How to Choose an Amazon Agency: What Full Service Must Include, and Who Should Not Hire One

Search for an Amazon agency and the first page is lists: the ten best, the top agencies of the year, six compared. They are useful for names and useless for the decision, because the decision is not which agency is best in general. It is which one will run your account, in your category, at your stage, and be answerable for what moves. This article is the method rather than the list: what to decide before you talk to anyone, the seven things to score every candidate on, the questions that sort the operators from the advisers, and the cases where the honest answer is not to hire one.

It is written by an agency, so read it as one. Where a criterion favours how we work, it says so, and you can weigh it accordingly. The rubric is the one we would use if we were on your side of the table.

Most bad agency choices are made before the first call, when a brand goes looking for help with everything and gets sold the part the agency happens to sell. Start narrower. There are three jobs on an Amazon account, and they need different people.

The first is advertising: campaign structure, bids, budgets and the search terms behind them. If that is the whole problem, the choice is between running it in house, a software tool, or a managed service, and the three are compared here. The second is the listing and the catalogue: titles, images, A+ content, variations and the keyword work that moves organic rank, which we set out separately. The third is the account itself: inventory and FBA, account health, compliance, pricing and the reporting that says whether any of it is making money. Full service means all three, run together, because a stock decision changes what advertising should do and a listing change changes what the ads convert.

Write down which of the three you are hiring for, and whether the others are covered. An agency that is strong at one will happily take the money for all three. Our own Amazon account management is the third kind: we run the account, and the first two inside it. That is the right choice for a brand that wants one team accountable for the whole channel, and the wrong one for a brand that only needs its ads tuned.

The phrase is used loosely, so here is the test. A full service agency operates the listings and the catalogue, runs the advertising, manages inventory and the FBA settings that decide fees and stock, watches account health and answers policy notices, and reports on contribution rather than on activity. If any of those five is missing from the proposal, it is not full service, however the proposal is titled.

The last one matters most and is the one most often missing. Revenue is easy to report and easy to grow by spending more. What you need to see every month is what the channel returned after Amazon's fees, advertising and fulfilment, and the fees are where most of the margin quietly goes. An agency that cannot show contribution by product is managing your account without knowing whether it is working.

Account health belongs in the list because it is the one job that can end the others. A suppressed listing or a suspended account stops the advertising and the sales at once, and the notices arrive with deadlines. What the metrics are and how reinstatement works is worth reading before you ask a candidate who watches them and how fast they answer.

Score every candidate on the same seven things, in writing, before the proposals arrive. The weights are yours, but the seven are the ones that predicted the outcome in the accounts we have taken over from other agencies and the ones we have handed on.

One: operators or advisers. Does the team work inside your Seller Central and advertising console, making the changes and owning the result, or does it send recommendations for your team to carry out? Both are legitimate. Only one of them is accountable for what moves. Two: category experience. Ask for two accounts in your category, or in a category with the same rules, and what changed on them. Supplements, food and beauty sell under compliance rules that a generalist learns on your account. Three: the access model. Your account, your brand registry, your advertising console, in your name, with the agency as a user you can remove. An agency that wants the account under its own credentials is asking for leverage it should not have.

Four: how the work is planned. A plan that names the competing sellers and the positions that can realistically change is a plan. A plan that promises growth in general is a sales document. Five: reporting. Contribution by product, monthly, with the fees and the advertising taken out, and an honest line about what did not work. Six: the people. Who actually does the work, how many accounts they carry, and whether you meet them before you sign. Seven: the exit. What you keep when you leave, which should be everything, and how long the notice is. An agency confident in its work does not need a long contract to keep you.

Score each from one to five, weight the ones that matter to you, and the arithmetic will usually surprise you: the agency with the best deck is rarely the one with the best numbers.

Ask every candidate the same questions and write the answers down, because the differences between them are the decision. Which sellers hold the positions we want, and why? A good answer names them and says what their advantage is made of. A worrying answer talks about keywords in general. What would you change in the first thirty days, and what would you leave alone? The second half is the test. An agency that would change everything has not read the account. What have you told a client not to do? An operator has stories; an adviser has principles.

How will we know in ninety days whether this is working, and what happens if it is not? The answer should be a number you agree now and a conversation you have then. Who on your team will be in our account on a Tuesday afternoon, and what else are they responsible for? You are hiring people, not a brand. What do we own if we leave in six months? The only acceptable answer is everything: the account, the listings, the campaigns, the data and the creative.

One more, which sounds unfair and is not: what would make you turn our account down? An agency that has never turned an account down takes every account, and yours will be treated like every account.

An agency is the wrong answer more often than agencies say. If you have not yet had a first sale, the work is a launch, and a launch needs a founder's attention more than an operator's process; the path from nothing to a first order is worth walking yourself, or with a specialist for the launch alone. If your margin after Amazon's fees is thin, an agency's fee does not fit inside it, and the first job is the product economics, not the management. If your goal is to keep a channel at its current level rather than to take share, a lower cost provider will serve you better than a team built to compete. And if you want the capability in house, hire and train, and use an agency for the audit and the handover rather than the running.

The brands an agency serves well are the ones already selling, with a product that can earn more market share than it has, in a category with sellers ahead of them who can be caught. That is a narrower group than the lists suggest, and it is the group we work for.

Whoever you choose, the first month has a shape, and a candidate who describes it unprompted is telling you something. It starts with a read of the account before any access is granted: the category, the competing sellers, and a straight statement of what is wrong and what is not. Then access, in your name. Then a baseline, written down, of the numbers you will judge the work by. Then the first changes, which should be the cheapest ones with the largest effect, and wasted advertising spend is usually first. The report at the end of the month should compare against the baseline and say what comes next.

If a proposal skips the read and the baseline and goes straight to the retainer, that tells you something too. The read is how we start every account, and it is the reason we can say no to the ones we should not take.