Marketplace advertising approval often fails in the least dramatic way possible. Nobody makes a reckless decision. Nobody “forgets” the account. The operator sees an Amazon campaign that needs a bid cut, a bol Sponsored Products term that deserves more budget, or a MediaMarkt category push that should pause because stock is suddenly tight. Then the work sits in a grey zone: is this an optimisation, an agency decision, a finance decision, or a client approval?
That grey zone gets expensive once a brand spends from roughly €5K per month on Amazon, bol and MediaMarkt ads. At that level, a single budget move can change contribution margin, stock availability, organic ranking, retail-media learning and next week’s cash position. The wrong approval process does not only slow the team down. It quietly decides who is allowed to expose profit.
The named mistake I see is using one approval rule for every advertising change. “Ask the client before major changes” sounds sensible until nobody agrees what major means. A €0.18 CPC increase on a high-volume keyword may be more material than launching a small test campaign. Pausing a campaign can protect margin, but it can also surrender a keyword you had been defending for three months. Letting automation reduce a bid by 12% can be safe on a harvest campaign and dangerous on a launch SKU that needs ranking evidence.
My stance: every managed marketplace advertising service needs an approval ladder. Not a bureaucracy ladder. A profit ladder. It defines which bid, budget, pause, restart, target, campaign-role and automation changes can be handled automatically, which need operator review, which need commercial approval, and which require a full budget-court decision before spend moves.
FiveX helps because the approval level should not be based on campaign spend alone. It should see SKU profitability, stock cover, repricing context, return pressure, marketplace performance and AI recommendations in one operating view. The best approval workflow is not the one with the most sign-offs. It is the one that gives fast permission to safe work and slows down only the decisions that can damage profit.
What current marketplace ad advice gets right
The market has improved a lot here. Pacvue talks about approval-based actions inside an AI-powered commerce media operating system, plus budget pacing, dayparting and retail-aware execution using inventory, pricing and profitability signals. That is the right direction: media decisions should not live in an ad-platform bubble.
Flywheel’s retail media positioning is strong on automation, intraday bidding, placement targeting and incrementality. It also makes the useful point that search is intent-rich and that placement performance needs proper measurement. Podean’s RPM approach explicitly connects retail media with non-media signals such as low stock, out-of-stock data, price changes and merchandising. BidX focuses more on automation efficiency: faster campaign creation, bid and budget optimisation, keyword suggestions and analytics.
The newer Amazon PPC workflow content from Trellis and SalesDuo is practical too. Trellis recommends weekly reviews with exceptions, search terms, pacing, bids, placements and an action log. SalesDuo separates what should always be automated, what should be automated with oversight, and what should stay manual, such as campaign structure, launch strategy and ACOS target-setting.
That advice is useful. The gap is approval economics. Most resources say “use oversight”, “document changes” or “approve sensitive actions”. Fewer explain how to decide whether a specific Amazon, bol or MediaMarkt action deserves automation, operator judgment, finance review or client sign-off based on the euros at risk. That is the gap FiveX can own.
The approval ladder: four levels of spend permission
A good approval ladder gives every recurring advertising action a default route. The trick is to route by commercial exposure, not by how dramatic the platform button looks.
Level 0: auto-execute
These are low-risk changes where the data is fresh, the SKU has margin, stock is healthy and the change sits inside pre-approved rules. Examples: reducing a bid by 8% after 14 days above target ACOS on a harvest campaign, lowering a daily budget on a campaign that repeatedly exhausts learning spend without orders, or adding a search term to a watchlist instead of blocking it.
Level 0 needs hard guardrails: maximum percentage change, minimum data volume, margin floor, stock-cover floor, no active price-change lock, and a change log. Automation is welcome here. It is repetitive work with clear permission.
Level 1: operator review
These decisions require marketplace judgment but not a commercial committee. Think bid moves on high-volume terms, search-term isolation, campaign-role cleanup, budget reallocation between campaigns inside the same channel, or restarting ads after a stockout recovery. The operator can approve if FiveX shows margin, stock, performance and campaign role still agree.
