Lists of the “top Amazon marketplace agencies” are useful for discovery. They are not useful for the decision that actually costs money.
A marketplace advertising agency can look excellent on paper and still be the wrong operator for a €5K+ account on Amazon, bol.com or MediaMarkt. The pitch deck may show strong ROAS. The website may mention DSP, AMC, Sponsored Products, creative, global expansion and proprietary software. The case studies may be real. None of that proves the person who will touch your bids knows when not to spend your money.
That is the mistake I see too often: brands choose an agency by reputation, then discover the operating model only after budget has already moved. By then the account has new campaign names, new bid logic, new negatives, new automation rules and a slightly awkward Slack channel where everyone is pretending the first month is “learning”.
My stance is simple: before you sign a marketplace ad retainer, run a proof week. Not a free audit. Not a vague strategy call. A structured five-day operator test where each shortlisted agency has to make the same decisions from the same evidence pack, under the same margin and stock constraints.
The best agency is not the one with the longest capability list. It is the one whose decisions survive contact with your contribution margin, stock cover and marketplace reality.
What competitor agency guides get right — and what they miss
During research for this article, I reviewed marketplace agency comparisons and agency pages from Podean, SalesDuo, Flapen, BidX partner pages, Smash.vc, Deltamatix and due-diligence guides such as Adverio. The good ones make three valid points.
- Scope matters. A full-service marketplace agency, a specialist PPC operator, an automation platform and an internal brand manager solve different problems.
- Operational depth matters. Advertising cannot be separated from retail readiness, catalog quality, stock, fulfilment, pricing and listing conversion.
- Questions matter. You should ask how the agency handles TACoS, ROAS, underperformance, reporting, missed targets and fee incentives.
Those are useful filters. But they still leave a dangerous gap: they test how well an agency explains its method, not how well it makes decisions when the account is messy.
Real marketplace advertising is messy. A product has 18% contribution margin after a coupon, not the 31% margin from last quarter. A bol.com hero SKU has 11 days of stock and LVB capacity is tight. A MediaMarkt listing has strong clicks but weak conversion because the warranty positioning is unclear. An Amazon campaign shows 4.2 ROAS, but 74% of the sales are branded and the TACoS line is still rising.
A pitch call rarely reveals how an operator handles that. A proof week does.
The proof week: five days, one evidence pack, zero dashboard theatre
The proof week is a short paid or unpaid evaluation project before the retainer starts. It should not give candidates access to move bids live. It should give them enough real evidence to show how they think.
The goal is not to get free consulting. The goal is to compare operator judgement.
Give each agency the same packet:
- 30 days of ad spend, sales, ACOS, ROAS, clicks and CPC by campaign and SKU.
- SKU-level gross margin, marketplace commission, fulfilment cost, return reserve and current break-even ACOS.
- Stock cover by marketplace: Amazon FBA/FBM, bol.com/LVB/FBR and MediaMarkt available units.
- Top search terms, wasted-spend terms and branded/non-branded split.
- Current campaign roles: defend, harvest, launch, learn, clearance or fix.
- Known business constraints: hero SKUs, forbidden margin floors, upcoming promotions, inventory risks and commercial priorities.
In FiveX, this is exactly the kind of evidence we would pull from profitability dashboards, advertising reports, product groups, stock insights, search-term data and change logs. The useful part is not one more dashboard. The useful part is that every candidate is forced to argue from the same facts.
Day 1: ask for the “do not touch” list
The first output should not be a growth plan. It should be a “do not touch” list.
Ask the agency: Which campaigns, SKUs, targets or budgets would you refuse to scale this week, and why?
This question separates operators from optimists. A weak answer starts with opportunity. A strong answer starts with permission.
Example: NovaNest sells home-storage products on Amazon and bol.com with €8,000 monthly ad spend. One Amazon Sponsored Products campaign for a drawer organiser shows 28% ACOS against a target of 30%. On a normal ROAS screen, it looks fine. But the proof pack shows the SKU has only 14% loaded contribution margin after a temporary 10% coupon and FBA fee update. The real break-even ACOS is 18%. Stock cover is 19 days. Scaling that campaign is not optimisation; it is buying negative-margin demand into a stock constraint.
The right proof-week answer is: cap the campaign, separate branded and generic terms, keep only the terms with clear repeat evidence, and re-open scaling after the coupon expires or margin is reset. If an agency says “ACOS is within target, increase budget”, they have failed the first day.
Day 2: force campaign roles before bid changes
On day two, ask the agency to label every active campaign with a role before suggesting any bid move.
Use simple roles:
- Defend: protect brand or core shelf space when margin permits it.
- Harvest: capture proven demand with strict CPC and stock controls.
- Learn: buy evidence, not volume.
- Launch: support a new SKU until evidence matures.
- Fix: keep minimal spend while retail readiness improves.
- Clear: move stock when margin math still allows promotion.
Why this matters: agencies often inherit campaigns with names that describe targeting, not purpose. “SP_AUTO_Exact_NL” tells you almost nothing about whether the campaign deserves money today.
FiveX product groups and campaign strategy labels make this easier in practice. A hero SKU can have a different budget permission than a clearance SKU. A launch SKU can tolerate a controlled learning loss. A margin-sensitive SKU cannot. The role has to be visible before automation or Ads AI recommendations are allowed to touch bids.
