Marketplace agencies do not usually lose client trust because one chart is ugly. They lose it because two reasonable people look at two reasonable dashboards and cannot agree what the number means.
The Amazon Ads console says attributed sales are up 28%. Seller Central shows business reports down 4% week on week. Walmart Connect has a different attribution window. Shopify includes discounts in one view and excludes them in another. Finance asks whether returns are netted out. The client founder asks why the agency is celebrating ROAS when the payout did not move. Nobody is being careless. The metric definitions are drifting.
The named mistake I see is treating reporting automation as reporting governance. An agency connects Amazon, Walmart, bol.com, TikTok Shop, Shopify, retail media and maybe a feed tool into a smart dashboard. That is useful. But if the team has not agreed what “sales”, “profit”, “ACOS”, “TACoS”, “stock cover”, “margin” and “incremental revenue” mean for this client, automation simply distributes confusion faster.
My stance: every marketplace agency with five or more people needs a metric dictionary inside its agency software stack. Not a dusty glossary in a shared folder. A living operating layer that defines each client-facing metric, names the source system, records the calculation, flags known caveats and decides which number wins when systems disagree.
This guide is for marketplace agencies in Germany, the US and other mature ecommerce markets managing Amazon, Walmart, bol.com, Kaufland, Otto, Target, TikTok Shop, Shopify or Mirakl clients. If your account managers spend Monday explaining why last month’s “revenue” changed after returns, this is for you. And yes, I am saying the boring dictionary might be more commercially valuable than the shiny dashboard. Marketing heresy. Useful heresy.
What competitor advice gets right
The marketplace software category has improved a lot. MerchantSpring is strong on agency reporting: multi-client workspaces, white-label dashboards, scheduled reports, custom layouts and AI-supported analysis from connected marketplace data. Their agency message is practical: scale the agency, not the reporting stack, and stop rebuilding exports before every client conversation.
Pacvue goes deeper on retail media execution. Its agency positioning focuses on cross-retailer campaign management, standardized workflows, automation, benchmarks, iROAS and client-facing dashboards. That is valuable when a team manages Amazon, Walmart, Target, Instacart and other networks with different media mechanics.
ChannelEngine and Rithum are strong on connected commerce operations. They talk about centralized reporting, SKU-level profitability, FBA and WFS fees, settlement reports, pricing guardrails, inventory and operational issues. Productsup and Channable focus more on product data and feed operations: faster activation, reusable transformation logic, channel readiness, fewer listing errors and better data quality for marketplaces and AI discovery.
Reddit and agency forums add a more human signal. Agencies complain that reporting takes too long, clients do not always open live dashboards, custom Looker Studio builds become hard to maintain, and monthly PowerPoint packs still survive because clients want a narrative, not only access. Amazon seller discussions also show a recurring pain: profitability tools are only trusted when users can see exactly how fees, refunds and returns are calculated.
All of that advice is useful. The missing layer is definition control. Most content helps agencies collect, visualize and send data. Much less explains how to prevent a client from debating the meaning of the metric after the report lands.
The unique angle: the metric dictionary is a trust system
A metric dictionary is not a list of KPI names. It is the contract behind every recurring client conversation. For each metric, it answers six questions:
- What does this metric mean? Plain-language definition.
- Where does it come from? Source system, report and refresh cadence.
- How is it calculated? Formula, filters, attribution window and currency logic.
- What is excluded? VAT, sales tax, cancelled orders, returns, coupons, organic sales, vendor chargebacks or marketplace-funded discounts.
- Who is allowed to use it for decisions? Account manager, retail media lead, client finance, founder or marketplace operator.
- What happens when two systems disagree? Decision hierarchy and reconciliation owner.
That last question is the expensive one. Marketplace work has normal disagreement built in. Amazon Ads attribution will not match settlement cash. Seller Central business reports will not match accounting revenue. TikTok Shop GMV will change as refunds arrive. Walmart item performance will not always agree with a finance export. If the agency waits until the QBR to decide which number is “real”, the meeting becomes a courtroom.
FiveX helps here by connecting marketplace, advertising, inventory and profitability data into one operating view. But the bigger point is not “one dashboard to rule them all”. It is one agreed logic layer, so the dashboard can support decisions instead of starting debates.
