Marketplace ad cannibalization is the quiet tax inside many healthy-looking Amazon, bol and MediaMarkt ad accounts. The campaign dashboard says the ads are working. ACOS is acceptable. ROAS is tidy. Spend is pacing. Everyone relaxes.
Then finance closes the month and asks the awkward question: why did paid sales rise while total contribution margin barely moved?
The named mistake is confusing defended revenue with created demand. A defensive Sponsored Products campaign on your own brand term may be necessary. A top-of-search bid on a keyword where you already rank number one may be smart during a competitor attack. A MediaMarkt retail media placement may protect shelf share during a launch week. But if that spend becomes permanent without a cannibalization review, you start paying marketplace rent on sales you were already likely to win.
My stance is simple: for brands spending €5K+ per month across Amazon, bol and MediaMarkt, cannibalization should not be discussed as a vague attribution worry. It needs a weekly profit court. Every defensive campaign has to prove one of three things: it protects profitable share, it creates incremental demand, or it should release budget to a SKU that can actually grow.
What the current advice gets right
The market is finally moving past ROAS worship. Autron explains Amazon PPC cannibalization clearly: paid ads can capture clicks from keywords where your listing already has strong organic visibility, while TACoS shows whether ad spend is expanding total revenue or only changing the paid-versus-organic mix. Feedvisor makes the useful point that paid sales are not automatically bad; they can support sales velocity and ranking, but brands should ask whether they would have generated the same sales without advertising. BidX gives TACoS a practical role in spotting dependency on advertising and connecting paid activity to organic sales. Ad Badger’s PPC Den discussion adds the operator nuance: when ad sales triple and organic sales fall, the cause may be branded overspend, shrinking search volume, lost rank, or a genuinely protective ad role.
That is all useful. The gap is that most cannibalization content stays inside Amazon and inside the advertising console. It talks about organic rank, paid sales ratio, TACoS and holdout testing. Important, yes. But a brand running an Advertentie Service brief in NL/BE has a wider problem: the same euro can defend Amazon branded search, push bol Sponsored Products, test MediaMarkt retail media, or sit out because the SKU has no stock cover left.
The missing layer is not “does PPC cannibalize organic?” The sharper question is: which defensive ad spend still deserves profit permission this week?
The three types of cannibalization that matter
Operators should separate cannibalization into three buckets. If you mix them together, every debate becomes philosophical and nobody changes the budget.
1. Search cannibalization
This is the classic version. You pay for clicks on a keyword where your SKU already has strong organic rank and strong conversion. Sometimes that is fine. If a competitor is bidding aggressively above you, a defensive ad can protect share. If you own the organic top spot, the Buy Box, the best price and enough reviews, a full-power bid may be buying your own demand back.
2. SKU cannibalization
This happens when ads shift demand from one of your products to another without improving total profit. A sponsored bundle may steal orders from a higher-margin single item. A discounted accessory may take clicks from a premium kit. The ad account celebrates revenue, but the product mix gets worse.
3. Channel cannibalization
This is especially relevant for Amazon, bol and MediaMarkt management. Extra spend on one marketplace can shift demand from another channel where the same product had better contribution margin, lower returns, more stock or a stronger strategic role. The campaign looks efficient in isolation. The business loses the better order.
FiveX is useful here because the platform connects marketplace analytics, ad performance, product profitability and inventory data in one operating view. You cannot judge cannibalization from the Amazon Ads console alone. You need the SKU’s margin, stock cover, return rate, price position and channel role next to the ad result.
Build a weekly cannibalization profit court
A profit court is not a dramatic meeting. No robes. No gavel. Sadly. It is a simple weekly review where defensive spend must bring evidence before it keeps budget.
Use four questions:
- What demand are we defending? Brand search, generic keyword rank, product-page share, competitor ASIN traffic, category shelf visibility or launch velocity?
- What would likely happen without this spend? Would the SKU keep organic rank, lose top-of-search visibility, hand orders to a competitor, or simply save money?