Day 3: test cross-marketplace budget judgement
A true marketplace advertising service should not treat Amazon, bol.com and MediaMarkt as three disconnected tabs. The proof week should include at least one cross-marketplace decision.
Example: BrewMate sells espresso accessories. Monthly ad spend is €6,200: €3,900 on Amazon, €1,700 on bol.com and €600 on MediaMarkt. Amazon reports 5.1 ROAS on a branded campaign. bol.com reports 3.2 ROAS on generic “melkopschuimer rvs” terms. MediaMarkt reports low volume but strong conversion on a premium bundle.
The lazy answer is to keep Amazon funded because ROAS is highest. The operator answer is to inspect incrementality and margin capacity. In this case, 81% of the Amazon campaign sales are branded repeat searches, the SKU has 24 days of FBA stock and the campaign is already defending rank. The bol.com generic terms bring new demand at 24% ACOS against a 32% loaded break-even ACOS, with 61 days of stock. MediaMarkt has a higher basket but only 12 units available.
The proof-week decision: move €500 from Amazon branded defence into bol.com generic harvest, keep MediaMarkt capped until replenishment lands, and set a review trigger after 100 paid clicks. That is not anti-Amazon. It is pro-profit.
Day 4: ask for the first seven changes — with expiry dates
By day four, the agency should propose a first change set. Limit it to seven changes. This prevents the classic “we will restructure everything” move that creates activity but destroys the account’s memory.
Every proposed change needs five fields:
- What changes: bid, budget, target, negative, campaign state, placement or role.
- Why now: the evidence that justifies the move.
- Profit permission: current margin, break-even ACOS or max CPC logic.
- Risk: stock, offer, listing, review, attribution or learning risk.
- Expiry: when the decision must be reviewed or reversed.
This is where FiveX change logs and automation rules become valuable. If a bid increase was allowed because a SKU had 73 days of stock and 29% contribution margin, that permission should not silently survive after stock drops to 16 days or the margin falls to 19%.
Good operators attach review dates to decisions. Average operators attach confidence to decisions. Confidence is nice. Review dates protect money.
Day 5: compare the agencies by decision quality, not charm
At the end of the proof week, score every agency using the same rubric.
| Proof-week signal | What good looks like | Red flag |
|---|---|---|
| Profit discipline | Uses SKU margin and break-even ACOS before scaling | Optimises only to campaign ROAS or platform ACOS |
| Stock awareness | Caps spend when stock cover or fulfilment capacity is weak | Mentions inventory only after you ask |
| Campaign role clarity | Labels each campaign by job and permission | Suggests restructures without role logic |
| Cross-marketplace judgement | Moves budget toward profit capacity, not platform politics | Protects the biggest channel by default |
| Change governance | Gives expiry dates, rollback rules and owners | Promises “continuous optimisation” without decision memory |
If two agencies are close, pick the one that made fewer but clearer decisions. Marketplace ad accounts rarely fail because the agency did too little in week one. They fail because someone moved too much before the economics were agreed.
The €5K threshold: when a proof week becomes non-negotiable
For small ad accounts, a heavy selection process can be overkill. If spend is €800 per month and the main problem is campaign hygiene, a specialist audit or software-led workflow may be enough.
At €5K+ monthly spend, the risk changes. A 15% misallocation is €750 per month. A bad retainer can cost more than the media mistake. A wrong campaign restructure can erase search-term memory that took months to build. A stock-blind budget increase can turn a profitable SKU into a stockout followed by expensive restart work.
Example: LumiClean runs €5,400 monthly across bol.com and MediaMarkt for cleaning electronics. Candidate A recommends a full account rebuild and a 20% budget increase because ROAS is above target. Candidate B recommends freezing two MediaMarkt campaigns, splitting bol.com generic and branded terms, and cutting €420 from a SKU whose return reserve recently increased from 6% to 14%. Candidate A sounds more ambitious. Candidate B is the operator I would trust with the first month.
How FiveX fits into the proof week
FiveX is not here to replace judgement. The point is to make judgement auditable.
For brands comparing marketplace advertising services, FiveX can support the proof week in four practical ways:
- Profitability dashboards show the SKU-level margin, fees, return reserves and break-even ACOS behind ad decisions.
- Advertising reports connect spend, revenue, TACoS, ACOS, ROAS, campaign roles and product performance across marketplaces.
- Inventory insights stop candidates from recommending spend increases on SKUs that cannot fulfil the demand.
- Change logs and automation rules make every bid, budget and pause decision reviewable instead of becoming invisible optimisation debt.
That also protects the agency. A good operator should want a clear evidence pack, clear permissions and clear review dates. It reduces politics. It prevents “but last month’s ROAS looked fine” conversations. It gives the agency room to do the right thing, including recommending less spend when less spend is the profitable move.
The final test: would you let them spend less?
The agency you choose should be able to grow the account. But first, they must prove they can protect it.
So ask one final question at the end of the proof week: where would you spend less in the first 30 days?
If the answer is vague, be careful. If the answer is specific, grounded in SKU margin, stock cover, campaign role and evidence expiry, keep talking.
Top Amazon marketplace agencies can be impressive. Full-service teams, proprietary platforms, global coverage and polished reporting all have value. But for a €5K+ Amazon, bol.com or MediaMarkt ad account, the deciding factor is simpler: can the operator prove which euros deserve permission today?
Start there. The rest of the pitch gets much easier to judge.