Level 2: commercial approval
These decisions change the business case, not just the campaign. Moving budget from Amazon to bol, scaling a MediaMarkt category push, changing target ACOS after a cost increase, keeping ads live during low stock, or accepting a short-term learning loss should go through commercial approval. Finance, ecommerce or the brand lead may need to confirm whether the trade-off is intentional.
Level 3: budget court
These are decisions where the account could look better while the business gets worse. Level 3 covers material monthly budget increases, margin-negative launch campaigns, defensive spend that cannibalises organic demand, automation rules that touch many SKUs, and campaign freezes after a major operational shock. The output is not “yes” or “no”. It is a written decision: approve, cap, test, delay, replace, or stop.
Example 1: NorthPeak Home and the Amazon CPC lift
NorthPeak Home sells a €42 kitchen organiser on Amazon.nl. The SKU has a 31% contribution margin before ads, 46 days of stock and a harvest campaign running at 24% ACOS. A priority non-branded keyword has 620 clicks over 14 days, converts at 8.1%, and Amazon’s suggested bid rises from €0.72 to €0.94 after a competitor runs a deal.
A basic workflow might ask the client before increasing bids because the keyword is important. A better approval ladder says this is Level 1 operator review, not Level 2. The max CPC math still works: €42 selling price multiplied by 31% pre-ad margin gives €13.02 of contribution before ads. At 8.1% conversion, break-even CPC is about €1.05 before allowing for returns and ranking strategy. The operator approves a smaller lift from €0.72 to €0.82, sets a 72-hour review, and caps spend at €180.
The decision does not need a meeting. It needs proof. FiveX would show the SKU margin, stock cover, paid/organic movement and campaign role in the same place, then record why the operator chose €0.82 instead of blindly following the suggested €0.94.
Example 2: BrightNest Beauty and the bol low-stock veto
BrightNest Beauty spends €6,800 per month across bol and Amazon. Its bol Sponsored Products campaign for a €19.95 serum looks efficient: 21% ACOS, 7.4% conversion and a search term that generated 38 orders in two weeks. The account manager wants to move €400 from Amazon discovery into bol because the numbers are cleaner.
On campaign metrics alone, that looks like a Level 1 decision. The approval ladder upgrades it to Level 2 because stock cover has dropped from 29 days to 10 days after a wholesale order pulled inventory away from bol. LVB availability is still active today, but replenishment will not land for 12 days. Scaling bol spend would not just create sales; it could break delivery promise, reduce conversion and force the brand to throttle a channel that was already working.
The commercial decision becomes: protect visibility, but do not scale. The operator keeps branded and exact harvest terms live, cuts discovery by 35%, holds the €400 in reserve and asks ecommerce to confirm replenishment timing. FiveX inventory insights and advertising analytics make the veto obvious before the team celebrates a cheap ACOS that the warehouse cannot support.
Example 3: VoltEdge and the MediaMarkt launch loss
VoltEdge launches a €129 docking station on MediaMarkt marketplace with a monthly ad budget of €5,500 across Amazon, bol and MediaMarkt. The MediaMarkt test spends €52 per day for ten days, produces €1,430 in attributed sales and reports 36% advertising cost of sales. The SKU margin after marketplace fees and fulfilment is 27%, so the campaign is not profitable on first-order math.
Should the operator pause it? Not automatically. The campaign is a launch lane, not a harvest lane. The approval ladder routes it to Level 3 budget court because the team is choosing whether to buy evidence, not merely whether to fix a bid.
