How to Use Amazon Marketing Cloud: 2026 Performance Guide

how to use amazon marketing cloud to analyze campaign performance

TL;DR

Amazon Marketing Cloud (AMC) is a privacy-safe analytics environment that helps advertisers understand campaign performance beyond what standard Amazon Ads reports can show. It reveals how shoppers interact with multiple ad types before purchasing, which campaigns assist conversions, and which audiences are worth activating. This guide explains how to use Amazon Marketing Cloud to analyze campaign performance through a practical framework: start with a business question, pick the right analysis, interpret the output, take action, and validate the results.

What Is Amazon Marketing Cloud?

Amazon Marketing Cloud is Amazon Ads’ privacy-safe clean-room analytics environment. It lets advertisers analyze pseudonymized Amazon Ads signals (impressions, clicks, detail page views, purchases) alongside approved first-party inputs, then returns aggregated insights rather than individual shopper records.

In practical terms, AMC helps brands study how groups of shoppers interact with ads across the Amazon Ads ecosystem before they buy, return, subscribe, or drop off. It does not serve ads or manage bids. It is an analytics and audience-building layer that sits on top of campaign data.

This distinction matters because advertisers frequently confuse AMC with other tools. Here is what AMC is not:

Not a campaign manager. That is Ads Console.
Not a real-time reporting feed. That is Amazon Marketing Stream.
Not a media-buying platform. That is Amazon DSP.
Not a CRM. You cannot see individual shopper names, emails, or browsing histories.
Not a BI dashboard. It is a query environment that produces outputs you then interpret.

Understanding this is the first step toward knowing how to use Amazon Marketing Cloud to analyze campaign performance effectively. AMC answers questions that standard tools cannot, but it works best alongside those tools, not as a replacement.

If your Amazon ad campaigns already run across multiple formats, AMC can show how those formats work together. If you only run one small Sponsored Products campaign, the tool may not change many decisions yet.

Why Standard Amazon Ads Reports Are Not Enough

Standard Amazon Ads reporting is essential for daily campaign management. It shows bids, budgets, keyword performance, search term data, placement results, and campaign-level ACOS or ROAS. For most day-to-day optimization decisions, it works fine.

The problem is what it misses. Standard reporting uses a 14-day last-touch attribution approach, which means the last ad a shopper clicked before purchasing gets all the credit. Every other touchpoint in the journey is invisible.

This creates blind spots:

Standard reports show	AMC can reveal
Last-touch attributed sales	First-touch, equal-weight, and position-based attribution views
Campaign-level ACOS/ROAS	How campaigns work together across the full shopper journey
Search terms and targets	Which paths involve branded vs. non-branded searches
Clicks and orders	Detail page views, add-to-cart events, halo ASINs, new-to-brand patterns
Individual campaign performance	Cross-campaign and cross-media overlap effects

Consider a Sponsored Display campaign with poor last-click ROAS. Standard reports suggest cutting it. But AMC might show that shoppers exposed to that campaign frequently return through Sponsored Products four or five days later and convert. Practitioners on Reddit report exactly this scenario, with one commenter noting that AMC revealed a “worst-looking” campaign was actually assisting conversions days later, and likely would have been paused without the journey data.

The gap between what standard reports show and what AMC reveals is where budget decisions go wrong. If you want to understand how to analyze campaign performance in Amazon Marketing Cloud, the starting point is recognizing what your current reports cannot tell you.

How to Access AMC in 2026

Older guides often say AMC requires a DSP contract or substantial ad spend. That changed. Amazon announced in September 2025 that AMC became available to all advertisers running sponsored ads campaigns, including Sponsored Products, Sponsored Brands, Sponsored Display, and Sponsored TV. The AMC option appears under Measurement & Reporting in Ads Console.

Amazon also introduced no-code analysis templates for campaign performance and audience insights at the same time, along with AI-powered assistance. Users who want more depth can access the query editor and write SQL directly.

