Agency client onboarding for marketplace ads: the 90-day profit baseline

Agency client onboarding for marketplace ads is usually treated like an access checklist: get Amazon Ads, get Seller Central, get the product feed, schedule the kickoff, launch the first campaign. That looks efficient on a project plan. It is also the fastest way to inherit somebody else's margin problem.

Marketplace advertising is not Google Ads with product pages attached. The ad account is only one piece of the operating system. If an agency starts bidding before it understands contribution margin, stock cover, fulfilment fees, return rates and marketplace-specific constraints, the first month can look successful in ROAS while quietly making the client poorer.

The best onboarding process for a marketplace ads client therefore has one job: build a profit baseline before you scale activity. Access matters. A friendly kickoff matters. But the real difference between a professional marketplace agency and a busy campaign shop is whether the team can answer this question in week one:

Which SKUs are allowed to receive incremental ad spend, which SKUs need protection, and which SKUs should not be advertised at all right now?

That is the angle most generic onboarding advice misses. They focus on portals, forms and first impressions. Useful, yes. But for agencies managing Amazon, bol.com, Walmart, Mirakl or retail media accounts, onboarding is where profit guardrails are created. Skip that work and your specialists spend the next 90 days explaining why a campaign with a 4.8 ROAS still disappointed the finance team.

The named mistake: launching before the SKU margin map exists

The mistake has a name inside many agencies, even if nobody says it out loud: the inherited campaign trap.

A new client signs. They already spend €18,000 per month on Amazon Ads across 240 campaigns. Sales look decent. The client wants quick wins. The agency imports the account, pauses obvious losers, increases bids on strong ROAS campaigns and sends a positive first update. Everyone feels momentum.

Then the margin file arrives two weeks late. Suddenly the account looks different. The hero product with 6.2 ROAS has only 17% gross margin after Amazon referral fees and FBA. Returns sit at 11%. The break-even ACOS is not 22%, as the old report assumed, but closer to 12%. The agency has just pushed more budget into the one SKU that looked good in the ad dashboard and bad in the P&L.

That is not a tactical error. It is an onboarding error.

Marketplace agencies should never judge inherited campaigns only by ACOS, ROAS or revenue. During onboarding, every active advertised SKU needs a simple status:

• Scale: contribution margin is healthy, stock cover is sufficient and the marketplace is strategically important.
• Fix first: advertising may work, but content, pricing, Buy Box, feed errors or stock issues limit performance.
• Protect: keep only defensive or branded coverage because margin or inventory is fragile.
• Stop: do not advertise until the economics change.

FiveX helps agencies create that view by connecting marketplace, advertising, stock and profitability data in one workspace. Instead of waiting for three spreadsheets and a late finance export, the team can see product-level margin next to ad spend and sales from the start.

The 90-day onboarding system for marketplace ads

A strong onboarding process does not need to be complicated. It needs to be sequenced correctly. The order matters: commercial truth first, campaign changes second, scale third.

Days 0-7: access, data integrity and the first profit snapshot

The first week is not for clever bidding. It is for establishing what the agency is allowed to trust.

Collect access to Amazon Ads, Seller Central or Vendor Central, the relevant marketplace accounts, feed tooling, analytics, Shopify if it exists, and any BI or finance export the client uses. For European clients, include bol.com, Kaufland, Otto, MediaMarkt or Mirakl-based marketplaces where relevant. For US clients, add Walmart Marketplace and retail media networks when they are part of the scope.

But access is only the surface. The agency should validate four datasets before making material changes:

1. Advertising data: spend, sales, ACOS, ROAS, campaign structure and search term history.
2. Marketplace sales data: ordered revenue, units, refunds, commissions, fulfilment fees and promotions.
3. Product economics: COGS, landed cost, gross margin, contribution margin and minimum acceptable profit per unit.
4. Operational constraints: stock cover, lead time, Buy Box ownership, listing quality and suppressed products.

The first deliverable should be a one-page profit snapshot, not a 40-slide audit. Show the client which products currently receive spend, which products actually make money, and where the two do not match.

Example: A German home & kitchen brand spends €12,400 per month across Amazon.de and Kaufland. The inherited Amazon report shows 24% ACOS and €51,700 attributed sales. After mapping COGS, FBA fees and returns, the agency finds that €4,100 of the spend goes to five SKUs with less than €1.50 contribution margin per unit. Pausing those SKUs and moving only €2,800 to three higher-margin bundles protects roughly €1,300 in monthly profit without reducing total ad visibility much. That is a week-one win a client can feel.

Days 8-30: restructure only where the business case is clear

The second phase is where many agencies overdo it. They rebuild naming conventions, split campaigns, add negatives, change bidding rules and adjust budgets all at once. The client sees activity, but nobody can isolate what worked.

Better: create a controlled change list. Every campaign change should have a reason connected to one of four outcomes:

• protect profitable base sales;
• reduce spend on negative-margin demand;
• increase share on SKUs with stock and margin room;
• collect clean data for a decision the team cannot make yet.

