Marketplace advertising teams are trained to find winners. That is useful. It is also incomplete.
Once an Amazon, bol or MediaMarkt account passes roughly €5K per month, the bigger profit leak is often not that winners are underfunded. It is that weak SKUs keep receiving just enough budget to prove, again and again, that they are weak. They sit in automatic campaigns, inherited category targets, retargeting pools, Sponsored Products groups or retail media tests because nobody has created a formal way to remove them from the ad system.
The named mistake is using ads as a permanent life-support machine for assortment debt. A product has thin contribution margin, mediocre conversion, awkward stock, a weaker offer than another marketplace, or a return pattern that eats the sale after attribution looks fine. The campaign report says “learning”. The operator says “we need more data”. The catalogue team says “it is part of the range”. Six weeks later, the account has spent €1,850 proving something the P&L already knew.
My stance: managed marketplace ad accounts need an assortment pruning room. Not a one-off SKU cleanup. Not a brutal “pause everything under target ACOS” rule. A weekly decision layer that classifies which products deserve ad support, which deserve a controlled learning budget, which should move to another marketplace, and which should be removed from paid demand until the product economics change.
This is especially relevant for NL/BE brands running Amazon, bol and MediaMarkt together. A SKU can be a bad advertising candidate on Amazon but a useful bol defender. It can be a weak standalone product on MediaMarkt but a profitable bundle component. It can have acceptable ACOS in one dashboard and still be destroying contribution margin once fulfilment, returns, marketplace fees and stock opportunity cost are included. Assortment pruning is not about shrinking the catalogue. It is about stopping paid media from subsidising the wrong part of it.
What the existing advice gets right — and what it usually misses
The public marketplace advertising advice has improved a lot. BidX explains Amazon PPC structure, target ACOS, max CPC formulas and the need for retail readiness before campaigns scale. Their controlling guidance is strong on the relationship between price, conversion rate, target ACOS and maximum CPC. Podean’s retail-metrics content rightly points to inventory, Buy Box rate, conversion, profitability and SKU-level management as core marketplace disciplines. Sellerboard connects Amazon PPC to profit through break-even ACOS and TACOS. Adverio goes further into underperforming SKUs, contribution margin, aged inventory and return rates. Pare’s budget allocation framework reminds teams that profitability is often an allocation and structure problem, not just a bid problem.
All of that is useful. The gap is the operating decision between “this SKU is underperforming” and “what exactly happens to its ad permission tomorrow morning?” Most guides diagnose. Fewer define the room where diagnosis becomes budget removal, learning quarantine, channel shift or catalogue escalation.
That is the angle FiveX can own: ad-funded assortment pruning. The question is not only whether a campaign is efficient. The sharper question is whether the product still deserves paid demand at all.
The pruning room starts with product permission, not campaign permission
A campaign can look clean while hiding weak product economics. A broad Amazon campaign may contain 40 advertised SKUs. A bol Sponsored Products group may promote several variants. A MediaMarkt placement may be bought for a category, while only three products actually have enough margin and stock to absorb the clicks.
So the pruning room starts one level lower: product permission.
Every advertised SKU gets one of five labels:
- Scale: margin, stock, offer strength and evidence are good enough to receive more budget.
- Defend: the SKU deserves visibility, but only on branded, exact or high-intent demand.
- Learn: the SKU may deserve budget, but only inside a capped experiment with a clear evidence goal.
- Fix: advertising is paused or capped until price, content, reviews, fulfilment or stock improves.
- Prune: the SKU should leave paid campaigns for now because the economics do not justify more demand.
This is where FiveX’s product profitability view becomes practical. Instead of asking an ad operator to guess whether a 31% ACOS is acceptable, the operator sees contribution margin, purchase price, shipping cost, fulfilment cost, return reserve and marketplace fees next to campaign spend. A SKU with 44% pre-ad margin and 72 days of stock can tolerate a different test than a SKU with 17% loaded margin and 11 days of stock. The bid is not the first decision. Permission is.
Example 1: the Amazon kettle that keeps “almost working”
Imagine a kitchen brand spending €7,500 per month on Amazon Ads. One electric kettle sells for €39.95. After referral fees, fulfilment, payment costs, expected returns and COGS, the loaded contribution margin before ads is €8.40 per unit, or 21%.
The Sponsored Products campaign reports 28% ACOS on this kettle over 30 days: €1,260 spend, €4,500 attributed sales and 113 units sold. At first glance, the operator might cut bids slightly and keep learning. But the pruning room does the actual math. At 28% ACOS, ad cost is €11.20 per €39.95 sale. The SKU only has €8.40 of pre-ad contribution margin. Every ad-attributed unit is roughly €2.80 underwater before overhead. Organic rank has not improved; TACOS for the product is still 19%; stock cover is 18 days.
The decision is not “reduce bids by 12%”. The decision is Prune from scale campaigns. Keep €150 for exact branded defence if needed, remove the SKU from broad discovery and category targets, and send the product to a fix lane: price test, bundle option or COGS review. FiveX campaign change logs make that decision visible later, so nobody re-adds the kettle to a broad campaign two weeks later because last month’s report said it had sales.
Example 2: the bol accessory that deserves a defend lane, not a growth lane
Now take a phone-accessory seller on bol. A USB-C cable has solid demand, but it is a low-ticket product: €12.99 selling price, €2.10 loaded pre-ad contribution margin after platform costs, fulfilment and return reserve. The account spends €5,400 per month across bol Sponsored Products. The cable campaign spends €620 and returns €2,480 attributed revenue at 25% ACOS.