For SQL, Amazon provides an Instructional Query Library with prebuilt examples for common use cases like path to conversion, frequency analysis, and text filtering. And at CES 2025, Amazon announced a generative AI SQL generator for AMC that lets advertisers describe an audience or use case in natural language and receive SQL output.

The practical takeaway: you can start with no-code templates, but serious AMC analysis still needs someone who understands campaign structure, attribution logic, and how to validate what comes out. AI can write a query. It cannot decide whether the business question is the right one.

A note about agencies and DSP seats: if you work through an agency’s DSP seat, access considerations may apply. Practitioners on Reddit have flagged that AMC can sit at the seat level and see every advertiser in that seat, which creates questions about data visibility. The right question to ask any partner is: “Will we have visibility into the AMC instance, query outputs, and audience logic, or only screenshots in a deck?”

The AMC Performance Loop: A Practical Framework

Most guides list what AMC can do. Fewer explain how to turn AMC output into campaign decisions. The framework below, the AMC Performance Loop, organizes the process into six steps.

Step 1: Start With a Business Question

Do not open AMC and run reports. Start with a decision you need to make. Good questions include:

Should we cut this low-ROAS Sponsored Display campaign?
Are Sponsored Brands assisting Sponsored Products conversions?
Are branded campaigns inflating ROAS by capturing existing demand?
Which campaigns bring in new-to-brand customers?
Are we over-serving ads to the same shoppers?
Which ASINs create repeat buyers?
Is upper-funnel media supporting branded search lift?

Step 2: Choose the Right Analysis

Match the question to the right AMC analysis:

Business question	AMC analysis	Key metrics
Which touchpoints happen before purchase?	Path to conversion	Path count, assisted purchases, time to conversion, NTB share
Are two campaign types stronger together?	Media overlap	Conversion rate by exposure group, unique vs. overlap reach
Are upper-funnel ads assisting sales?	Custom attribution	First-touch credit, equal-weight credit, assisted conversions
Are we over-serving ads?	Reach and frequency	Frequency buckets, CVR by frequency, cost per purchase at each bucket
Are branded campaigns just capturing demand?	Branded vs. non-branded journey	NTB rate, non-branded-to-branded transitions, branded search lift
Who should we retarget?	Audience analysis	DPV no purchase, add-to-cart no purchase, exposed no purchase
Which ASINs deserve launch support?	New-to-brand / gateway ASIN	NTB orders, repeat rate, cross-sell behavior

Step 3: Run a Template, Instructional Query, or Custom SQL

New users should start with Amazon’s no-code templates, which now provide visualized insights for campaign performance and audience analysis. For deeper questions, adapt queries from the Instructional Query Library to your specific campaigns, date ranges, and attribution models.

Practitioners on Reddit who learned AMC recommend starting with the Instructional Query Library to see what results come back, then modifying queries for specific needs. The learning curve is real, but it is shorter than starting from a blank SQL editor.

One important prerequisite: your campaign naming and structure must be clean before running path analysis. If campaign names do not map to strategic roles (brand defense, category discovery, competitor conquesting, retargeting), the AMC output will be hard to interpret. Map campaign IDs to strategy groups before you start.

Step 4: Interpret by Campaign Role

Avoid generic “ROAS went up or down” interpretation. Instead, interpret relative to each campaign’s role:

If a campaign has poor last-click ROAS but high first-touch or assisted-conversion value, it is likely an awareness or consideration driver. Do not cut it without understanding its assist contribution.
If a campaign performs only when the shopper is already branded-searching, it may be demand capture rather than demand creation.
If conversion rate rises until four to six exposures then flattens, cap frequency or shift budget to incremental reach.
If Sponsored Products single-touch paths dominate, upper-funnel spend needs stricter justification.

That last point deserves emphasis. A practitioner in a ThinkNectar webinar observed that for one real brand, the top path was simply Sponsored Products single-touch, driving far more revenue than multi-touch paths. AMC does not always prove your funnel is complex. Sometimes it proves the opposite.