For an agency with multiple specialists, this is where software discipline matters. A senior strategist may set the commercial guardrails, while a PPC specialist makes campaign changes and an account manager prepares client communication. If those people work from different exports, small errors become expensive. One person sees ACOS, another sees stock, and nobody sees that the product will be out of stock in nine days.

FiveX gives agencies a shared operating layer for those decisions: ad performance, inventory signals, product profitability and AI recommendations in one place. The hook is not “more dashboards”. Agencies have enough dashboards, bless them. The hook is fewer unsafe decisions.

Days 31-60: build the client rhythm around decisions, not screenshots

By month two, onboarding should shift from setup to rhythm. This is where the account becomes manageable at agency scale.

A good weekly marketplace ads meeting should not walk through every chart. It should answer five questions:

1. Which SKUs earned more budget this week?
2. Which SKUs lost permission to spend?
3. Which campaign changes improved contribution margin, not only ROAS?
4. Which stock or Buy Box issues blocked growth?
5. Which client decision is needed before the agency can move?

This rhythm protects both sides. The client sees how decisions are made. The agency avoids becoming a reporting factory. And the account manager has a clear story: here is the money we protected, here is the growth we unlocked, here is what we need from you.

Example: A US supplement brand spends $32,000 per month across Amazon Sponsored Products and Walmart Sponsored Search. The old report celebrates a blended 5.1 ROAS. During onboarding, the agency separates products into replenishable hero SKUs, trial-size acquisition SKUs and low-margin multipacks. The trial-size SKU runs at 38% ACOS but creates repeat orders within 45 days, so the agency keeps it live with a capped budget. The multipack runs at 19% ACOS but loses $0.80 per order after fees and coupons, so the agency stops non-brand spend. Same ad account. Completely different decisions.

Days 61-90: turn the baseline into scalable rules

By the third month, the agency should have enough evidence to automate parts of the operation. Not everything. Marketplace advertising still needs judgement, especially around promotions, stock delays, price changes and competitor moves. But repeatable guardrails should become rules.

Examples:

• Pause non-brand campaigns when stock cover drops below 14 days.
• Reduce bids when contribution margin falls below €3 per unit after returns.
• Increase budget only when ACOS is below break-even and sell-through supports the extra demand.
• Flag campaigns where ad-attributed sales rise but total marketplace sales do not.
• Review SKUs with strong ROAS but declining Buy Box ownership.

This is where FiveX's AI recommendations and advertising automation become useful for agencies. The system can surface anomalies, margin risks and optimisation opportunities across clients, while the specialist keeps control over the final action. That balance matters. Automation without profit context is just a faster way to make the wrong decision.

What competitors cover well, and what agencies still need

Generic client onboarding tools are good at reducing admin. They help agencies collect access, standardise questionnaires, create portals and make the first client experience feel polished. That matters. If a client waits three weeks for the right person to grant Google or Meta access, trust erodes before the work begins.

Marketplace platforms and feed tools cover another important part: product data, channel setup, listing quality and feed automation. That also matters. Bad product data can kill ad performance before a bid is placed.

The gap is the middle layer: the commercial operating model for marketplace ads. Agencies need onboarding that connects access, feed readiness and campaign structure to SKU-level profit. Without that layer, the team can still look organised while making unprofitable decisions.

So the practical stack is:

• Access collection: make permissions simple and secure.
• Project workflow: keep tasks, owners and deadlines visible.
• Feed and listing quality: fix the products that cannot convert.
• Marketplace profit intelligence: decide where spend is allowed.
• Ad execution: optimise campaigns inside those guardrails.

FiveX sits in the fourth and fifth layers for marketplace agencies: profitability dashboards, ad automation, AI recommendations, repricing and inventory insights connected to marketplace performance. That is why it fits the onboarding moment so naturally. It helps the agency move from “we have access” to “we know what should happen next”.

The client-facing onboarding checklist

If you want a practical version, use this checklist before the first major optimisation sprint:

• Confirm commercial goals: profit, revenue, market share, launch, liquidation or defence.
• Confirm marketplace scope by country, account type and retail media network.
• Collect access without shared passwords.
• Map COGS, fees, returns, promotions and fulfilment costs per SKU.
• Calculate break-even ACOS or minimum ROAS by product group.
• Check stock cover and replenishment lead times.
• Classify advertised SKUs as scale, fix first, protect or stop.
• Audit campaign structure only after the SKU map exists.
• Create the first 30-day change log with expected business impact.
• Agree the weekly decision rhythm and escalation rules.

Notice what is missing: “optimise everything”. That is not an onboarding strategy. That is a caffeine habit with a login.

The final test: can the client explain the logic?

Good onboarding does not end with a neat dashboard. It ends when the client can explain the agency's operating logic back to you.

They should understand why one campaign was scaled while another was capped. They should know why a high-ROAS product was paused. They should see the connection between stock, margin and ad spend. And they should feel that the agency is protecting their business, not just spending their budget.

For marketplace agencies with five or more employees, that is the difference between onboarding as admin and onboarding as a profit system. The first creates tasks. The second creates trust.

And trust, very conveniently, renews.