There are two primary methods to execute this audit.

Method A: The Amazon Marketing Cloud (AMC) Query (Recommended)

If your brand spends enough to access Amazon Marketing Cloud, you have access to event-level data. You can write SQL queries to de-duplicate your customer paths.

To run an incrementality audit in AMC, execute a query that groups conversions into three distinct buckets:

• Sponsored Ads Only: Conversions where the shopper only interacted with Sponsored Products, Brands, or Display.
• Overlap (Sponsored Ads + DSP): Conversions where the shopper interacted with both.
• DSP Only: Conversions where the shopper only interacted with DSP.

Analyze the Overlap bucket. Within those overlapping conversions, determine what percentage of DSP sales were driven by View-Through (VTA) versus Click-Through (CTA).

Method B: The Geo-Targeted Holdout Test

If you do not have AMC access, you can run a clean geo-targeted holdout test. This method uses geographic isolation to measure real-world incrementality.

Tactical Adjustments: Reining in DSP Overlap

Once you audit your attribution and find your true MER, you will likely need to adjust your DSP setup to stop wasting budget on non-incremental views.

1. Tighten the Frequency Cap

By default, DSP campaigns can over-expose audiences, showing the same banner 10 or 20 times a day to a user who already bought your product. Set a strict frequency cap—such as 3 impressions per user per 24 hours—to prevent system fatigue and wasted ad spend.

2. Exclude Past Buyers from Retargeting

If your product has a 90-day repurchase cycle, there is no reason to retarget a shopper 3 days after their purchase.

Build a custom ASIN exclusion audience inside DSP containing anyone who purchased your product within your specific consumption window. This prevents DSP from claiming cheap VTA credit on customers who are already in their organic reorder phase.

3. Shift Budget from Retargeting to Prospecting

Retargeting campaigns always boast the highest self-reported ROAS in the DSP console because they target warm leads. They also have the highest overlap with Sponsored Ads.

Make Decisions on Total Account Economics

Amazon DSP is a powerful tool for scaling market share, but its native reporting is highly promotional. Evaluating it in isolation is a fast track to margin erosion.

Every dollar spent on DSP must justify its existence by improving your overall MER or driving clear, verified New-to-Brand customer acquisition. Run an attribution audit quarterly. If your total revenue does not move when DSP spend shifts, it is time to adjust your targeting, tighten your frequency caps, or reallocate those dollars to channels where you can prove incrementality.