Step 5: Take a Campaign Action

Analysis without action is a vanity exercise. After interpreting AMC output, possible actions include:

Reallocate budget between DSP, Sponsored Brands, Sponsored Display, and Sponsored Products
Build AMC audiences for retargeting or bid boosts
Exclude purchasers or overexposed users
Separate branded and non-branded campaign budgets
Promote gateway ASINs that attract new-to-brand buyers
Change retargeting windows based on actual time-to-conversion data
Build cross-sell campaigns from product-pair behavior

Step 6: Validate After the Change

AMC improves attribution analysis, but it is not a causal experiment. After making changes, validate through before/after cohort comparison, Amazon Marketing Stream for near-real-time monitoring, Ads Console for bid and search-term optimization, and contribution margin analysis to verify business-level impact.

A seller on Reddit who ran a switchback experiment on a six-figure Amazon account found that only 53.6% of ad-attributed sales were truly incremental. The rest would have happened organically. This is one user’s test, not a universal benchmark, but it illustrates why attribution should not be confused with proven incrementality.

Best AMC Analyses for Campaign Performance

Path to Conversion

Path-to-conversion analysis shows the sequence of ad touchpoints shoppers experienced before converting. A shopper might see a DSP ad, later click a Sponsored Brands ad, then buy after clicking a Sponsored Products ad. Standard last-click reporting credits only that final Sponsored Products click. AMC can show the full chain.

This is one of the most common ways to use Amazon Marketing Cloud to analyze campaign performance because it directly answers “which campaigns should get credit?”

What to do with the output: Do not pause a weak last-click campaign until checking its assist role. Separate opener campaigns (high first-touch contribution) from closer campaigns (high last-touch contribution). Adjust budgets by path contribution, not just attributed ROAS.

Media Mix and Overlap

Media overlap analysis compares performance for shoppers exposed to different combinations of ad types. In an Amazon Ads case study for ghd, EY Fabernovel used AMC to compare audiences exposed to DSP only, Sponsored Products only, and both. The conversion rate for audiences exposed to both was 50% higher than the Sponsored Products-only group, and 86% of conversion paths started with DSP.

These results are from a single advertiser and are not universal benchmarks. But they illustrate the kind of question AMC can answer. If your brand runs both DSP and sponsored ads, overlap analysis can reveal whether those channels reinforce each other or just duplicate reach.

What to do with the output: If overlap improves conversion rate, coordinate DSP and Sponsored Products rather than managing them as silos. If overlap is high but performance does not improve, reduce duplicate reach or adjust audience exclusions.

Custom Attribution

Custom attribution lets advertisers compare how credit is assigned across touchpoints. Instead of relying only on last-touch, AMC supports first-touch, last-touch, equal-weight, and position-based attribution models with customizable weights.

Amazon’s own documentation notes that last-touch attribution tends to favor lower-funnel media. This means upper-funnel campaigns may look worse than they actually are when judged by standard reports alone.

What to do with the output: Use last-touch to understand closing efficiency. Use first-touch to identify demand creators. Compare multiple models before cutting upper-funnel spend. If a campaign looks strong in first-touch but weak in last-touch, it may be a valuable opener that other campaigns are closing.

Reach and Frequency

Reach and frequency analysis shows how many unique users saw ads and how often. This is where brands uncover wasted spend. A campaign may keep serving impressions to the same shoppers long after conversion probability has flattened.

What to do with the output: Find the frequency bucket where conversion rate peaks. Reduce spend where frequency increases but conversion does not. Reallocate impressions toward incremental reach. Use exclusions for recent purchasers. Adjust DSP frequency caps or retargeting windows. For more on eliminating waste, see this guide on reducing wasted ad spend.

New-to-Brand Analysis

New-to-brand (NTB) analysis identifies which campaigns, ad types, ASINs, or paths attract customers who have not previously purchased from the brand. This matters because a high-ROAS branded campaign that only captures existing customers is not growing the business.

