Fix your "now-what" problem with the ACE framework. Learn to turn Amazon Marketing Cloud data into actionable media moves—no SQL required.
Key takeaways
The short version: Amazon Marketing Cloud (AMC) shows you the whole path to purchase, but the data rarely turns into a decision. ACE fixes that. Sort any growth problem into one of three jobs: Amplify (more from current buyers), Capture (convert demand efficiently), Expand (reach new buyers), and each job points to the AMC read to run and the move to make. No SQL required.
Most Amazon advertisers don't have a data problem. They have a now-what problem.
AMC shows how your ad events and purchase events fit together across Sponsored Ads, DSP, and conversions. The reads are powerful. But a gap sits between the data and the decision, and neither reporting access nor SQL skill closes it: the work of turning a growth problem into a concrete media action, fast enough to matter.
Most AMC projects don't fail because the data is weak. They fail because the team starts with the wrong question. They open a blank query editor and ask "what can I pull?" when the question is "what am I trying to fix?"
This guide starts with the fix.
If AMC has felt like more work than payoff, blame three things:
The result: many advertisers have AMC access, few have an AMC habit. The fix isn't more dashboards or a SQL hire. It's a method that ends in an action, and a way to run it without writing code.
AMC stitches your ad events and purchase events into whole-path reads; aggregated, anonymized, and privacy-safe by design. No PII leaves the clean room.
So you can answer questions a single-channel report can't. Which sequence of touches converts a new customer. When a hero-product buyer comes back. Whether your spend reaches new people at all. The hard part is knowing which question to ask first. That is ACE.
Nearly every AMC question fits one of three growth jobs.
Brands rarely ask for a model by name. They say "I'm stuck on…" and that sentence names the job. The job names the read. The read becomes the move.
Every example below uses one method: business question → AMC read → action. If the output doesn't change what you do next, it wasn't a useful read.
This table is the framework's core, the map from a plain problem to the read that answers it and the move it points to.
Find your row. Three real brands show you how to work down it.
The problem: "My hero product sells. How do I earn more revenue per customer?" A hero ASIN with high new-to-brand and low repeat is a strong front door with an empty hallway behind it.
The reads: Cross-Product Association shows what buyers purchase next. New-to-Brand shows whether the hero acquires or retains. Time to Conversion reveals the repurchase window. CLV shows what a hero buyer is worth over 6–12 months.
The move: Build a cross-sell audience of hero buyers who haven't bought the next SKU. Build a repurchase-window audience timed to the repeat peak. Test a virtual bundle where same-cart pair-lift is strong.
The proof: KAO (global CPG): In AMC Hub, KAO found an ~8-month repurchase cycle, a ~26% peak repeat rate in October, and a ~20-day window from new-to-brand to repeat. It aligned lifecycle campaigns and Subscribe & Save to those windows, turning DSP-driven demand into incremental Sponsored Products revenue and higher lifetime value. The hero stopped being a one-time sale and became the start of a relationship.
The problem: "Which combination drives conversions, and where am I wasting money?" When spend climbs but efficiency doesn't, the cause is allocation, budget stacked on the tactic that closes demand you already had.
The reads: Path to Conversion shows the sequences that drive sales. Multi-Touch Attribution credits the upper-funnel touches last-click ignores. Overlap Analysis separates synergy from cannibalization. Marketing Funnel Analysis finds the leak.
The move: Apply the MTA budget recommendation, shift from SP-only into upper-funnel DSP. Forecast the impact before touching the console. Build an exposed-not-converted audience to close the loop. Re-run Path to Conversion in 4–6 weeks.
The proof: Bernie's Best: In AMC Hub, Bernie's Best mapped its DSP path-to-conversion. Its most common journey — DSP into Sponsored Products retargeting — ran ~85% of paths but returned 10.9x ROAS. A deliberate nurture-loop sequence returned 26.1x; an awareness-led path, 13.6x. The most frequent path was not the most valuable. The fix wasn't more spend. It was rebuilding budget around the sequences that compound.
The problem: "How do I create new demand, and prove I'm expanding?" Flat new reach while spend rises isn't scale. It's saturation with better reporting.
The reads: Unique Reach shows the share of impressions reaching new uniques, and the cost to win them. Audience Labels surfaces adjacent segments you don't target yet. New-to-Brand shows which campaigns acquire and which cannibalize.
The move: Build a net-new prospecting audience from adjacent in-market segments. Exclude brand viewers and purchasers, so you measure net-new, not retargeting. Push to upper-funnel DSP, and judge the audience by new reach, not short-window ROAS. Net-new audiences pay back upstream.
The proof: Wuffes (DTC pet supplements): Entering a category run by legacy brands, Wuffes used AMC lookalikes and non-ad-exposed prospecting, built and activated in AMC Hub — to reach new shoppers. New-to-brand share doubled, from 6.5% to 16%. 63% of new-to-brand sales came from previously unexposed audiences, at a $0.27 CPC against a category average above $10. Expansion wasn't louder spend against the same pool. It was a move into new demand.
Each of those use cases ends in a move, and the move is where AMC usually breaks. A read you can't act on is just a slide. AMC value leaks in the distance between insight and activation: export, pivot, hand off, wait.
The Xnurta AMC Hub closes that distance. The same models live in a no-code gallery organized by business problem. The Hub doesn't stop at the read, it recommends the next best action, so you move from output to audience without guessing. Take the recommendation and build the audience, rule-based or lookalike, with AND / OR / Exclude logic and lookbacks from 7 days to 5 years, then push it to Sponsored Ads or DSP.
Three SQL queries, two exports, and a manual upload become a few clicks. (Custom reads still use native AMC SQL through Xnurta services; the daily work is no-code.)
That is the point of ACE: not to admire the data, but to act on it.
The framework isn't only for steady-state growth. Before the next traffic spike (Prime Day or any major event) audience strategy matters as much as keyword strategy. Each job has a play:
After the event, freeze your cohorts (turn auto-update off) and compare exposed, converted, and net-new buyers. Those cohorts seed your next launch. AMC is a loop, not a report.
You don't need SQL to do sophisticated AMC work. You need a method and a tool that does the heavy lifting. Bring one growth problem — a hero ASIN, a leaky funnel, a plateaued segment — to a working session with the Xnurta team. We'll map it to the right AMC read, show the workflow in the AMC Hub, and pin down the first action worth testing.
ACE stands for Amplify, Capture, and Expand, the three growth jobs almost every AMC question maps to. Amplify gets more from current buyers and products. Capture converts existing demand more efficiently. Expand reaches net-new buyers. Pick the job that matches your problem, run the AMC read it points to, and build one action.
Start with your problem, not the model. Use the AMC Model Map above: match your plain problem to an ACE job, and the table names the read (Cross-Product Association for cross-sell, Multi-Touch Attribution for funnel waste, Unique Reach for saturation) and the first move.
No. Native AMC uses SQL and exports, but the Xnurta AMC Hub runs pre-built models and creates audiences with no code. Native SQL is reserved for custom reads, handled through Xnurta services.
No fluff. Just what's working.
