This is the companion piece to “How to Evaluate Whether Your Amazon Agency Is Actually Working.” Part One asked whether the agency is doing the right work. Part Two asks whether the conditions exist for any agency to do the right work at all.
How to Make Your Amazon Agency Really Work
Part One of this series offers a framework for asking the hard questions of your Amazon agency: contribution margin per ASIN, incrementality, organic rank trajectory, the hypothesis being tested. That piece assumed something this piece questions: that the agency is the variable in the equation. Sometimes it is, but often it isn’t. And brands that switch agencies repeatedly without ever fixing the underlying pattern they’re experiencing are often the ones that never turned the diagnostic inward.
There are two failure modes that look identical from the outside. The first involves an agency that is genuinely better at sales and storytelling than at operations. This often looks like polished onboarding and beautiful decks, but declining performance month to month. The second is an agency doing competent work inside a relationship that doesn’t let that work succeed: stock-outs the agency couldn’t see coming, financial planning the agency wasn’t looped into, a buy plan the brand approved and then blamed the agency for executing, channel hygiene so porous that the buy box belongs to whoever felt like undercutting that week. Both produce poor outcomes, but only one is the agency’s fault.
Many established brands that have churned through two or three Amazon agencies have hired tactical when they needed strategic; the failure mode named in Part One. But a meaningful subset of them have done something else: they have built an internal environment in which no agency, however strategic, could have produced the outcomes the brand expected.
This piece looks at a harder conversation. How to tell the difference between an agency that is failing at execution and one that is being prevented from executing. Three diagnostics, applied to your own organization, not the agency, will help surface whether the partnership has the conditions it needs to thrive. Plus a frank discussion of agency economics: what a real Amazon team costs, and what kind of results a given spend level can credibly produce.
Is the Agency Bad at Operations, or Is the Relationship Bad at Letting Them Operate?
Some Amazon agencies are better at selling the project than running it: polished onboarding, with weak operational follow-through, and declining results once novelty fades. But many underperforming engagements involve agencies doing the right kind of work inside relationships that are doomed to fail: missing inventory the agency can’t forecast, financial planning the agency isn’t included in, and channel pricing chaos the agency has no authority to fix.
The sales-strong, ops-weak agency
This is a real pattern, and Part One describes its signature: massive slide decks, elaborate color-coded dashboards, ACOS trending in the right direction while the P&L goes the other. The agency wins the project on storytelling and loses it on operations. There is no shame in naming it when it’s happening, and the six questions from Part One are designed to surface it quickly.
The mirror pattern
But there is a second pattern that produces the same symptoms and almost never gets discussed. The agency is doing good work. The brand approves a buy plan in March and runs out of inventory in May… and the agency catches the blame for sales that fell off when stock did. The brand’s finance team commits to a Q4 promotional cadence in October without telling the agency until the campaign launches… and the agency is held responsible for traffic the listing was never resourced to convert. The brand’s wholesale team agrees to a price concession on a major retailer that drops the listed price below Amazon’s… and the agency is asked why the buy box keeps disappearing.
None of these are agency failures but all of them produce metrics that look like agency failures.
Three Diagnostics That Tell You Whether You’re Actually a Good Client
Three assessments, applied honestly to your own brand, surface whether your Amazon agency is set up to succeed: whether you’re asking real questions and demanding answers grounded in Amazon data rather than another dashboard, whether you’re operating as a true partner on inventory and financial planning and channel hygiene, and whether you’re paying enough to make competent execution economically possible. Two or three of these failing is the most common reason established brands rotate through agencies without improving outcomes.
Each of the next three sections is structured the same way: what the diagnostic is, what good looks like, what bad looks like, and a short note on the consequences of getting it wrong. Read them as a self-audit. If the answers concern you, your Amazon agency is probably not the variable that needs to change first.
Diagnostic One: Are You Asking the Hard Questions and Validating the Answers Against Amazon Data?
A good agency client asks the hard questions and refuses to accept “look at this dashboard” as the answer. The test is whether the response reconciles to Amazon data (Seller Central or Vendor Central reports, Brand Analytics, advertising console exports) rather than to a tool the agency built or a dashboard that sources from another dashboard. Many established brands stop one layer too early in the chain of evidence and accept narrative for proof.
