Amazon Marketing Services is still one of the most searched phrases in marketplace advertising, even though Amazon retired the AMS name years ago. That alone tells you something important: clients often ask agencies for “AMS management” when they really mean a messy mix of Sponsored Products, Sponsored Brands, Sponsored Display, DSP, Amazon Marketing Stream, Amazon Marketing Cloud, catalog fixes, retail readiness and reporting.
For a solo consultant, that language gap is mildly annoying. For a marketplace agency with five or more people, it becomes operational risk. One person sells the retainer. Another builds the campaigns. A third handles content. A fourth checks inventory. Finance later asks why ad spend rose while contribution margin fell. Everyone thought they were managing “AMS”. Nobody agreed what AMS was allowed to change.
The named mistake I see is treating AMS as an ad channel instead of a decision system. The agency opens campaigns, optimises bids, adds negatives and reports ACOS. Meanwhile the client’s margin file is outdated, the hero ASIN has nine days of sellable FBA stock left, Sponsored Brands is defending a term that organic already owns, and a DSP retargeting line is being judged with the same KPI as Sponsored Products. The ad work may be technically correct. The operating model is not.
My stance: marketplace agencies need an AMS profit permission room. Not another Amazon Ads glossary. Not a prettier PPC dashboard. A weekly decision layer that translates “AMS” into current Amazon Advertising products, assigns decision rights, ties budget to SKU-level profit, and prevents old terminology from hiding new commercial complexity.
This guide is written for agencies in Germany, the United States and cross-border teams managing Amazon-heavy brands. The public advice on AMS is useful for beginners. But agencies do not only need to know what Sponsored Products are. They need to decide which client, SKU, campaign type and budget has permission to scale without turning the retainer into a margin leak.
What current AMS guides explain well
The strongest public resources do a decent job of translating the old AMS vocabulary. Productsup correctly notes that Amazon Marketing Services is now usually understood as Amazon Advertising, while many marketers still use the AMS label. Channable’s Amazon Ads guides explain Sponsored Products, Sponsored Brands, Sponsored Display, campaign creation and automation workflows. Amazon’s own ACOS guide is clear on the basic formula: ad spend divided by ad-attributed revenue, then compared with profit margin and break-even ACOS.
MerchantSpring focuses on the agency reporting problem: consolidating Seller Central data, PPC performance, profitability, fees and returns across accounts instead of downloading spreadsheets by hand. Rithum’s retail media content adds another important layer: advertising performance during peak events cannot be separated from inventory, orders, pricing and fulfilment. Pacvue’s Amazon Marketing Stream material points to hourly signals and faster optimisation for sellers operating across regions. Agency-focused posts, such as AMZ Dudes’ hiring checklist, are right to warn about account-manager workload, weak reporting narratives, data ownership and percentage-of-spend incentives.
That is all useful. But most of it still answers the wrong agency question.
The common question is: “How do we run Amazon ads well?” The agency question is sharper: “Which Amazon advertising decision is allowed to move client money this week, who owns the evidence, and what commercial constraint can veto it?”
The gap: AMS advice rarely defines decision rights
AMS content usually describes ad products. Sponsored Products capture shopping intent. Sponsored Brands build brand visibility. Sponsored Display reaches shoppers on and off Amazon. DSP can support awareness, retargeting and upper-funnel audiences. AMC helps analyse paths and audiences. Marketing Stream gives more frequent performance signals.
That taxonomy is helpful, but it does not stop a five-person agency from making expensive mistakes.
Imagine a client asks, “Can we increase AMS spend by $10,000 this month?” A junior PPC specialist may look at last week’s 22% ACOS and say yes. The account manager may say yes because the client wants growth. The content specialist may say wait because the listing has weak images. The inventory operator may say no because replenishment lands in 18 days. Finance may say no because the SKU margin after referral fee, FBA fee, coupon and agency hours is only 17%.
If the agency has no decision-rights model, the loudest metric wins. Usually that metric is ACOS, because it is visible, fresh and easy to explain. But ACOS is not the same as profit permission.
An AMS profit permission room fixes that. It turns every Amazon ad decision into a simple operating question: what is the campaign trying to do, what evidence supports the move, what constraint can block it, and who has authority to release budget?
Build the AMS translation table first
The first job is vocabulary control. Clients still say AMS. Amazon says Amazon Ads. Operators talk about Sponsored Products, Sponsored Brands, Sponsored Display, DSP, AMC and Marketing Stream. Finance talks about margin and cash. If those terms are not mapped, reporting becomes theatre.