That looks acceptable if the team uses a generic 30% target ACOS. It is not acceptable if the product’s loaded margin is only 16%. The cable is also a repeat add-on product with high branded search presence. The pruning room gives it a Defend label, not Prune. It stays live on exact brand and product-name terms with a €0.18 max CPC, but leaves generic “usb c kabel” discovery where the average CPC is €0.41 and the margin cannot absorb the auction.
The result is not less sophistication. It is better role discipline. FiveX search-term reporting helps separate add-on defence from generic growth attempts. FiveX Ads AI recommendations can still suggest bid changes, but the SKU label limits which recommendations are allowed to apply. Automation should not turn a margin-thin add-on into a growth bet just because the campaign found revenue.
Example 3: the MediaMarkt smart-home bundle that should shift channel
A smart-home brand tests MediaMarkt retail media with a starter budget of €1,200 for a sensor bundle. The bundle sells at €89.99 and has €24.50 loaded pre-ad contribution margin. Early results are not terrible: €920 spend, €3,680 attributed revenue, 41 orders and 25% ACOS.
The dashboard would probably keep it alive. The pruning room asks a better question: is MediaMarkt the strongest paid-demand destination for this bundle right now? Stock cover is 14 days at current run rate. Amazon has 48 days of stock for the single sensor, but the bundle page has only six reviews on MediaMarkt and a slower delivery promise than the brand’s bol offer. The same product family converts better on bol when promoted as a two-pack with LVB delivery.
The decision is Shift and Learn. MediaMarkt keeps a €250 learning budget focused on high-intent category visibility, but €700 of the planned second tranche moves to bol for the two-pack while stock and review depth improve. This is pruning without panic. The product is not “bad”. The channel-role and offer readiness are not ready for scale.
The five checks before a SKU receives paid demand
The assortment pruning room needs a simple weekly scorecard. I like five checks because they force the conversation away from dashboard theatre.
1. Loaded contribution margin
Use the margin that survives real operations: COGS, marketplace fee, fulfilment, shipping, pick-pack, return reserve, payment cost and any recurring discount. If the SKU has 18% loaded margin, a 22% ACOS is not “slightly above target”. It is loss-making unless there is a deliberate rank or launch reason.
2. Stock and replenishment runway
Do not scale demand into a stockout. A SKU with 12 days of sellable stock should not receive the same discovery budget as a SKU with 70 days and confirmed inbound stock. FiveX inventory insights help turn this from a vague warning into a budget veto.
3. Offer strength per marketplace
Price, delivery promise, reviews, Buy Box or featured offer status, content quality and bundle clarity all change what a click is worth. If Amazon has the reviews but bol has the stronger delivery promise, the pruning decision may be channel-specific.
4. Evidence maturity
Separate 36 clicks from 900 clicks. A weak SKU with thin evidence deserves a capped learning test, not immediate deletion. A weak SKU with six weeks of spend, stable conversion and negative contribution deserves pruning.
5. Strategic role
Some products deserve defensive spend because they protect a hero product, support a bundle, block competitors or anchor category perception. That role must be written down. Otherwise “strategic” becomes the excuse every weak SKU uses to stay funded.
What to remove first
Start pruning where the financial exposure is high and the regret risk is low.
- SKUs with negative loaded contribution after ad spend for two consecutive review cycles.
- Products with low stock cover and no confirmed replenishment date.
- Variants that receive clicks but lose to a stronger sibling product on price, reviews or delivery.
- Low-ticket products spending in generic discovery terms while only defending branded demand profitably.
- Products with high return rates where attributed sales arrive before refund reality.
Do not start with the emotional catalogue debate. Start with paid exposure. If a SKU is unprofitable but not receiving ad spend, it is a catalogue problem. If it is unprofitable and receiving €600 a month in clicks, it is an advertising governance problem.
How FiveX supports the pruning room
FiveX is useful here because the decision crosses the normal borders between ads, stock and finance.
First, SKU profitability connects ad performance to actual contribution margin instead of platform ROAS. Second, inventory insights show whether demand can be fulfilled before a campaign receives more budget. Third, search-term and campaign reporting reveal whether the SKU is defending existing demand, learning new demand or paying for traffic another product should capture. Fourth, Ads AI recommendations and automation rules can work inside the labels set by the pruning room, rather than blindly chasing ACOS.
For Advertentie Service clients, this becomes part of the weekly operating cadence. We are not trying to make every SKU advertise. We are trying to make every advertising euro answer one question: does this product still deserve paid demand?
The operator rule
Here is the rule I would put on the wall: no SKU gets to spend because it exists in the catalogue.
It spends because it has margin permission, stock permission, offer permission and a clear role. If one of those breaks, the product moves from Scale to Defend, Learn, Fix or Prune. That may feel strict. Good. Marketplace auctions are strict too. They do not care that the product manager likes the SKU, that the catalogue looks fuller with it, or that last quarter’s campaign once had a good week.
The best ad operators are not only bid optimisers. They are assortment editors with a profit ledger. They know when to fund winners, when to protect strategic products, when to test patiently and when to stop paying to prove weak SKUs are weak.
That is how €5K+ marketplace ad accounts become calmer, cleaner and more profitable: not by giving every product another chance, but by giving every product the right kind of chance.