Scenario 1: the Amazon “sales” argument that steals the QBR
Imagine a US outdoor brand managed by a six-person marketplace agency. The client sells on Amazon and Shopify. In August, Amazon Ads reports $92,000 in attributed sales on $18,400 spend, so the campaign view shows a clean 20% ACOS. The account manager prepares a slide that says retail media is healthy.
Finance pushes back. Amazon settlement shows only $71,500 net payout for the same product family. There were $8,900 in FBA fees, $6,200 in referral fees, $4,600 in returns and $2,300 in coupon funding. The client’s question is fair: “If sales were $92,000, why did cash look like $71,500 before landed cost?”
Without a metric dictionary, the agency explains from memory. “Ads sales are attributed revenue, settlement is net, returns lag, coupons are separate.” True, but it sounds defensive. The conversation moves from strategy to trust repair.
With a metric dictionary, the slide says:
- Ad-attributed sales: Amazon Ads, 14-day attribution, gross order value, used for media efficiency only.
- Marketplace net revenue: settlement-based sales after Amazon fees, refunds and seller-funded discounts, used for profit review.
- Contribution margin after ads: net revenue minus landed cost and ad spend, used for budget permission.
The agency can then show the real decision: ACOS looked acceptable, but contribution margin after ads was only $7,800 on the product family. That supports a practical recommendation: keep branded defence live, cap generic discovery at $4,000 next month, and move $3,500 into two higher-margin SKUs with 34% contribution before ads.
FiveX hook number one: connect Amazon Ads, settlement-style profitability, product margin and SKU-level cost assumptions so the agency can show both the media view and the profit view without rebuilding the maths in a spreadsheet.
Scenario 2: the German marketplace “TACoS” number that hides stock pressure
Now take a German homeware client selling on Amazon.de, Kaufland and Otto. September revenue across the three channels is €186,000. Retail media spend is €22,300, so the agency reports blended TACoS of 12.0%. The client likes the number. Growth looks controlled.
But the product family is not controlled. Amazon has 21 days of stock. Kaufland has 8 days. Otto has 46 days because replenishment arrived earlier. A single blended TACoS number encourages the team to scale the best-looking campaigns, which are mostly Kaufland search ads. That is the exact channel with the thinnest stock cover.
The metric dictionary forces the right split:
- Portfolio TACoS: total ad spend divided by total marketplace sales, used for board-level trend only.
- Channel TACoS: spend divided by channel sales, used for budget allocation.
- Stock-adjusted TACoS permission: channel TACoS plus stock cover rule, used for daily spend permission.
The operating rule is simple: if stock cover drops below 14 days, prospecting campaigns cannot scale, even if TACoS is below target. In this case, Kaufland prospecting pauses at €420/day, Otto receives an extra €180/day for two weeks, and Amazon keeps branded protection because Buy Box share is stable.
FiveX hook number two: advertising automation should not only read ROAS or ACOS. It should read stock cover, channel mix, SKU margin and marketplace constraints before allowing budget to move.
Scenario 3: the TikTok Shop GMV story that becomes unpaid strategy work
A TikTok Shop test creates €38,000 GMV in ten days for a beauty client. The agency has a monthly retainer of €7,500 and a scoped reporting cadence of one monthly performance review. Suddenly the client wants three extra calls: one about creator commission, one about Amazon halo, one about whether to reorder 4,000 units.
If “GMV” is undefined, the agency absorbs the thinking. The social team celebrates demand. The marketplace team worries about returns. Finance asks whether TikTok vouchers were platform-funded or seller-funded. The account lead spends six unplanned senior hours aligning everyone.
The metric dictionary turns the conversation into scope and evidence:
- TikTok GMV: order value before returns and settlement deductions, used as a demand signal.
- Settled TikTok revenue: paid-out revenue after refunds, fees and vouchers, used for profitability.
- Cross-channel halo: change in branded Amazon and Shopify demand versus baseline, marked as directional unless incrementality test is active.
- Decision trigger: reorder decisions above €25,000 expected landed cost require a paid profit brief or pre-agreed automation workflow.
That is not bureaucracy. It protects agency margin and client money at the same time. FiveX hook number three: AI recommendations and reporting workflows can draft the profit brief from connected TikTok, Amazon, Shopify, ad spend, inventory and margin data, but the metric dictionary decides which numbers the AI is allowed to use and how confident the recommendation may be.