Amazon’s Poppi case study shows what NTB analysis can drive: after AMC-informed strategy changes, Poppi observed a 16x increase in new-to-brand customers, with 34% of Q1 2022 ad-attributed sales coming from NTB buyers and a 54% increase in Subscribe & Save count. These results are specific to Poppi’s campaigns and should not be treated as a general benchmark.

What to do with the output: Scale campaigns with high NTB quality even if short-term ROAS is lower. Identify gateway ASINs that bring in new buyers. Pair NTB insights with repeat-purchase and margin data to find ASINs that attract customers who actually come back.

Repeat Purchase and Lifecycle

AMC can reveal which campaigns, audiences, or first-purchase ASINs lead to second purchases, subscriptions, or cross-sells. For consumable or replenishment categories, this is where long-term value lives.

An agency-published case from Incrementum Digital describes a health and wellness CPG example where AMC segmented first-time buyers, created reorder-window audiences, and measured results. The claim is that repeat purchases increased 20% within six weeks without increasing total ad spend. Treat this as directional, not a guaranteed outcome.

What to do with the output: Build reorder-window audiences. Cross-sell complementary products. Compare first-purchase ASINs by repeat rate, not only first-order ROAS. A lower-margin ASIN that creates loyal repeat buyers may deserve more budget than a high-margin product that generates one-time purchases.

AMC Audiences

AMC is not just reporting. It lets advertisers build custom audiences and activate them in Amazon DSP, Sponsored Products, Sponsored Display, and Sponsored Brands. This is where analysis becomes action.

Examples of performance audiences:

Detail page viewers who did not purchase
Add-to-cart users who did not convert
Shoppers exposed to upper-funnel media but not yet converted
First-time purchasers who have not reordered within the replenishment window
Buyers of ASIN A who are candidates for ASIN B cross-sell
High-frequency purchasers

One constraint: rule-based AMC audiences require a minimum of 2,000 records, with Amazon recommending 2,500 or more to reduce failure risk. Lookalike audiences require a minimum of 500 and a maximum of 500,000. Small accounts may struggle to meet these thresholds.

A practitioner on LinkedIn proposed a staged approach to AMC audience building: start with practical audiences (add-to-cart but no purchase, detail-page viewers but no purchase), move into path-to-conversion and cross-sell analyses, then progress to multi-touch attribution, lifetime value, and first-party data integration. That crawl-walk-run model keeps teams from overcomplicating things too early.

Not sure whether your campaigns are structured to get value from AMC audiences or any advanced analysis?

Get a free Amazon brand audit to find out where your campaign architecture, reporting, and catalog stand today.

Amazon Marketing Cloud Glossary

AMC is full of terms that confuse even experienced advertisers. Here are the most important ones, defined in plain language with notes on why each matters.

Clean room. A privacy-safe environment where data can be analyzed without exposing individual identities. AMC is a clean room because advertisers never see individual shopper records, only aggregated outputs.

Event-level data. Individual ad and conversion events like impressions, clicks, detail page views, add-to-cart actions, and purchases. AMC uses these internally for analysis but does not expose them as individual-level output.

Pseudonymized user ID. An identifier used inside AMC to count unique users, measure frequency, and analyze journeys. It cannot be selected in the final output of analytics queries because of privacy rules, but it powers the analysis behind the scenes.

Aggregation threshold. A privacy rule that suppresses or nulls outputs that are too granular to meet minimum user counts. If your AMC report shows nulls or missing rows, it usually means the segment is too small, the date range is too narrow, or the query exposes dimensions with high threshold requirements.

Instructional Query Library. Amazon’s collection of prebuilt SQL examples for common AMC use cases. Good starting point for advertisers learning how to use Amazon Marketing Cloud to analyze campaign performance without writing SQL from scratch.

Path to conversion. The sequence of ad touchpoints a shopper experienced before purchasing.

Media overlap. The audience exposed to more than one ad type or campaign group. Overlap analysis reveals whether combined exposure improves outcomes.

Custom attribution. Changing how credit is assigned across touchpoints. Supports first-touch, last-touch, equal-weight, and position-based models.

