By Ted Gibson — RippleEffect
The Amazon agency red flags worth acting on are the checkable ones: a price you cannot learn without sitting through a sales call, a fee that rises with your ad spend and no clause covering what happens when spend should come down, a guarantee of rank or revenue, reporting built on advertising sales rather than total sales, and any arrangement that leaves your seller account or your brand registry in the agency's name. The softer signals tell you very little. A slick deck and a wall of logos are the cheapest things an agency owns, and the published red-flag guides are at their least reliable on the one test they all reach for.
The staffing test everyone quotes, and nobody agrees on
Every guide on the subject asks how many accounts each manager carries, and then gives a different answer. My Amazon Guy says ideally 8 to 10 or fewer. SupplyKick calls 10 to 20 brands per manager healthy. BrandGrowthIQ treats junior staff "often spread across 15-30 client accounts simultaneously" as the warning sign. Adverio draws the line at agencies with "50 or more clients per account manager". Same question, same market, the same year, and the thresholds sit six times apart end to end. None of the four publishes a method, so the ratio on its own settles nothing.
The version of the question that does work is specific and hard to dodge. Who does the daily work on my account, what is their name, how many other accounts are they carrying this quarter, and how much of that work did the person sitting in front of me do personally? An agency that answers with a name and a number has given you something you can hold them to in six months. An agency that answers with "our team" has avoided the question. Here the work is founder-led, so the person on the first call is the person who runs the channel, which makes that an easy question for us and is exactly why it is worth asking everyone.
Almost every number in this category is an agency's number about itself
Look at what the red-flag guides use as evidence. My Amazon Guy publishes a 95% client retention rate, more than $1.2 billion in revenue managed and over 400 brands supported. SupplyKick states $100m or more in Amazon revenue annually with 96% client retention. Marknology gives its retention as 93%. Each of those is self-published and unaudited, and each is used to imply a benchmark the reader should measure other agencies against. We could not find a neutral source anywhere in this market, no survey and no trade body, publishing what normal retention or normal churn actually looks like, which means a retention rate is a claim rather than evidence. What you can check is whether the numbers an agency prints are attached to a named client who has agreed to stand behind them, which is why ours carry the names.
On fees, the structure is the flag rather than the number
A percentage of your ad spend is the most common model and the one whose incentive points furthest from yours. BrandGrowthIQ puts it at typically 8 to 15% of monthly spend, which on a £10,000 monthly ad budget is £800 to £1,500, and on $10,000 is $800 to $1,500. The percentage is not the problem. The missing clause is what happens to the agency's income in the month when the right move is to cut spend by a third, and if nobody can answer that in one sentence the model has answered for them. The three charging structures and what each one rewards are set out in a separate post. Ours is published in full, including the one-off onboarding fee, because an agency that will not name its one-off costs before you sign has already decided which of you the surprise is for.
Reporting you cannot check yourself
Two tests here. First, does the reporting lead with advertising sales or total sales? Advertising sales flatters whoever bought the advertising, and an account can post a beautiful ACOS while the total goes sideways, which is the same trap that makes branded campaigns look like your best performers. Second, can you see the number they bill you on, inside your own account, without asking them for it? Our commission is calculated on Ordered Product Sales as reported in Seller Central Business Reports, a figure you can pull yourself on a Sunday evening. A billing base that exists only in the agency's spreadsheet is a red flag however reasonable the percentage looks.
Then the contract, which is where the real exposure sits: the initial term and what rolls after it, the notice period, what happens to your data and your access on the day you leave, and whose name is on the seller account and the brand registry. Get that last one wrong and the argument is not about a refund, it is about who owns the shop. Ours runs twelve months and then rolls, and SKU bands are reviewed quarterly and move down as well as up. An agency writing about agency red flags is its own small warning sign, so the fair way to read this is as a checklist to run on us: the pricing is on the website, the case studies are named, and every engagement opens with a free account audit you can judge before you have paid for anything.