Marketplace advertising KPIs for agencies: the scorecard that turns metrics into margin decisions

Marketplace advertising KPIs for agencies should do more than prove that campaigns were managed. They should show which client accounts are allowed to grow, which ones need protection, and which ones are quietly turning specialist time into unprofitable revenue.

That sounds obvious until you look at the average agency report. It has spend, sales, ROAS, ACoS, TACoS, clicks, CPC, CTR, conversion rate and maybe a few keyword winners. Useful metrics. But for a marketplace agency managing Amazon, bol.com, Walmart, Mirakl retailers or retail media networks across 8, 20 or 60 clients, those numbers are only the raw material. They are not the operating system.

The named mistake is what I call metric theatre: a client deck full of green arrows that still does not answer the commercial question, “Did this account become healthier this week?” A 6.4 ROAS can hide a negative-margin SKU. A falling ACoS can hide underinvestment on a hero product with 50 days of stock. A strong CTR can distract from a landing page that converts poorly because the Buy Box is unstable. Everyone has numbers. Not everyone has decisions.

My stance is simple: agencies should report marketplace advertising KPIs in three layers: media efficiency, marketplace reality and client profit. If a KPI cannot change a bid, budget, SKU priority, stock conversation, pricing action or client expectation, it belongs in the appendix, not in the weekly operating view.

What competitors explain well — and where the gap is

The market already has decent KPI content. MerchantSpring’s agency article covers the classic Amazon advertising metrics well: TACoS, ACoS, ROAS, CTR, CVR, CPC and ad spend efficiency. That is helpful for explaining what each metric means and why agencies need reporting discipline.

ChannelEngine’s marketplace KPI content is strong on seller health: delivery performance, cancellation rate, return rate, Buy Box, profitability by marketplace and operational standards. That matters because marketplaces do not judge accounts by ad results alone. A client can have beautiful advertising and still lose visibility because service metrics are weak.

Pacvue’s retail media positioning is sharp on connected commerce: ROAS, conversion rate, share of voice, retail sales signals, inventory and Buy Box intelligence. Rithum makes a good point that ROAS is not comparable across every retail media placement because funnel stage and ad format matter. Productsup approaches agencies from the product feed and client scale angle, where data quality and infrastructure become the basis for performance.

The gap is the agency decision layer. Most content explains metrics one by one. Agencies do not manage one metric at a time. They manage portfolios of clients, each with different margin structures, fee models, stock positions, marketplace maturity and tolerance for risk. The better question is not “Which KPIs should we track?” It is: Which KPI combinations should trigger action today?

The KPI stack agencies actually need

A practical agency KPI stack has four levels. Each level answers a different question.

• Campaign KPIs: Is the media doing what we asked it to do?
• SKU KPIs: Is the product able to turn demand into profitable orders?
• Account KPIs: Is the marketplace channel becoming healthier?
• Agency KPIs: Is the team spending time where it creates client value?

Most agencies are comfortable with the first level. The commercial advantage comes from connecting all four.

1. Campaign KPIs: still necessary, but not enough

Start with the standard advertising metrics because they are still the diagnostic language of campaign management.

• Spend: how much budget was used by marketplace, client, campaign and SKU.
• Ad sales: attributed revenue inside the marketplace attribution window.
• ACoS: ad spend divided by attributed ad sales.
• ROAS: attributed ad sales divided by ad spend.
• CTR: clicks divided by impressions; useful for search relevance and creative pull.
• CVR: orders divided by clicks; useful for offer quality, listing quality and conversion friction.
• CPC: cost per click; useful for auction pressure and bid efficiency.

The operator trap is treating these as success metrics by themselves. They are not. They are symptoms.

Example: an Amazon Sponsored Products campaign for “Klara Home 3L air fryer” spends €1,200 in a week, generates €7,800 in attributed sales and shows 15.4% ACoS. On the surface, excellent. But if the SKU has 18% contribution margin before ads, a €2.10 return reserve per unit and a €1 coupon, the campaign is probably too aggressive. The KPI that matters is not 15.4% ACoS. It is whether the campaign stays below the SKU’s break-even ACoS after returns and promotions.

