Amazon Advertising Attribution: How Sales Are Counted, What Gets Inflated, and What to Do About It

Published: February 05, 2019
Last updated: April 24, 2026

Table of Contents

• What Attribution Actually Means on Amazon
• Attribution Windows by Ad Type
• Click Attribution vs View Attribution
• Same-SKU vs Brand Halo: The ROAS Distortion You’re Not Tracking
• Five Attribution Pitfalls That Inflate Your Numbers
• How to Use Attribution Data to Optimize Campaigns
• Getting More From Amazon Marketing Cloud
• FAQ

---

Most sellers look at their ROAS and think they know which campaigns are working. They don’t. Amazon’s attribution system counts the same sale differently depending on which ad type generated it, which attribution window applies, and whether the shopper clicked or merely saw an ad. If you’re comparing Sponsored Products ROAS to Sponsored Brands ROAS without adjusting for these differences, your budget allocation is based on fiction.

What Attribution Actually Means on Amazon

Amazon uses a last-touch attribution model. When a shopper clicks on multiple ads before purchasing, only the most recent click gets credit for the sale. That’s it - no credit splitting, no multi-touch weighting in standard reports.

The attributed sale is the dollar value of products purchased within a defined window after that last click. The window length, what counts as a qualifying interaction, and which products get included all vary by ad type. Those three variables are where sellers get tripped up.

Attribution Windows by Ad Type

Not all ad types measure the same time horizon. Here’s the current breakdown as of early 2026:

Ad Type	Click Window	View Window	What Gets Counted
Sponsored Products	7 days (Seller Central) / 14 days (Vendor Central)	None	Advertised ASIN + other products from same seller
Sponsored Brands	14 days	ML-determined (shorter than 14 days)	Advertised product + all same-brand products, including 3P seller sales
Sponsored Display (CPC)	14 days	None	Same brand scope as SB
Sponsored Display (vCPM)	14 days	ML-determined (shorter than 14 days)	Same brand scope as SB
Amazon DSP (Store ads)	14 days	ML-determined (shorter than 14 days)	Configurable

The Seller Central vs Vendor Central difference for Sponsored Products is one most sellers miss entirely. A 3P seller running the same SP campaign as a 1P vendor sees fewer attributed sales simply because the window is half as long. For higher-priced products where purchase decisions take a week or more, an estimated 20-30% of SP conversions fall outside that 7-day window.

Click Attribution vs View Attribution

Click attribution is straightforward: shopper clicks your ad, buys within the window, sale gets counted. View attribution is where it gets murkier.

View-through attribution credits a sale when a shopper sees your ad without clicking, then purchases later. Before 2026, this used a blanket 14-day window - glimpse a display ad, buy anything from the brand within two weeks, and the ad got credit. The industry rightly criticized this as overcounting.

As of January 1, 2026, Amazon replaced this with a “shopping-signal enhanced last-touch” model. Machine learning now evaluates whether the ad view actually influenced the purchase by analyzing browsing behavior, category search patterns, and shopping journey stage. The effective window is shorter and dynamic.

This change only affects view-based campaigns: Sponsored Brands (vCPM), Sponsored Display (vCPM), and DSP store ads. Click attribution is completely unchanged. DSP ads served offsite still use the legacy 14-day view window.

The practical impact: If your SB or SD campaigns showed a sudden drop in attributed conversions in early 2026, that’s not a performance decline - it’s a measurement correction. Amazon now provides dual reporting: standard metrics use the new model, while “all views” metrics preserve the old 14-day window for comparison.

Same-SKU vs Brand Halo: The ROAS Distortion You’re Not Tracking

This is the single biggest source of misinterpretation in Amazon advertising metrics.

Sponsored Products counts only “same-SKU” sales - the advertised product plus other products sold by the same seller. A shopper clicks your SP ad for headphones and buys a phone case from a different seller of your brand? That case sale is not attributed to your SP campaign.

Sponsored Brands and Sponsored Display count brand halo sales - any purchase of any product with your brand name, from any seller. A shopper clicks your SB ad and buys your product from a reseller? That reseller’s sale inflates your attributed revenue, even though you earned nothing from it.

For a brand with 50 resellers, this distinction is enormous. Consider a scenario: you spend $1,000 on SB ads. Attributed sales show $10,000 - a 10x ROAS that looks phenomenal. But $4,000 of those sales went to third-party resellers. Your actual revenue from those ads was $6,000, making your true ROAS closer to 6x.

The fix: Track same-SKU metrics separately from total metrics on SB and SD campaigns. If you have significant 3P distribution, same-SKU ROAS is the number that reflects your actual return.

Managing attribution across multiple ad types and accounting for brand halo effects requires real-time data integration. Feedvisor’s advertising optimization connects pricing and advertising data to identify which campaigns drive profitable, incremental sales - not just inflated ROAS.

Five Attribution Pitfalls That Inflate Your Numbers

1. Double-counting across campaigns. If a shopper clicks both an SP ad and an SB ad before purchasing, the sale appears in both campaign reports. Sum your SP + SB + SD attributed sales and you’ll get a total that exceeds actual revenue by 20-40%. Never add attributed sales across ad types. Use TACoS (Total Ad Spend / Total Revenue) instead - it’s immune to double-counting.