- What is the SKU allowed to pay? Use loaded contribution margin, not a neat ad-console ACOS target. Include marketplace commission, fulfilment, returns, discounts, agency/service cost allocation and stock risk.
- Where would the next euro work harder? Another keyword, another SKU, another marketplace, a listing fix, a price correction or no spend at all?
The output should be one of five labels: Defend, Trim, Test-off, Shift or Stop.
FiveX advertising automation can then turn those labels into action rules: cap branded CPCs, reduce bids when organic rank is stable, shift budget when stock cover falls, and flag campaigns where paid sales rise while total SKU contribution does not. The human still makes the commercial call; automation stops the account from drifting back into old habits by Thursday.
Example 1: LumaHome defends less on Amazon brand search
LumaHome sells a compact air purifier on Amazon.nl for €89. The loaded contribution margin before ads is €24.50 per unit after referral fees, FBA, returns and a small coupon reserve. The account spends €620 per month on branded search for “LumaHome air purifier”. The campaign reports 9.8% ACOS and looks wonderful.
The profit court adds context. The SKU ranks first organically for the brand term, holds the Buy Box, has 4.6 stars, and receives roughly 870 total monthly units. Branded ads claim 71 ad-attributed units, but total units barely change in weeks when branded spend is higher. Worse, non-brand category terms are underfunded even though the SKU has 46 days of stock cover.
The decision is Trim. Keep a low defensive presence at €120 per month with a CPC ceiling, but move €500 to two non-brand tests: “air purifier bedroom” and “air purifier pollen”. If total branded units fall by fewer than 10 units while the new tests produce at least 18 incremental units at a target ACOS below 22%, the trim becomes permanent.
This is not anti-brand advertising. It is anti-laziness. Defensive spend is allowed when it protects profit. It is not allowed to sit there forever because the ROAS makes everyone feel clever.
Example 2: NordTrail stops bol Sponsored Products from eating scarce stock
NordTrail sells rain jackets on bol.com in NL and BE. The hero SKU has a €64.95 price, 18% loaded contribution margin after bol commission, LVB costs, payment fees and returns. Sponsored Products spend is €1,150 per month on “regenjas heren” with 4.1 ROAS. On the surface, fine.
The FiveX profitability dashboard shows the less glamorous truth. The SKU has only nine days of stock cover in the Netherlands, Belgian conversion is 31% lower because size availability is patchy, and organic rank in NL is already position two for the main term. Paid clicks are accelerating sell-through in the market where stock is shortest, while the Belgian campaign buys lower-quality demand.
The decision is Shift. NL generic bids are reduced by 40% until inbound stock is confirmed. BE spend moves from broad “regenjas heren” to a narrower “waterdichte regenjas heren zwart” query where available sizes match demand. €300 of the saved budget moves to a higher-margin softshell SKU with 52 days of stock cover and 28% contribution margin.
The ad manager did not “pause a winner”. They protected a winner from becoming a stockout, then used the same budget where the business could still fulfil profitably. That is the difference between bid management and marketplace ad management.
Example 3: MysaTech separates MediaMarkt conquesting from Amazon substitution
MysaTech sells a USB-C docking station on Amazon, bol and MediaMarkt. Amazon has the strongest organic demand but a lower contribution margin after a €12 coupon. MediaMarkt has lower volume, but the product performs well when bundled with a laptop accessory and returns are lower.
The team wants to scale MediaMarkt retail media by €2,000 after a sponsored placement shows 5.6 ROAS. Nice number. The profit court asks whether this is new demand or a channel shift from shoppers who would have bought on Amazon anyway. FiveX marketplace analytics shows that the MediaMarkt lift is strongest on weekends, while Amazon branded search dips at the same time. The bundle attach rate is only 12%, below the 25% required for MediaMarkt to beat Amazon contribution margin.
The decision is Test-off, but carefully. MediaMarkt spend is held for weekdays and reduced on weekends for two weeks. Amazon branded search is capped, not removed. The test asks one narrow question: when MediaMarkt weekend ads are reduced by €500, does total docking-station contribution fall by more than €500? If not, the weekend budget was probably substituting demand rather than creating it.