The dashboard-about-a-dashboard problem
Most Amazon agencies have invested heavily in client-facing dashboards. That is not a bad thing, visualization is genuinely useful. Plus, the alternative (raw CSV exports) doesn’t scale to executive review. The risk is that the dashboard becomes the answer rather than the summary of an answer. The question “why is contribution margin down on these six SKUs?” gets answered by pointing at a dashboard panel labeled “contribution margin trend.” That isn’t an answer. It’s the same question, rendered visually.
The hard version of the question is: what specifically changed, in which Amazon data source, that produced the trend you’re showing me? Was it a fee change visible in the FBA fee preview report? A return rate spike visible in the FBA customer returns report? A buy box loss visible in the Brand Analytics Item Comparison report? A spend reallocation visible in the bulk advertising download? Real answers cite the source. Dashboard-only answers cite the dashboard.
What good looks like
A strong client asks the question, listens to the agency’s answer, and then asks where in Amazon’s data the answer can be verified. A strong agency responds by naming the report, the date range, and the specific rows or columns that support the claim, and offers to walk through the source data rather than just the summary view. The exchange is collaborative: the client isn’t trying to catch the agency, and the agency isn’t defensive about being asked to show its work.
What bad looks like
The bad version takes one of two forms. In the first, the client never asks the harder question. The dashboard panel feels like an answer, the meeting moves on, and three months later contribution margin is still sliding and nobody has named why. In the second, the client asks the harder question but accepts a reframing of the same dashboard as validation. “Here’s the contribution margin view broken out by ASIN” is not validation of a contribution margin claim. It’s the same answer at finer granularity. The validation is the underlying Amazon data the dashboard was built from.
Why it matters
An agency operating without source-data discipline will eventually produce a narrative that drifts from reality, because the dashboard is a model of Amazon’s data and models develop blind spots. A client operating without source-data discipline has no way to catch that drift until the P&L makes it impossible to ignore. At this point the agency’s effectiveness has usually been declining slowly but surely for two or three quarters. The clients that really get value from their agencies ask for the source citation as a matter of course, and agencies that thrive alongside their clients welcome it.
Diagnostic Two: Are You Operating as a True Partner, Particularly on Inventory, Financial Planning, and Channel Hygiene?
The three operational responsibilities that brands most frequently fail at (and blame their agency for) are keeping Amazon in stock, looping the agency into financial planning early, and managing pricing and channel conflict across non-Amazon sales channels. These items are all outside the scope of most agency contracts. But each of them, when neglected by the brand, produce metrics that appear as if the agency has failed.
Inventory: the buy plan you approved is now your responsibility
A pattern repeats across underperforming engagements. The agency builds a demand forecast, presents it as part of a quarterly buy plan, and the brand approves it. Two months later, faster-than-expected sell-through on three SKUs leads to a stock-out the brand’s internal team didn’t flag… because nobody on the brand side was responsible for monitoring the agency’s forecast against actual sales velocity. Revenue falls. Organic rank slips while inventory is unavailable. The agency is asked why performance dropped.
The agency built the forecast. The brand approved it, owns the supply chain, and has visibility into manufacturing lead times the agency does not. When a stock-out happens, the question isn’t whether the forecast was wrong; forecasts are wrong by definition. The question is whether anyone on the brand side was watching the gap between forecast and actuals in real time, and whether the brand has a process for telling the agency early so paid spend can be ramped down, listings repositioned, and the loss contained rather than compounded.
A good client either keeps Amazon in stock or communicates clearly when they won’t. They treat their Amazon agency as a partner in the inventory conversation rather than a downstream consumer of decisions made elsewhere.
Financial planning: loop the agency in before the plan is final
The second most common partnership failure is treating the agency as the executor of decisions made without their input. The CFO sets a Q4 marketing budget in August. The merchandising team commits to a holiday cadence in September. The agency hears about both in October when they’re asked to execute. Every constraint that should have been negotiated when there was time to negotiate it becomes a done deal that the agency is now responsible for making work.
The agency should be in the room (or at minimum, on the call) for financial planning that affects their scope. Not because they should set the budget, but because they have a perspective on what a given budget can credibly produce. Their input may be wrong, and the brand may overrule it. That is fine. What is not fine is hiring an agency for its expertise and then asking that expertise to validate decisions it had no chance to inform. Cluing the agency in early is one of the highest-leverage moves an Amazon client can make. It costs nothing and dramatically changes the quality of execution.