The budget court asks four questions: is MediaMarkt a strategic channel for this electronics SKU, is stock cover above 30 days, did branded search or organic PDP traffic lift during the test, and what loss limit was pre-approved? In this case stock cover is 41 days, PDP visits rose 18%, but only 9 of 27 assisted orders were new-to-brand and the pre-approved learning loss was €450. The test has already used €520 of loss. Decision: stop scaling, keep a €15/day defensive trickle, move €300 back to Amazon exact match, and reopen only if conversion reaches 4.5% or MediaMarkt lowers CPC below €0.68.
That is the value of a ladder. It stops a launch campaign from being killed too early by harvest logic, but it also stops “learning” from becoming a polite word for unmanaged loss.
How to build the approval ladder
Start with the decisions you already repeat every week. Do not begin with org charts. Begin with money movement.
- Bid changes: define auto limits by campaign role, data volume, margin floor and CPC ceiling.
- Budget moves: separate same-channel reallocations from cross-marketplace reallocations.
- Pauses and restarts: require evidence of stock, offer health, search intent and campaign role before spend resumes.
- Target ACOS changes: treat every target as a margin promise with a reason, owner and expiry date.
- Automation rules: route any rule touching many SKUs or high spend through Level 2 or Level 3 before activation.
- Learning spend: pre-approve loss limits, review dates and success criteria before the test starts.
Then add escalation triggers. A normally safe action should climb the ladder when any of these appear: stock cover below 14 days, contribution margin below 18%, CPC up more than 25% week over week, price changed in the last seven days, return rate above category expectation, Buy Box or offer position unstable, or the action touches more than 20% of monthly spend.
This is also where FiveX’s AI recommendations become more useful. The AI should not simply say “increase bid” or “move budget”. It should say: “This recommendation is Level 1 because margin and stock are healthy,” or “This is Level 3 because it changes the monthly budget exposure and the SKU has only 11 days of stock.” That turns AI from a suggestion machine into an operating assistant.
The approval fields your change log should capture
Every approved change needs a record that can survive the next weekly meeting. Keep it short enough that operators actually use it.
- Action: bid, budget, pause, restart, target, placement, keyword, automation rule.
- Marketplace: Amazon, bol, MediaMarkt or cross-marketplace.
- SKU or SKU group affected.
- Campaign role: defend, harvest, launch, learn, scale, fix.
- Approval level: Level 0, 1, 2 or 3.
- Reason: margin, stock, CPC, conversion, search term, offer, promotion, competitor, learning.
- Exposure: estimated spend, margin at risk and time window.
- Owner and approver.
- Review date and rollback rule.
The review date is not admin. It is the safety mechanism. A Level 2 decision without a review date becomes a permanent strategy by accident. A Level 3 learning test without a loss limit becomes a monthly surprise. Very few brands need more meetings. Many need fewer decisions left open forever.
What a good ad service should own
If you outsource marketplace advertising, the service should not ask approval for everything. That would be slow and slightly exhausting. But it should be very explicit about which decisions it can take alone and which decisions need commercial permission.
At €5K+ monthly spend, I would expect an agency or managed service to own Level 0 and Level 1 inside agreed boundaries. I would expect Level 2 to be handled in a weekly or twice-weekly decision rhythm with ecommerce or finance. I would expect Level 3 to be rare, written and tied to budget exposure. If every meaningful change is Level 3, the service has no mandate. If everything is Level 0, the brand has no governance.
The sweet spot is boring in the best way: routine optimisations move quickly, commercial trade-offs are visible, and nobody needs to reconstruct why €750 shifted from bol to Amazon three weeks ago.
Final thought: approval is part of optimisation
Approval workflows have a reputation for slowing marketers down. Sometimes they do. But in marketplace advertising, approval is not a separate admin layer. It is part of optimisation because the approval route decides how much evidence a decision needs before money moves.
The operator voice version is simple: do not let the platform button define the risk. Let contribution margin, stock, evidence maturity and spend exposure define the risk.
That is how an approval ladder protects speed and profit at the same time. Amazon, bol and MediaMarkt ad accounts can still move quickly. They just stop pretending that every bid change, budget shift and campaign pause deserves the same level of permission.