Retail media packages do not usually fail on launch day. They fail three weeks later, in the meeting where everyone politely accepts the retailer report.
The package ran. The screenshots look tidy. Amazon shows attributed sales. bol.com reports clicks, CTR and ACoS. MediaMarkt shares exposure, category context and a neat performance summary. Someone says, “Looks promising — should we do the next one?”
That is the dangerous moment.
For a marketplace advertising service managing €5K+ across Amazon, bol.com and MediaMarkt, the job is not only to approve packages before they launch. The job is to run a package verdict room after they end. Not a reporting readout. A verdict room.
Here is the operator stance: a retail media package has not earned renewal because it produced attributed revenue. It has earned renewal only when it proves what it added, what it displaced and what profit survived after the operational aftershock.
This article gives you the framework we use: five evidence checks, five verdicts and three named examples with real numbers. The goal is simple: stop turning pretty post-campaign reports into automatic next-month spend.
The named mistake: renewing the report, not the result
The most common mistake is what I call report renewal. The team renews because the package report contains enough positive numbers to feel safe.
That sounds reasonable until you look at what the report usually leaves outside the frame.
- It may show attributed sales, but not how much would have happened through organic rank or always-on Sponsored Products anyway.
- It may show ROAS, but not whether the package borrowed budget from a campaign with better contribution profit.
- It may show impressions, but not whether the hero SKU ran down to eight days of stock cover.
- It may show click-through rate, but not whether the package changed search-term quality or simply harvested brand demand.
- It may show new shoppers, but not whether returns, discounts and marketplace fees left enough contribution margin.
Competitor advice often explains measurement, incrementality or dashboards well. Flywheel rightly pushes AMC-powered incrementality and iROAS. BidX explains flexible Amazon dashboards with ACOS, ROAS, total profit and AMC metrics. bol.com is moving its measurement closer to IAB retail media standards. Skai, Feedvisor and other retail media voices keep reminding advertisers that attribution is not causation.
All useful. But the missing operator layer is the renewal decision itself. After the report lands, someone must say: renew, resize, absorb, quarantine or stop. That decision should be written down before the next retailer package gets budget.
What the package verdict room must decide
A verdict room is a 45-minute decision meeting with one rule: no one is allowed to say “good ROAS” until the team has named the baseline, the sacrifice and the profit impact.
The meeting needs four inputs:
- The retailer report: spend, impressions, clicks, attributed sales, ACoS/ROAS, dates, placements and format delivery.
- The FiveX profitability view: SKU contribution margin, fulfilment cost, fees, discounts, returns reserve and net profit after media.
- The advertising change log: which bids, budgets, campaigns or marketplaces were capped to fund the package.
- The operational view: stock cover, Buy Box or offer status, price changes, delivery promise, content issues and return signals during the package window.
This is where FiveX is useful in a very practical way. The platform can connect ad spend with SKU economics, stock cover and marketplace performance, so the verdict is not trapped inside a retailer dashboard. The retailer can tell you what the package claimed. Your operating system must tell you whether the claim deserved money.
The five evidence checks
1. Delivery integrity: did we receive what was sold?
Start with the boring check. Boring checks save budget.
Was the package delivered on the promised dates? Did the homepage slot, Sponsored Brands burst, newsletter mention, category tile or product-targeting layer actually run? Was spend paced correctly or did it burn through too early? Did the package point to the intended SKU, store page or assortment?
If delivery was materially different from the proposal, the verdict cannot be “renew”. It can only be “reconcile first”. A package that underdelivered may still create sales, but the renewal decision should not treat incomplete delivery as a clean proof point.
2. Measurement contract: what exactly is being counted?
Before judging performance, write down the measurement contract in plain language:
- Sales scope: advertised SKU only, total brand, halo SKUs or category-level sales?
- Window: click-only, view-through, 7 days, 14 days or retailer-specific?
- Customer type: new-to-brand, returning buyers, category buyers or unknown?
- Returns: included, excluded, delayed or not visible yet?
- Comparison: no control, pre/post, geo holdout, audience control or matched baseline?
If the package uses broad brand sales with a generous view window, the attributed result may be directionally useful but not renewal-grade evidence. That does not make the retailer wrong. It means the operator must label the confidence level.
In FiveX, this is where we like to separate reported revenue from decision revenue. Reported revenue is what the platform credits. Decision revenue is the portion you are willing to use when approving the next euro.
3. Displaced-spend test: what did the package steal from?
Every fixed package has an opportunity cost. If you spend €3,000 on a MediaMarkt category bundle, that money usually comes from somewhere: Amazon exact-match campaigns, bol Sponsored Products, a launch test, defensive brand search or margin-safe remarketing.
The verdict room must compare the package against the spend it displaced, not against doing nothing.
Ask one uncomfortable question: If we had left the same budget in the best eligible always-on campaigns, what would we reasonably expect?
The answer does not need fake precision. A range is enough. If the displaced Amazon campaign normally returns €0.48 contribution profit per click and the package produced €0.19 after returns, the package needs a strategic reason to survive. Visibility alone is not a strategy; visibility with new demand, retailer leverage or category defence can be.
4. Profit survival: what contribution margin remains after the report?
ROAS is a starting line, not a verdict.
The package verdict should calculate contribution after product cost, marketplace commission, fulfilment, discounts, return reserve and media cost. If the package changed the product mix, do not use an average margin. Use the actual SKUs sold.
A package can produce lower ROAS and still deserve renewal if it attracts new category buyers into high-repeat SKUs. Another can show beautiful ROAS while harvesting existing branded demand on a low-margin bundle. The verdict room protects the P&L from both errors.