Create a translation table inside the agency operating system. It should have four columns:
- Client phrase: “AMS”, “Amazon PPC”, “retail media”, “DSP”, “brand defense”, “launch budget”.
- Actual Amazon product: Sponsored Products, Sponsored Brands, Sponsored Display, DSP, AMC analysis, Marketing Stream signal.
- Primary decision: demand capture, brand protection, product launch, retargeting, share-of-voice push, audience learning.
- Profit permission owner: PPC lead, account manager, marketplace strategist, inventory owner or finance approver.
Here is the trade-off: this makes the agency slower for one afternoon. It makes the agency faster for the next six months. Without the table, every “AMS performance” meeting starts by re-litigating what is included. With it, the team can say, “This request is not one AMS request. It is a Sponsored Products scaling request, a Sponsored Brands defense request and a DSP learning request. They need different proof.”
FiveX helps here by giving agencies a shared marketplace analytics layer instead of letting every role work from a different export. Campaign, product, stock, revenue and margin signals can sit in the same client workspace, so the translation table is connected to current evidence rather than a static onboarding document.
Example 1: PeakHome DE and the false-positive Sponsored Products win
PeakHome DE is a fictional but typical German home-and-living brand managed by a six-person marketplace agency. The client spends €32,000 per month on Amazon Ads. A cookware ASIN shows €9,800 ad-attributed revenue from €2,450 spend in seven days. ACOS is 25%. The client’s target ACOS is 28%, so the PPC specialist proposes increasing the daily budget from €350 to €520.
On the surface, that looks reasonable.
The profit room adds three checks. First, the current contribution margin after referral fee, FBA fee, average coupon cost and COGS is 21%, not the old 30% still sitting in last month’s sheet. Second, FiveX inventory visibility shows nine days of FBA stock at the current sales pace and 27 days of stock in the 3PL. Third, organic rank improved on two main terms, so part of the ad-attributed lift may be demand that the brand would have captured anyway.
The decision changes. The agency does not scale Sponsored Products immediately. It keeps the daily budget at €350, opens a replenishment task, lowers bids on the two terms where organic rank is already strong, and moves €1,200 of test budget to a related ASIN with 34% contribution margin and 41 days of stock.
The lesson is not “never scale at 25% ACOS”. The lesson is that 25% ACOS has no decision right until margin, stock and organic context approve it.
Example 2: TrailKettle USA and the DSP KPI mismatch
TrailKettle USA sells outdoor cooking gear and works with an eight-person US marketplace agency. The client asks the agency to “add AMS retargeting” before a spring campaign. The team allocates $18,000 to Sponsored Products and $6,000 to Amazon DSP retargeting. After two weeks, Sponsored Products runs at 31% ACOS. DSP shows a weaker direct ROAS, so the client wants to cut it and move everything into Sponsored Products.
That reaction is understandable — and too simple.
The AMS profit permission room separates campaign roles. Sponsored Products must prove profitable demand capture at SKU level. DSP must prove incremental audience value, frequency discipline and assisted conversion, not the same last-click ACOS target. The agency reviews new-to-brand share, branded search lift, add-to-cart audiences, returning shopper windows and stock cover on the promoted bundle.
The evidence shows that DSP retargeting is not ready to scale, but it is not useless. Frequency is too high on one audience, the creative pushes a two-pack bundle with only 16 days of stock, and the audience excludes a profitable accessory ASIN with 39% margin. The agency cuts DSP from $6,000 to $3,500, caps frequency, swaps the bundle creative for the accessory, and keeps $2,500 in reserve until the next weekly review.
The named mistake would have been judging every AMS tactic with the same ACOS ruler. Sponsored Products, Sponsored Brands, Sponsored Display and DSP do not deserve identical KPIs just because a client groups them under “AMS”.
Example 3: Northstar Pets and the brand-defense trap
Northstar Pets is a pet-supplement brand with strong branded search. The agency spends $14,000 per month on Amazon Ads, including $3,200 on Sponsored Brands for branded terms. The campaign reports a beautiful 9% ACOS. In the monthly client call, the account manager highlights it as a win.
The profit room asks a less flattering question: what would happen if that branded budget did not exist?
The team checks organic position, competitor conquesting, Subscribe & Save share, branded CPC trends and TACOS. Branded Sponsored Brands is useful on two terms where competitors are aggressive. On three other terms, the brand owns top organic placement, competitor pressure is low and paid clicks are mostly subsidising demand that would probably convert anyway.
