Marketplace ad accounts rarely waste money in one dramatic place. They usually leak it in small, polite ways.
An Amazon exact campaign wins a sale that a broad discovery campaign was also chasing. A bol Sponsored Products campaign in the Netherlands keeps buying a query that Belgium already proved unprofitable. A MediaMarkt retail media push sends traffic to a product that Amazon is also defending with branded search. Every campaign can look reasonable in isolation. Together, they create an internal auction.
The named mistake is treating campaign overlap as a hygiene issue instead of a profit decision. Teams find duplicate keywords, remove a few, add some negatives and call the account cleaner. That helps, but it misses the operator question: which campaign, marketplace and SKU is actually allowed to own this buyer intent?
My stance: €5K+ marketplace ad accounts need an overlap map. Not just a duplicate-keyword export. A working map that assigns ownership for search terms, product targets, branded demand, country splits and cross-marketplace roles before bids move. If two campaigns are chasing the same shopper, one of them needs a job, a ceiling or a stop sign.
This matters for managed Amazon, bol and MediaMarkt accounts because overlap is not always bad. Some overlap is deliberate defence. Some is learning. Some is leakage. The work is not to remove every collision. The work is to decide which collisions deserve budget.
What the existing advice gets right
The strongest Amazon PPC cannibalisation guides explain the classic mechanics well. SellerSprite defines keyword cannibalisation as multiple listings, campaigns or ad groups competing for the same search terms. It separates PPC-versus-PPC, PPC-versus-organic and ASIN-versus-ASIN overlap, then recommends keyword ownership, negative keywords and weekly maintenance.
SellerMetrics goes deeper on search-query cannibalisation. Its useful point is bid tiering: exact campaigns should own the highest-intent queries, while auto and broad campaigns should discover demand with lower bids and negatives. That is good PPC discipline.
BidX also gives practical structure advice. Their campaign-structure article recommends separating auto, broad, exact and product targeting so budgets can be controlled at campaign level. Their mistakes article pushes advertisers to set goals, optimise listings and avoid running ads without a strategy.
Podean’s agency-selection article adds the agency layer. It correctly argues that Amazon success needs more than media buying because retail operations, product pages, forecasting and replenishment affect advertising outcomes.
Reddit threads and seller discussions add the messy operator reality: sellers worry that if they group many products together, four or five products steal impressions from the rest; if they split everything too far, they lose control and spend too much time policing match types.
The missing layer is cross-marketplace profit ownership. Most advice stops at Amazon keyword hygiene. But a brand spending from €5K across Amazon, bol and MediaMarkt has a bigger problem: the same buyer intent can be duplicated across match types, products, countries, marketplaces and agencies. A negative keyword can prevent one collision. It cannot decide whether the next euro should defend Amazon rank, grow bol Belgium or test MediaMarkt retail media.
The overlap map: five ownership fields
An overlap map is a decision table. Every meaningful query, product target or audience cluster gets five fields:
- Intent owner: which campaign is allowed to buy the primary version of this demand?
- Marketplace role: is Amazon, bol or MediaMarkt defending, harvesting, learning, launching or redirecting demand?
- SKU permission: does the product have enough margin, stock, offer strength and review depth to receive spend?
- Overlap status: is the collision intentional, temporary, accidental or banned?
- Review date: when does the permission expire?
That last field is underrated. A campaign overlap can be correct on Monday and wrong by Friday. Stock changes. Competitors discount. CPCs rise. A bol NL campaign that was allowed to defend a search term before a promo may need to release it once Amazon has stronger stock and better conversion again.
FiveX fits naturally here because the overlap decision needs more than ad metrics. In FiveX, operators can compare ad spend, revenue, TACoS, SKU contribution margin and inventory signals in one place instead of debating from three platform dashboards. The point is not prettier reporting. It is giving the campaign owner permission evidence before money moves.
Example 1: TerraCook and the Amazon broad campaign that kept stealing the exact winner
TerraCook sells a €49.95 ceramic pan set on Amazon with a 31% contribution margin after fulfilment and marketplace fees. The account spends €7,200 per month. The exact campaign for “ceramic frying pan set” has a €0.78 CPC, 11.8% conversion rate and 24% ACOS. The broad discovery campaign also keeps matching that query at €0.64 CPC, but converts at 5.1% and sits at 44% ACOS.
A basic duplicate cleanup would add the exact term as a negative in broad. Correct, but incomplete.
The overlap map asks why the broad campaign is still allowed to touch that intent. In this case the answer is: it is not. The exact campaign owns the query because it has conversion history, a clear margin ceiling and enough stock: 41 days. The broad campaign keeps discovery rights only for non-owned variations such as “non stick pan gift set” and “induction pan bundle”. Its daily budget is capped at €38 until it proves a new query with at least 6 orders and ACOS below 32%.
The FiveX hook is the change log. When the operator adds the negative and caps the broad campaign, the reason should not be “optimisation”. It should be logged as: “Exact owns core intent; broad exceeded learning loss; stock and margin allow exact to scale.” That history protects the account when someone asks three weeks later why discovery traffic dropped.
Example 2: LunaSkin and the bol NL/BE split that looked like one campaign problem
LunaSkin sells a vitamin C serum on bol.com. NL spends €3,900 per month; BE spends €1,700. The same search term, “vitamine c serum gevoelige huid”, appears in both country splits. NL has a €0.42 CPC, 9.4% conversion rate and 21% ACOS. BE has a €0.39 CPC, 4.2% conversion rate and 47% ACOS.