2. Organic cannibalization. Branded keyword campaigns often show ROAS of 10-20x because the shoppers were already searching for your brand. Industry estimates suggest 25-60% of branded keyword ad clicks would have resulted in organic purchases anyway. If your TACoS rises as ad spend increases but total revenue stays flat, you’re paying for sales you’d have gotten free.

3. Attribution window mismatches. Comparing a 7-day SP campaign to a 14-day SB campaign is apples to oranges. The SB campaign inherently captures more conversions because it has twice the window. This doesn’t mean SB is more efficient - it means the measurement is more generous.

4. Post-January 2026 metric confusion. Standard metrics now use the shopping-signal enhanced model (shorter windows, fewer view-attributed conversions). Comparing 2026 standard metrics to pre-2026 data shows an artificial decline of roughly 15-30% in view-attributed conversions. Use “all views” metrics for historical comparison and standard metrics for forward-looking decisions.

5. Subscribe & Save and repeat purchases. If a customer clicks your ad and their Subscribe & Save order auto-renews within the attribution window, that renewal counts as an attributed sale. You didn’t drive that purchase - the subscription did.

How to Use Attribution Data to Optimize Campaigns

Stop optimizing individual campaign ACoS in isolation. The metric that matters at the business level is TACoS.

TACoS benchmarks: 10-15% for established brands with 30%+ gross margins, 15-25% during growth phases where customer acquisition justifies higher spend. Below 25% gross margin, even 10% TACoS starts eating into profitability - run the numbers on your own margins before treating these ranges as safe.

Wait for the full attribution window before making bid changes. For SP campaigns on Seller Central, that means at least 8 days. For SB and SD, at least 15 days. Amazon’s data can take up to 12 hours to update and adjusts for canceled orders within 72 hours. Optimizing on incomplete data leads to over-correction.

Decision framework for bid adjustments:

ACoS Trend	TACoS Trend	What It Means	Action
Stable or falling	Falling	Ads are driving incremental growth	Increase bids and budget
Rising	Rising	Inefficient spend	Reduce bids, add negative keywords
Low	Stable or rising	Likely organic cannibalization	Reduce branded keyword bids
Rising slightly	Falling	Scaling phase - acceptable	Maintain; monitor for inflection

New-to-brand metrics (available for SB, SD, and DSP - not SP) tell you whether your ads are acquiring new customers or re-engaging existing ones. A new-to-brand order rate below 20% on non-retargeting campaigns is a red flag for cannibalization. Category-targeting SB campaigns typically show 60-80% NTB rates; branded keyword campaigns often fall to 20-40%. Your customer acquisition cost (Ad Spend / NTB Orders) should be weighed against customer lifetime value - if NTB customers go on to become repeat buyers, a higher initial acquisition cost pays for itself.

Getting More From Amazon Marketing Cloud

If you’re spending over $50K/month across ad types, run the “Overlap by Ad Type” query in AMC first. When SP and SB share more than 40% of conversion paths, cut SB branded bids by roughly 20% and watch whether SP performance stabilizes before reallocating that budget. That single analysis pays for the time investment.

AMC provides what standard last-touch reports cannot: de-duplicated conversions, multi-touch attribution models, and cross-campaign overlap analysis. The key finding: true incremental ROAS is typically 30-50% lower than standard last-touch ROAS, because standard attribution includes sales that would have happened organically.

Since September 2025, no-code query templates are accessible directly in the Ads Console - no SQL required for overlap analysis, conversion path reports, or NTB journey analysis. For off-Amazon traffic, Amazon Attribution provides separate measurement that feeds into AMC for cross-channel visibility.

FAQ

How long is the attribution window for Amazon Sponsored Products?

Sponsored Products uses a 7-day click attribution window for Seller Central (3P) accounts and a 14-day window for Vendor Central (1P) accounts. There is no view-through attribution for SP - only clicks count.

Why did my Sponsored Brands ROAS drop in 2026?

Amazon implemented a shopping-signal enhanced attribution model on January 1, 2026, which shortened view-through attribution windows using machine learning. This reduced reported view-attributed conversions by an estimated 15-30%. Your actual sales likely didn’t change - the measurement became more conservative. Check the “all views” metrics to compare against the previous methodology.

Can the same sale be attributed to multiple Amazon ad campaigns?

Yes. If a shopper interacts with multiple ad types before purchasing, the sale can appear in each campaign’s reports. This means summing attributed sales across Sponsored Products, Sponsored Brands, and Sponsored Display will overstate your actual ad-driven revenue by 20-40%. Use TACoS (Total Ad Spend / Total Revenue) for accurate business-level measurement.

What is the difference between same-SKU and brand halo attribution?

Same-SKU attribution counts only sales of the advertised product (used by Sponsored Products). Brand halo attribution counts sales of any product with your brand name, including sales by third-party resellers (used by Sponsored Brands and Sponsored Display). SB and SD ROAS can appear significantly higher than SP ROAS for the same campaign effectiveness.

What is a good TACoS for Amazon sellers?

It depends on your margins. Established brands with 30%+ gross margins typically target 10-15% TACoS, while growth-phase brands may run at 15-25%. TACoS is more reliable than per-campaign ACoS because it includes organic sales and avoids double-counting. If your TACoS is rising while total revenue is flat, your ads are likely cannibalizing organic sales rather than driving incremental growth.