Channel management: buy box is won and lost outside the Amazon conversation
The third operational failure that gets misattributed to the agency is porous channel management. The brand’s wholesale team agrees to a price concession on a major retailer. A direct-to-consumer flash sale drops the brand’s own website below MAP. A reseller buys excess inventory at distributor pricing and lists it on Amazon below the brand’s offer. Buy box disappears, conversion drops, paid advertising stops working because the offer isn’t winning the buy box anymore. The agency is asked why the Amazon business is suddenly underperforming.
The agency has no authority over any of these decisions. Most agencies don’t even have visibility into them in real time. A brand that hasn’t built channel management discipline (clear MAP policy, enforced reseller controls, communication between the Amazon team and every other channel’s pricing decisions) is asking its Amazon agency to compete in a fight they can’t enter. That is the brand’s job. When it isn’t being done, the metrics that suffer first are the metrics the agency is on the hook for.
Why it matters
Each of these three failures is invisible in the agency report. The agency sees the result (declining velocity, slipping rank, buy box loss) and reports on it accurately. The brand sees the report and concludes the agency is failing. Both sides are looking at real data but neither is looking at the cause. Without partnership on inventory, financial planning, and channel hygiene, the whole engagement is structured to produce the exact symptoms that lead to an agency being replaced… by another agency that will encounter the same conditions and produce the same symptoms.
Diagnostic Three: Are Your Expectations Calibrated to What You’re Actually Paying?
Amazon agency expectations should be calibrated to the economics of running a competent Amazon team. A high-quality in-house ecommerce manager earns well into six figures, and a competent Amazon function typically requires a manager plus support. This is easily $200,000 to $400,000 a year in fully loaded cost. Paying an agency less than $3,000 a month means buying a fraction of one capable person’s time, which produces either slower results or visibly poor results quickly. The expectation gap is one of the most common reasons brands feel their agency is underperforming.
The economics of a real Amazon team
A capable in-house Amazon manager (the kind who can run advertising, manage listings, handle catalog hygiene, and produce a credible monthly review) earns six figures in most markets, often $120,000 to $180,000 with experience. Add a coordinator or specialist for execution support, and the all-in cost of a minimal internal Amazon function lands between $200,000 and $400,000 a year before tools, training, or the senior oversight strategic decisions actually require.
That is what is being substituted for when a brand hires an agency. An Amazon agency is renting expertise to brands that do not hire in-house. The math of agency pricing reflects this: the agencies that produce strategic-partner-level work are pricing their engagements against the cost of the team they are replacing, not against a discounted version of it.
What sub-$3,000-a-month buys
At under $3,000 a month ($36,000 a year) the brand is buying a small fraction of one capable person’s time, spread across the agency’s other clients. That is a real service, and there are competent agencies operating at that price point. But the realistic deliverable is execution maintenance… keeping campaigns running, keeping listings live, responding to clear questions… not strategic diagnosis, not forensic margin analysis, not the multi-month research that produces structural change. Expecting the latter from the former is the gap that drives most “my agency isn’t producing results” conversations.
The honest read on a sub-$3,000-a-month engagement is that it will either move slowly or fail visibly. Both are predictable outcomes of the price point but neither is a definite failure of the agency.
The mid-tier reality
Most brands paying $5,000 to $15,000 a month are buying meaningful agency attention. This is enough to support real execution, with some strategic capacity if the agency is set up for it. That is the price tier where the difference between tactical and strategic agencies matters most: the brand is paying enough that mismatch is expensive, but not so much that the agency’s economics force a particular operating model. The diagnostics from Part One are most useful here.
The strategic engagement reality
Engagements that include real diagnostic work like forensic margin reconciliation, advertising incrementality testing, multi-month research before significant tactical intervention, typically price north of $15,000 a month, often substantially more. That price reflects the senior time required to do the work credibly. Brands that require this level of work but are unwilling to pay for it usually end up cycling through cheaper engagements that produce activity reports instead. The cost of three failed cheap engagements often exceeds the cost of one engagement priced honestly.