This is where a managed Advertentie Service should earn its fee. The answer is not “MediaMarkt works” or “Amazon works”. The answer is which marketplace deserves the next euro under today’s margin, offer and bundle conditions.
The metrics to put in the court file
Do not overload the review with every metric. The point is decision speed. Use a one-page file per SKU or campaign cluster:
- Paid share of total sales: if it rises while total units stay flat, check cannibalization.
- TACoS trend: rising TACoS with flat revenue is a warning; falling TACoS with growing total sales is healthier.
- Organic rank band: top 1-3, top 4-10, page two, or not visible. Exact rank is noisy; the band is enough for decisions.
- Loaded break-even ACOS: after fees, fulfilment, returns, discounts and a service-cost allowance.
- Stock cover: because paying to accelerate a SKU with seven days of stock is usually just expensive panic.
- Channel role: defend, scale, learn, clear, launch or harvest.
- Next-euro alternative: where the same budget would go if this campaign lost permission.
FiveX AI recommendations can help surface the cases humans should review first: branded campaigns with rising paid share, generic keywords where organic rank is already strong, SKUs with falling contribution margin, and channel shifts where one marketplace grows while another shrinks. That is much more useful than asking an operator to inspect every campaign every Monday morning like it is a tiny spreadsheet museum.
When cannibalization is acceptable
Let’s be practical. Not every cannibalized sale is a sin. Sometimes you knowingly pay for demand you might have won because the alternative is worse.
Defensive brand campaigns can make sense when competitors are bidding on your brand and your organic listing is pushed below sponsored results. Top-of-search spend can be rational when a priority SKU is climbing from position five to position two and the extra velocity improves organic rank. Product-page targeting can protect accessories or bundles from cheaper substitutes. During launches, a higher TACoS may be acceptable because the SKU is buying review velocity and search history.
The operator rule is not “never cannibalize”. The rule is name the reason, cap the cost, set an expiry date. A defensive campaign without an expiry date becomes a subscription to your own demand.
How FiveX runs this for Advertentie Service accounts
For Advertentie Service clients, the cannibalization review sits inside the weekly marketplace ad cadence. First, we map SKU economics: product margin, marketplace fees, fulfilment, returns, discounts and stock cover. Then we connect that to Amazon, bol and MediaMarkt ad performance. Finally, we label campaign clusters by role: defend, grow, learn, clear or stop.
The value is not that every bid move becomes complicated. Quite the opposite. Once the rules are visible, many decisions become easier. A branded Amazon campaign with stable organic rank and rising paid share gets trimmed. A bol SKU with low stock cover loses scale permission. A MediaMarkt placement with high bundle contribution keeps budget. A generic term with weak organic rank but strong incremental units gets more room.
That is the work: not making ads look efficient, but making marketplace demand more profitable.
The weekly checklist
- List the top 10 campaigns by spend across Amazon, bol and MediaMarkt.
- Mark which ones are defensive: brand, high-organic-rank generic, product-page protection or retargeting.
- Add SKU contribution margin, stock cover, return rate and channel role.
- Compare paid sales movement with total SKU sales movement for the last four weeks.
- Assign Defend, Trim, Test-off, Shift or Stop.
- Document the reason and the expiry date.
- Move freed budget to the best next-euro alternative, not to the loudest marketplace rep.
If your agency or internal team cannot show this review, they may still be optimizing campaigns. They are not yet operating marketplace advertising as a profit system.
Final thought
Cannibalization is not a reason to panic and slash advertising. It is a reason to grow up operationally. Mature marketplace accounts need defensive spend, conquesting, launch support and rank protection. They also need the discipline to ask whether last month’s protective spend still deserves this month’s budget.
The best Advertentie Service work is not heroic bid tweaking. It is the weekly habit of putting every meaningful campaign in front of the same commercial question: are we creating profitable demand, protecting profitable demand, or paying for demand we already owned?
Answer that honestly and the budget usually knows where to go.