Lookback window. How far back in time AMC examines ad touchpoints. Amazon expanded the ad-traffic lookback window from 13 months to 25 months in supported markets as of November 2025, enabling year-over-year seasonal analysis.

New-to-brand (NTB). A customer who has not purchased from the brand within Amazon’s applicable NTB window. Important for separating acquisition from retention.

Halo ASIN. A product that receives attributed sales even though a different product was promoted. AMC can reveal which promoted ASINs drive halo sales across the catalog.

AMC audience. A custom audience built from AMC logic and activated in Amazon Ads campaigns. Can be rule-based or lookalike.

Amazon Marketing Stream. Amazon Ads’ API reporting tool for hourly campaign updates. It is not the same as AMC. Marketing Stream is for near-real-time monitoring; AMC is for historical, cross-campaign deep dives.

For more foundational terminology, the Amazon advertising glossary covers the broader ecosystem of ad types, metrics, and campaign concepts.

AMC vs. Amazon Marketing Stream vs. Ads Console

Most searchers need this comparison. These three tools are complementary, not competing.

Tool	Best for	Timing	Use it when…
Ads Console	Campaign, keyword, search term, placement, budget, ACOS/ROAS management	Daily / ongoing	You need to optimize bids, budgets, targets, negatives, and campaign settings
Amazon Marketing Stream	Near-real-time performance monitoring through API	Hourly	You need intraday pacing, budget alerts, hourly CTR/CVR changes, or launch-day monitoring
Amazon Marketing Cloud	Historical clean-room analytics, journeys, attribution, overlap, audience creation	Periodic deep dives	You need to understand paths, assists, audience overlap, NTB, repeat behavior, and audience activation

If you have noticed discrepancies between what your ad platform reports and your actual backend orders, that is a related but separate issue. This guide on ad platform vs. backend order discrepancies covers common causes.

Common Mistakes When Using AMC

Running Reports Without Decisions

AMC should answer a business question. Running every available template to create another dashboard wastes time. Start with a decision, then find the analysis that informs it.

Treating Last-Click ROAS as the Whole Truth

Amazon’s own custom-attribution documentation notes that last-touch attribution favors lower-funnel media. Cutting a campaign because it has poor last-click ROAS, without checking its assisted-conversion value, can remove a campaign that was quietly supporting other campaigns’ results.

Confusing Attribution With Incrementality

AMC improves attribution and journey understanding. It does not automatically prove that ads caused sales that would not have happened otherwise. Incrementality still requires experiments, holdouts, or other causal methods. The Reddit seller whose switchback experiment found only 53.6% of attributed sales were truly incremental is a useful reminder: platform-attributed credit and actual causal impact are different things.

Ignoring Privacy Thresholds

Small segments, narrow date windows, or overly granular dimensions may return nulls or be suppressed entirely. This is not a bug. It is AMC’s privacy design working as intended. Widen date ranges, reduce granularity, or aggregate dimensions to get usable outputs.

Using Messy Campaign Taxonomy

If campaign names are random strings of keywords and dates, path-to-conversion analysis becomes unreadable. Before running AMC analysis, map campaigns by funnel role: brand defense, category discovery, competitor conquesting, retargeting, launch support. Include ASIN metadata like category, margin, and replenishment cycle.

Building Audiences Too Small to Activate

Rule-based AMC audiences need at least 2,000 records, with 2,500 or more recommended. Building a hyper-specific audience of 800 people will fail. Check audience size estimates before submission.

Scaling Without Checking Margin and Inventory

A campaign path can look strong in AMC but still be bad business if the ASIN has weak contribution margin, low inventory depth, or stockout risk. Before scaling traffic based on AMC findings, verify inventory readiness and restock schedules.

When Should Your Brand Use AMC?