This is where FiveX helps agencies move beyond ad-console reporting. FiveX connects advertising data with product profitability, marketplace fees and operational cost assumptions, so the team can see whether “good” ACoS is actually good for that client’s margin.

2. TACoS: the account dependency metric

TACoS is ad spend divided by total marketplace revenue. Agencies should use it to understand whether advertising is building the account or simply buying the same sales every week.

But TACoS needs context. A rising TACoS can be healthy during a launch, a Prime Day push, a bol.com category entry or a share-of-voice offensive. A falling TACoS can be unhealthy if the account is starving new products and coasting on old organic rank.

For agency reporting, split TACoS into three views:

• Account TACoS: total ad spend divided by total account revenue.
• SKU TACoS: SKU-level ad spend divided by total SKU revenue.
• Objective TACoS: TACoS by campaign role: launch, rank defence, profit harvesting or liquidation.

Scenario: “Nordlicht Coffee” sells on Amazon.de with €180,000 monthly revenue and €21,600 ad spend. Account TACoS is 12%. The client asks the agency to reduce it to 9%. That sounds reasonable until the split shows two realities: branded defence campaigns run at 4% TACoS, while launch campaigns for compostable capsules run at 28% TACoS but are moving organic rank from position 28 to position 9. Cutting all spend evenly would protect the monthly report and damage the growth plan. The right action is to cap low-margin launch SKUs, not punish every campaign.

3. Break-even ACoS: the margin guardrail agencies should personalize per SKU

Break-even ACoS is the maximum ACoS a product can carry before advertising turns contribution margin negative. The simple version is contribution margin percentage before ads. The agency version is more precise:

Break-even ACoS = contribution margin after marketplace fees, fulfilment, COGS, returns, discounts and variable operating costs.

This KPI should not be one target per client. It should be one target per SKU-channel combination. The same product may tolerate 22% ACoS on Amazon FBA, 14% on bol.com with LVB, and 9% on a Mirakl retailer because commission, fulfilment and returns differ.

The trade-off: SKU-level guardrails take more setup work. But they prevent the most expensive agency habit: applying one target ACoS to an entire catalogue. A blended 25% target can overfund weak-margin accessories and underfund high-margin bundles.

FiveX can turn this into a rule layer: campaigns can be flagged when ACoS exceeds the SKU’s margin ceiling, when a product has enough margin to scale, or when the current target is commercially impossible. That gives specialists a decision queue instead of another spreadsheet tab.

4. Stock and Buy Box KPIs: because you cannot advertise what you cannot sell

Advertising KPIs become misleading when stock and Buy Box are ignored. Agencies should include at least four operational signals next to ad performance:

• Days of stock cover: expected days until sell-out at current velocity.
• Buy Box win rate: percentage of time the client owns the purchasable offer.
• Suppression or listing issue rate: products live in ads but not healthy in commerce.
• Return rate: products where advertising creates revenue but post-purchase economics deteriorate.

Scenario: “PeakTrail Gear” sells hiking poles on Amazon.com and Walmart. The Amazon campaign has 5.1 ROAS and 19% ACoS, comfortably below the 31% break-even ACoS. A junior specialist wants to raise budget from $350 to $700 per day. The stock view shows only nine days of FBA inventory left and the next inbound shipment is 18 days away. Scaling is not performance marketing. It is stock-out acceleration. The correct KPI action is “protect rank with capped spend” until replenishment lands.

That is a perfect place for FiveX’s inventory insights and AI recommendations. Instead of asking a specialist to remember every client’s stock situation, FiveX can surface products where ad scaling conflicts with stock cover, Buy Box stability or repricing pressure.

5. Share of voice and placement KPIs: useful only when tied to a role

Share of voice, impression share and placement mix are powerful, especially for agencies managing competitive categories. But they should be interpreted by campaign role.

For a launch campaign, rising top-of-search impression share may justify a temporarily worse ACoS. For a profit harvesting campaign, the same placement premium may be wasteful. For a branded defence campaign, low CPC and high conversion might be the goal, not maximum exposure.