Why it matters
Calibrating expectations to economics is the cheapest form of partnership discipline available. It costs the brand nothing to know what their spend is realistically buying. It saves an enormous amount of grief — both the agency’s and the brand’s — when the engagement is set up with a shared understanding of what good performance looks like at that price point.
The Honest Diagnostic Is Bilateral
When an Amazon engagement isn’t working, the honest diagnostic is bilateral. The agency may be better at sales than at operations. Part One was written for you to surface that case quickly. Or the brand may be running an environment in which no agency could succeed: dashboards accepted as evidence, financial decisions made without agency input, stock-outs treated as agency failures, channel chaos blamed on Amazon performance, expectations set against pricing that cannot fund them. The brands that get sustained value from their agency relationships apply the same forensic rigor to both sides of the table.
Our Second-Opinion Review is built around exactly that bilateral read. It is not just an assessment of the current agency. It is an assessment of the partnership: the agency’s work, the conditions the brand has created for that work, and what would need to change on either side for the engagement to produce what the business needs.
Other articles that may interest you:
Why Your Amazon Business Is Bleeding Money — And Why Your Last Agency Didn’t Find It
Your Amazon Ads Are Training the Algorithm — Make Sure They’re Teaching It the Right Lessons
How to Evaluate Whether Your Amazon Agency Is Actually Working
Frequently Asked Questions
How do I make my Amazon agency work for me?
Three disciplines determine whether any Amazon agency can produce results: asking hard questions and demanding answers grounded in actual Amazon data rather than dashboards summarizing other dashboards, operating as a true partner on inventory and financial planning and channel hygiene rather than treating the agency as a downstream executor of decisions made without them, and calibrating expectations to what you are actually paying. A brand that gets all three right will outperform with most competent agencies. A brand that gets two or three wrong will struggle with any agency, no matter how strong.
Why isn’t my Amazon agency producing results?
Underperformance usually comes from one of two patterns. Either the agency is better at sales and storytelling than at operations — the failure mode Part One of this series surfaces with six diagnostic questions — or the agency is doing competent work inside a relationship that was structurally set up to fail. Stock-outs, financial planning the agency wasn’t included in, and channel pricing chaos all produce metrics that look like agency underperformance but originate on the brand side.
Is it the agency’s fault when Amazon runs out of stock?
It is the agency’s responsibility to build a credible demand forecast and flag inventory risk early. It is the brand’s responsibility to approve the buy plan, execute the supply chain, and monitor actual sales velocity against forecast in real time. When a stock-out happens, the brand cannot reasonably hold the agency accountable for the revenue impact unless the agency failed to warn — but the brand can and should hold itself accountable for whether the partnership had a process for catching the gap before it became a crisis.
Should my Amazon agency be involved in financial planning?
The agency should not set the budget, but they should be in the room when the budget is set — or at minimum, looped in before it is finalized. Hiring an agency for its expertise and then asking that expertise to execute decisions it had no chance to inform is one of the most common partnership failures in Amazon engagements. Cluing the agency in early on financial planning costs the brand nothing and dramatically improves the quality of execution.
How does channel management affect my Amazon agency’s performance?
Buy box is won and lost outside the Amazon conversation. When wholesale pricing concessions, direct-to-consumer discounting, or reseller activity push the brand’s offer below the buy box threshold, conversion drops and paid advertising stops working — regardless of how well the Amazon agency is executing. Porous channel management is one of the most common causes of metrics that look like agency underperformance, and the agency typically has neither the authority nor the visibility to fix it.
How much should I pay an Amazon agency?
A capable in-house Amazon manager earns well into six figures, and a real Amazon function typically costs $200,000 to $400,000 a year all-in. Agencies are pricing against this cost. Sub-$3,000-a-month engagements buy a fraction of one capable person’s time and reasonably produce execution maintenance, not strategic diagnosis. Mid-tier engagements at $5,000 to $15,000 a month buy meaningful attention. Engagements that include forensic margin work and incrementality testing typically price north of $15,000 a month.
How do I tell if I’m the problem in the agency relationship?
Three self-audits surface it. Are you accepting dashboards as evidence rather than reconciling claims to underlying Amazon data? Are you treating the agency as a downstream executor on inventory and financial planning rather than a partner in those decisions? Are your performance expectations calibrated to what you are paying, or to what you wish you were paying? If two or three of those concern you, the agency is probably not the variable that needs to change first.