Good Fit

Use AMC when:

You run multiple Amazon ad types (Sponsored Products, Sponsored Brands, Sponsored Display, DSP)
You want to understand upper-funnel contribution
You need to separate demand creation from demand capture
You care about new-to-brand growth, not only immediate ROAS
You need repeat-purchase or subscription analysis
You want to create behavior-based audiences
Standard reports are causing unclear budget decisions
You have enough traffic and conversions for meaningful aggregated analysis

Not Yet

Do not prioritize AMC when:

You run one small Sponsored Products campaign
You do not have enough conversions for stable outputs
Your campaign naming and structure are messy
You need daily bid optimization, not journey analysis
You are not prepared to act on the findings

Small Seller Reality Check

Access does not equal usefulness. A practitioner on Reddit suggested that AMC becomes especially valuable for brands spending more than $30K per month. That is not an official threshold, and smaller brands can still benefit from basic templates. But AMC’s power scales with data volume, campaign complexity, and the team’s ability to turn findings into actions.

Connecting AMC Insights to Profitable Amazon Growth

AMC is most useful when its insights connect to campaign architecture, TACOS, contribution margin, inventory readiness, and weekly optimization governance. A path-to-conversion report is only valuable if the team knows what to change after reading it.

That connection between analysis and execution is where many brands stall. The analysis exists, but nobody changes the campaigns, adjusts the audiences, or validates the results.

Talk to EZCommerce about Amazon campaign performance to see how AMC insights can connect to structured campaign architecture, margin-aware optimization, and weekly governance.

FAQs

Do you need SQL to use Amazon Marketing Cloud?

Not always. Amazon now offers no-code templates and AI-powered query assistance for sponsored ads advertisers. The Instructional Query Library provides prebuilt SQL examples you can modify. But custom campaign-performance analysis still benefits from SQL knowledge or a partner who can adapt queries to your campaign structure, attribution logic, and ASIN taxonomy.

Is Amazon Marketing Cloud free?

AMC access is available through Amazon Ads for qualifying advertisers without a separate subscription fee. The real cost is the time, expertise, data setup, and governance needed to turn AMC outputs into decisions. Some advanced features, partner tools, or managed workflows may involve additional costs depending on setup.

Can small Amazon sellers use AMC?

Yes. Sponsored ads access is broader than it used to be. But small accounts may not have enough data volume for meaningful outputs. Privacy thresholds and minimum audience sizes can limit what smaller advertisers can analyze or activate. AMC becomes more powerful once you have enough traffic across multiple campaign types to compare journeys and audiences.

Is AMC better than Amazon Ads Console?

It is not better. It answers different questions. Ads Console is better for day-to-day campaign management: bids, budgets, search terms, placements. AMC is better for historical, cross-campaign, audience, and attribution analysis. Amazon Marketing Stream is better for hourly reporting and intraday optimization. Use all three for different purposes.

Does AMC prove incremental sales?

Not by itself. AMC can improve attribution and journey understanding, but incrementality still requires experiments, holdouts, switchback tests, or geo tests. Attribution analysis tells you which campaigns touched a sale. Incrementality testing tells you whether those sales would have happened without ads.

What campaign-performance metrics should you inspect in AMC?

Useful metrics include reach, frequency, impressions, clicks, detail page views, add-to-cart events, purchases, sales, units, cost per DPV, add-to-cart rate, purchase rate, ROAS, ACOS, new-to-brand share, repeat-purchase behavior, promoted vs. halo ASIN sales, and conversion rate by exposure path. The right metrics depend on the business question you started with.

How often should you analyze AMC?

Use AMC for periodic deep dives rather than every bid change. A practical cadence is monthly or quarterly for strategic analysis, plus event-based reviews after product launches, Prime Day, holiday campaigns, major DSP flights, or large budget shifts. Use Ads Console and Marketing Stream for daily and weekly optimization.

What should you do before running your first AMC analysis?

Clean your campaign naming. Map campaigns to strategic roles (brand defense, category, competitor, discovery, retargeting). Organize ASIN metadata by category, margin, and lifecycle stage. Decide on a clear business question. Without clean structure and a clear question, AMC outputs will be difficult to interpret and even harder to act on.