The lazy answer is to lower the BE bid. The better answer is to check whether Belgium is allowed to own that intent at all.
The overlap map shows three constraints. First, the BE offer has a two-day slower delivery promise than NL. Second, the SKU margin is lower in BE because of a local promotion: 24% instead of 34%. Third, review depth is weaker: 38 reviews visible versus 214 in NL. The search term is not bad. The Belgian offer is not ready to buy it aggressively.
The decision: NL remains the harvest owner. BE moves to “watch” status with a €0.22 max CPC and a €12 daily learning cap. The term can graduate only after the BE offer reaches next-day delivery or conversion improves above 6.5% for two consecutive weeks.
This is where FiveX product profitability and inventory insights matter. A bol Ads platform view can show ACOS. It will not always show the margin and delivery-context argument cleanly enough for a weekly client decision. In FiveX, the ad operator can bring product margin, country performance and stock cover into the same conversation before the bid changes.
Example 3: VoltHub and the MediaMarkt launch that duplicated Amazon defence
VoltHub sells a USB-C wall charger at €29.95. Amazon is the mature channel: €9,400 monthly ad spend, 18% TACoS and stable organic rank on branded and category queries. MediaMarkt is new: €1,600 launch budget for retail media, mainly category placements around “USB-C charger fast charging”.
Two weeks into the launch, total paid sales look up. But the overlap map reveals a problem. MediaMarkt is buying broad category demand at €0.71 CPC while Amazon is also defending the same category with Sponsored Products at €0.58 CPC. Amazon has 63 days of stock and 3,800 reviews. MediaMarkt has 19 days of stock, fewer visible reviews and a higher return reserve because electronics customers compare warranty details carefully.
The right move is not to stop MediaMarkt. The launch has strategic value. But MediaMarkt should not compete with Amazon for the same broad category pool yet.
The decision: Amazon keeps category harvest ownership. MediaMarkt shifts into a narrower launch lane: bundle queries, retailer-specific placements and a €240 weekly test tranche. If MediaMarkt generates 20 orders with return reserve below 8% and stock cover above 28 days, it earns a bigger role. Until then, it does not buy the same category intent Amazon already converts profitably.
That is the operator voice in practice. Channel expansion is exciting. Paying twice for the same shopper while the weaker shelf has lower stock is not strategy. It is an expensive rehearsal.
How to build the overlap map in a weekly routine
1. Pull search terms and targets into one view
Start with the last 14 to 30 days of search terms, keywords, product targets and campaign performance. Do not only look for identical text. Group close variants, plurals, branded terms, category terms and product-use cases. “USB C charger”, “fast charger USB-C” and “phone charger 30W” may not be duplicates, but they can serve the same buyer intent.
2. Label every campaign role before judging performance
A 45% ACOS can be acceptable in a launch lane and unacceptable in a harvest lane. A branded defence campaign can overlap with organic rank by design. A discovery campaign should overlap temporarily, then pass winners to an owner. Without role labels, the team argues about metrics instead of decisions.
3. Add SKU economics before adding negatives
Before you block a term in one campaign, check the SKU that will inherit the traffic. Does it have margin headroom? Enough stock? Strong reviews? Stable price? If not, you may be moving spend from a messy campaign to an unready product.
4. Use negatives as routing rules, not punishment
Negative exacts and phrases are not a moral judgement on a term. They are traffic-routing rules. “This campaign is not the owner anymore” is a much better reason than “bad keyword”. That wording keeps future operators from reopening the same leak because they misunderstand the original decision.
5. Put expiry dates on overlap permissions
If broad discovery is allowed to overlap an exact owner for seven days, write the review date. If MediaMarkt is allowed a launch collision for a campaign test, define the order, margin and stock conditions that end the test. Overlap without expiry becomes account debt.
What to automate and what to keep human
Automation is useful once the ownership model is clear. FiveX Ads AI recommendations can propose bid changes, and automation rules can pause or reduce spend when ACOS, spend, stock or margin thresholds break. But automation should not decide the political question: which channel owns this demand?
Keep these decisions human:
- moving a high-volume search term from discovery to harvest ownership;
- letting a new marketplace compete with an existing profitable channel;
- overriding a campaign because stock, margin or offer strength changed;
- reopening a previously banned overlap after a price or fulfilment change.
Automate the guardrails around those decisions. Let the system catch spend above the cap, flag terms that appear in multiple owners, surface stock-risk conflicts and record bid changes. Let the operator decide the trade-off.
The weekly overlap-map checklist
- Which search terms appear in more than one campaign, match type, country or marketplace?
- Which of those overlaps are intentional defence, temporary learning or accidental leakage?
- Which campaign is the owner for each high-spend intent cluster?
- Which SKU has the strongest margin, stock and offer readiness to receive the traffic?
- Which negatives are traffic-routing rules, and is the reason logged?
- Which overlap permissions expire this week?
- Which automation rule could enforce the ceiling without hiding the strategic decision?
If your agency cannot answer those questions, the account is not being actively managed. It is being tidied.
Final thought: clean accounts are not the goal
A clean marketplace ad account can still be commercially wrong. You can have tidy campaign names, perfect match-type separation and beautiful reports while Amazon, bol and MediaMarkt quietly compete for the same shopper with different margins and stock positions.
The goal is not cleanliness. The goal is controlled intent ownership.
For brands spending from €5K per month, Advertentie Service should not only optimise bids. It should decide which campaign deserves the query, which marketplace deserves the shopper, and when overlap is learning versus leakage. That is how marketplace advertising becomes a profit system instead of a set of competing tabs.