How to build a marketplace agency metric dictionary
1. Start with the ten metrics clients argue about
Do not try to document everything on day one. Start with the metrics that create friction: sales, net revenue, gross margin, contribution margin, ACOS, ROAS, TACoS, ad-attributed sales, stock cover and return rate. Add marketplace-specific metrics later: Buy Box share, organic rank, retail readiness, listing suppression, refund lag, settlement variance and creator commission.
2. Create a source hierarchy
For every commercial decision, name the winning source. For example: Amazon Ads wins for campaign optimization, settlement data wins for cash reconciliation, FiveX profitability wins for SKU budget permission, ERP landed cost wins for product cost, and client finance wins for final booked revenue. This avoids the classic Monday argument where every department brings its favourite export.
3. Separate reporting metrics from decision metrics
Some metrics are useful for storytelling but dangerous for action. Portfolio TACoS is good for trend conversations. It is too blunt for daily bid moves. GMV is good for demand detection. It is too early for margin approval. Ad-attributed sales are useful for media diagnosis. They are not the same as profit.
4. Add caveats directly into dashboard labels
If a number excludes returns, say so next to the number. If attribution is 14 days, put it in the label. If stock cover uses the last 14 days of unit sales, make that visible. Hidden caveats create avoidable Slack archaeology.
5. Give every disputed metric an owner
A dictionary without ownership becomes polite documentation. The retail media lead owns ACOS and campaign attribution. The marketplace operations lead owns stock cover and listing health. Finance owns settled revenue and landed cost. The account director owns the client-facing explanation when definitions change.
6. Review definitions when the client changes channel mix
A client that only sells on Amazon can survive with simpler definitions. Add Walmart, TikTok Shop, Kaufland, Otto, bol.com or Mirakl retailers and the old definitions start creaking. Any new marketplace, fulfilment model, currency, agency scope item or paid media channel should trigger a metric dictionary review.
The trade-off: slower setup, faster trust
The objection is obvious: building a metric dictionary takes time. True. A good first version can take three to six hours per agency template, plus one client-specific hour during onboarding. That feels expensive when the team is busy.
But compare it with the hidden cost of definition drift. One disputed QBR can burn four senior hours. One wrong budget decision can waste €2,000 in retail media before the next weekly review. One unclear profitability report can turn a renewal call into a forensic accounting session. The dictionary pays back when it prevents even one of those moments.
The operator move is not to make reporting heavier. It is to make the default explanation stronger. When a client asks, “Why is this number different?”, your team should not improvise. They should point to the agreed definition, show the reconciliation path and move the conversation back to the decision.
What to put in your agency software
A practical metric dictionary should live where the work happens. In FiveX, that means connecting the definitions to dashboards, profitability views, alerts, AI-generated commentary and client reporting packs. The minimum useful setup is:
- A metric name and client-friendly description.
- Source system and refresh cadence.
- Formula, filters, currency and attribution window.
- Decision use: reporting, optimization, budget permission, finance reconciliation or renewal evidence.
- Known caveats and confidence level.
- Owner and escalation path.
- Last reviewed date.
That structure gives agencies something competitors rarely talk about: repeatable trust. New account managers onboard faster. Clients get fewer conflicting answers. AI reporting becomes safer because the model uses approved definitions. Automation gets better guardrails because “good performance” is tied to the right commercial metric, not the most convenient one.
Final thought
Dashboards make marketplace work visible. Metric dictionaries make it governable.
For marketplace agencies, that distinction matters. Your team is not only reporting Amazon, Walmart, bol.com, Kaufland, Otto, TikTok Shop or Shopify performance. You are translating messy channel data into commercial decisions clients can trust. The more channels you manage, the more that trust depends on definition discipline.
So before buying another reporting add-on, ask a less glamorous question: can every person on the account explain what this number means, where it came from, what it excludes and whether it is allowed to approve spend?
If the answer is no, you do not have a dashboard problem. You have a dictionary problem. Fix that first. FiveX can help by turning connected marketplace, advertising, inventory and profitability data into a shared operating view — with the definitions, alerts and AI-assisted reporting needed to keep agency recommendations clear, defendable and profitable.