The agency does not kill brand defense. It makes it conditional. Two branded campaigns keep budget. Three move into a watchlist with a 40% budget reduction and competitor-pressure trigger. The freed $1,280 goes into non-brand category testing with a hard break-even ACOS ceiling.
This is where FiveX’s ad automation and rule logs become useful. Agencies can document why a budget moved, which condition triggered it, and whether the next action should be manual approval, bid change, pause or test extension. That record matters when the client asks, “Why did you reduce the campaign with the best ACOS?”
The four boards inside an AMS profit permission room
A good permission room does not need to be complicated. It needs four boards that the agency reviews on a fixed cadence.
1. The scope board
This board answers what “AMS” includes for the client. Some retainers cover only Sponsored Products and Sponsored Brands. Some include Sponsored Display. Some include DSP coordination but not execution. Some include AMC analysis once per quarter. The scope board prevents service creep and protects the agency’s margin.
For each client, list campaign types, marketplaces, deliverables, reporting frequency, approval thresholds and excluded work. If DSP is “advisory only”, say so. If Amazon Marketing Stream is used for monitoring but not automated hourly bid changes, say so. The board should be boring enough to survive a busy Friday.
2. The unit-economics board
This board decides whether a SKU can receive ad budget. It needs current selling price, referral fee, fulfilment cost, COGS, average promo cost, return impact, contribution margin and break-even ACOS. If the margin file is older than a defined threshold — for example 14 days for volatile categories — the board marks the SKU as “limited permission”.
FiveX’s P&L and SKU profitability views are built for this layer. The point is not to admire margin in a dashboard. The point is to give the PPC lead a current ceiling before bids move.
3. The demand-and-stock board
This board checks whether the agency is allowed to buy more demand. It combines campaign performance with inventory runway, inbound shipments, fulfilment route, Buy Box health, content readiness and organic ranking. A product with 11 days of sellable stock should not receive the same budget decision as a product with 54 days of stock and stable conversion.
The hard trade-off: sometimes the most efficient campaign should not scale. Agencies hate that because it feels like leaving revenue on the table. But buying profitable-looking demand into a stockout can destroy rank, waste learning, frustrate the client and create emergency work that the retainer never priced.
4. The client-decision board
This board records which decisions the agency can make alone and which require approval. A bid change within an agreed ACOS band may be agency-owned. A $7,500 monthly budget increase, DSP launch, promo-funded campaign or margin-negative launch push should require client sign-off. The board should include response-time expectations. If approval is not received by Thursday 12:00, the spend does not move. Silence is not consent.
FiveX supports this with campaign logs, ad recommendations, product strategies and reporting exports that help agencies turn decisions into proof rather than memory. That is the difference between “we optimised your AMS” and “we released €3,000 because these five SKUs met margin, stock and demand conditions.”
When AMS budget should be blocked
Not every strong campaign deserves more money. Agencies should define hard blockers before emotions enter the call. Common blockers include:
- Contribution margin below break-even ACOS.
- COGS or fulfilment cost missing for the promoted SKU.
- Less than 14 days of sellable stock with no confirmed inbound date.
- Suppressed Buy Box, weak delivery promise or listing quality issue.
- Campaign role unclear: launch, defense, harvesting, retargeting or awareness.
- Client approval required but not received.
- Automation rule conflict, such as one rule raising bids while another caps budget.
These blockers are not bureaucracy. They are how agencies protect their own service quality. A marketplace agency does not lose trust because it refuses a budget increase. It loses trust when it increases spend, then later explains that the economics were never checked.
The bottom line
Amazon Marketing Services is an old name. The agency risk behind it is very current.
Clients use “AMS” as shorthand for Amazon growth. Agencies hear an advertising brief. Amazon offers a growing set of ad products, reporting signals and audience tools. Somewhere between those three layers, budget starts moving. If the agency has no profit permission system, ACOS becomes the default judge and the retainer absorbs every missing piece of context.
The better model is simple: translate AMS into actual campaign products, assign decision rights, connect budget to SKU-level economics, check stock before demand scales, and document every release or hold. That is how an agency turns Amazon Advertising from a busy optimisation workload into a repeatable profit operating system.
FiveX fits naturally in that operating layer. It gives marketplace agencies one place to connect advertising performance, product profitability, stock signals, automation rules and client reporting. Not so the team can stare at more dashboards. So the next “Can we scale AMS?” question receives a confident answer: yes, no, or not yet — with proof.