Report placement KPIs with labels:

• Defend: protect branded demand and competitor conquesting.
• Build: earn ranking and category visibility.
• Harvest: capture profitable demand efficiently.
• Clear: move overstock without pretending it is normal growth.

This prevents the client conversation from becoming “Why did ROAS fall?” and turns it into “We paid for visibility because this SKU had 42% margin, 60 days of stock and a launch target.” Much better. Fewer dramatic client calls, too. Always a bonus.

6. Client action KPIs: the missing agency layer

Some KPI problems cannot be fixed inside the ad platform. If conversion rate drops because content is weak, if ACoS rises because a competitor cut price, or if TACoS climbs because stock was unavailable for two weeks, the agency needs a client action KPI.

Track the number and age of open commercial actions:

• price decision needed;
• stock decision needed;
• content fix needed;
• promotion decision needed;
• margin data missing;
• approval needed for budget reallocation.

This is not admin. It is profitability protection. If an agency flags that Client B needs to approve a €4,000 budget shift from Amazon Sponsored Brands to Walmart Sponsored Products, and the approval sits for ten days, the next report should show the opportunity cost. Agencies that make dependencies visible protect both client outcomes and their own team from unfair blame.

7. Agency efficiency KPIs: measure the work behind the work

For a five-person agency, KPI reporting must also protect capacity. Track:

• Accounts per specialist: segmented by complexity, not just count.
• Manual reporting hours: time spent exporting, formatting and explaining data.
• Exception-to-action time: hours between a KPI breach and a specialist decision.
• Automation coverage: percentage of safe recurring decisions handled by rules.
• Client escalation rate: recurring issues that need client input.

The goal is not to make specialists robotic. It is to remove low-value repetition so they can spend more time on judgement. FiveX’s multi-client dashboards and automated recommendations are designed for exactly this: one agency view across client accounts, with margin, ads, inventory and profitability signals in the same place.

A practical weekly KPI dashboard for marketplace agencies

If I had to design one weekly client view, it would have seven blocks:

1. Commercial summary: revenue, ad spend, contribution margin after ads and account TACoS.
2. Decision queue: scale, protect, fix, harvest or stop.
3. Campaign diagnostics: ACoS, ROAS, CTR, CVR, CPC and spend pacing.
4. SKU margin guardrails: break-even ACoS, current ACoS and margin remaining.
5. Marketplace reality: stock cover, Buy Box, returns, price movement and listing health.
6. Budget movement: where budget moved and why.
7. Client actions: decisions required from the client before the next optimisation cycle.

Notice what is missing: a 40-slide tour of every campaign. The dashboard should not prove that the agency looked at everything. It should prove that the agency knew what mattered.

The KPI rulebook: what should trigger action?

Here is a simple starting rulebook agencies can adapt:

• If ACoS is above break-even for seven days and stock is healthy, review bids, queries and conversion blockers.
• If ACoS is below target and stock cover is above 30 days, test controlled budget increases.
• If TACoS rises while organic sales fall, investigate dependency on paid demand.
• If ROAS improves while total revenue falls, check whether the account is becoming too conservative.
• If conversion rate drops by more than 20% week over week, inspect price, Buy Box, reviews and listing changes before cutting bids.
• If a SKU has fewer than 14 days of stock, cap growth campaigns unless the client explicitly accepts the stock-out risk.
• If margin data is missing, do not label campaigns “profitable”. Label them “performance visible, profit unknown”.

That last label is important. Agencies build trust when they say what they know and what they do not know.

Final thought: better KPIs make agencies harder to replace

Clients can get campaign metrics from every ad console. They can get charts from dozens of reporting tools. What they cannot easily get is a partner who connects advertising to margin, stock, marketplace health and decisions.

That is the opportunity for marketplace agencies. Do not win the KPI conversation by adding more metrics. Win it by making the metrics more useful. Show the client where money is being created, where it is leaking, where their team must act, and where the agency has already protected profit.

FiveX helps agencies build that operating layer: multi-client marketplace dashboards, advertising performance, SKU profitability, inventory signals, repricing context and AI recommendations in one place. The result is a simpler weekly conversation: not “Here are the numbers”, but “Here is what we should do next.”