This is one of the cleanest FiveX hooks: when SKU profitability, ad spend and returns sit in one view, the team can stop arguing about whether 5.2x ROAS is “good” and start asking whether €1 of package spend created enough contribution profit to repeat.
5. Operational aftershock: did the package damage the next month?
Retail media packages can create a hangover.
They can drain stock, shift demand to the wrong marketplace, train campaigns on weak search terms, trigger stockouts, expose a poor delivery promise, or make a retailer expect another package next month. The post-campaign verdict must include these aftershocks.
Check stock cover seven days after the package. Check whether Buy Box or offer strength changed. Check if organic rank improved or simply spiked and faded. Check whether returns are tracking above the SKU baseline. Check if always-on campaigns lost learning because budgets were capped.
FiveX inventory insights and change logs help here because they show whether the package result was a clean win or a win with a bill arriving later.
The five verdicts
The output of the verdict room should be one of five labels.
Renew
Use this only when delivery was clean, measurement confidence is acceptable, contribution profit survived, the package beat or justified its displaced-spend alternative and there is no stock or offer hangover. Renewals should include a new hypothesis, not just “same again”.
Resize
The package worked, but only partly. Maybe one SKU, placement, marketplace or audience carried the result. Keep the useful part, cut the rest. This is where FiveX advertising automation can help by turning the verdict into new caps, bid ceilings and budget release rules instead of another manual reminder.
Absorb
The package produced learning, but the best next step is not another package. Move the winning search terms, audiences or SKUs into always-on campaigns where pacing, bids and margin controls are stronger.
Quarantine
The report looks promising, but returns, stock impact, attribution confidence or displaced-spend comparison is not mature enough. Freeze renewal until a specific date or condition. Quarantine is not indecision; it is disciplined waiting.
Stop
The package failed the profit case or created operational damage. Stop means the next proposal must be materially different before it gets budget again.
Three named examples
Example 1: NovaBrew renews only the profitable half
NovaBrew spends €6,500 per month across Amazon and bol.com for coffee accessories. bol.com offers a €2,400 seasonal visibility package. The report shows €13,200 attributed revenue and a 18.2% ACoS. Nice enough for a cheerful slide.
The verdict room finds the split. Two grinder SKUs drove €9,100 of the attributed revenue at 34% contribution margin before media and 14 days of stock cover after the campaign. A low-margin milk frother drove €4,100 at only 17% contribution margin and a 9% return reserve. Funding the package also capped an Amazon exact-match campaign that usually creates about €620 contribution profit per week.
Verdict: resize. NovaBrew approves a €1,200 follow-up only for the grinder SKUs, with a stock floor of 21 days and a rule that Amazon exact-match budgets cannot be capped to fund it. The frother is removed from the package and moved into a listing-fix lane.
Example 2: StudioNest absorbs the learning instead of renewing MediaMarkt
StudioNest sells smart lighting on Amazon and MediaMarkt. A €3,800 MediaMarkt package runs around a category event. The retailer report shows 420,000 impressions, 5,900 clicks, €18,600 attributed sales and 20.4% ACoS.
On the surface, that looks acceptable. The post-campaign analysis shows something more useful: 68% of package sales came from two starter kits with enough margin, but most converting queries were generic “smart lamp set” and “wifi lamp starter pack” terms that StudioNest was barely covering in always-on search. Returns are normal. Stock cover remains 32 days. The package itself did not clearly beat the displaced Amazon Sponsored Products budget, but it revealed a demand pocket.
Verdict: absorb. StudioNest does not renew the fixed MediaMarkt package. Instead, it moves €900 per month into always-on MediaMarkt search coverage for those two starter kits, adds a €0.58 max CPC ceiling, and uses FiveX AI recommendations to monitor whether the new generic terms keep contribution profit above €0.31 per click.
Example 3: ArcLight quarantines a beautiful Amazon report
ArcLight sells premium desk lamps and runs €11,000 monthly marketplace ad spend. Amazon proposes a Sponsored Brands and video package for €5,000. The report lands with €41,000 attributed sales, 12.2% ACoS and strong new-to-brand numbers. Everyone wants to renew.
The verdict room slows the room down. The hero lamp entered the campaign with 27 days of stock cover and ended with 11. The top-selling variant now has a 19-day replenishment lead time. The package also used a broad brand sales scope, so halo sales from existing organic rank are mixed into the result. Returns are not mature yet because the SKU has a 21-day return curve.
Verdict: quarantine. Renewal is frozen for 18 days. The condition is simple: stock cover must return above 28 days, returns must stay below 7%, and decision revenue is capped at 55% of reported revenue unless Amazon provides a cleaner audience or query split. If those conditions pass, ArcLight can resize into a €2,500 test. If not, the money goes back to always-on campaigns.
The operating cadence
The verdict room should happen within five working days of receiving the retailer report, then reopen once delayed returns or stock data mature. For most €5K+ accounts, that means two moments:
- Day 5 verdict: delivery, early revenue, displaced spend and immediate stock impact.
- Day 21 verdict: returns, contribution profit, organic lift, repeatable learnings and renewal decision.
Do not let the next package proposal arrive before the previous package has a verdict. That is how retail media calendars become subscription plans in disguise.
Final thought: make retailers earn the next package
Good retailer relationships matter. Amazon, bol.com and MediaMarkt can all bring useful visibility, category moments and shopper data. The point is not to reject packages by default.
The point is to stop treating a retailer report as the final truth.
A serious marketplace advertising service should protect the next euro with evidence. What changed because the package ran? What did it displace? Which SKU margin survived? What stock, return or search-term aftershock did it create? And what exact verdict does that evidence deserve?
When those questions are answered, package renewals become calmer. Some packages earn more budget. Some shrink. Some become always-on campaigns. Some wait. Some stop.
That is not anti-growth. It is growth with a memory.