Choose on diagnosis, not on promises. Give each candidate read-only access, describe your worst symptom, and see whether they identify the cause or jump to a fix. The right manager names the mechanism, shows the data behind it, and tells you what they would not touch yet.
Why diagnosis is the whole test
Amazon accounts fail in predictable ways, and almost every visible symptom has several possible causes. Sales dropping could be rank loss, a price change by a competitor, a suppressed variation, an image test that went backwards, or inventory that ran thin three weeks ago and cost you velocity. The symptom looks identical from the dashboard in all five cases.
That is why interviewing on tactics is useless. Everyone can describe campaign structures. Very few can look at a specific account and say which of the five causes is operating here, and prove it. Choosing an account manager is choosing a diagnostician.
The diagnostic table
Bring your real symptom to the interview and see how close the candidate gets.
| Symptom | Likely causes, in order | Who should fix it |
|---|---|---|
| Traffic steady, sales falling | Conversion rate decline, price gap, review or rating shift, competitor image upgrade | Listings and creative, before advertising |
| Impressions falling, ACoS stable | Rank loss on head terms, budget capped early, inventory thinned | Advertising plus inventory planning together |
| ACoS rising, sales flat | Bid creep, wasted search terms, targets never reset after launch | Advertising, with a written target per product stage |
| Good ads, poor profit | Fee and freight arithmetic, price too low, high return rate | Finance and catalog work, not campaign tweaks |
| Ratings sliding | Product or packaging issue, expectation gap in the listing, unit variability | Sourcing and content, urgently |
| Nothing works in a small category | The market is too small to pay for its own customer acquisition | Nobody. Change the product decision |
A strong candidate will move down that middle column with you rather than reaching for the tool they like best. A weak one recommends restructuring campaigns before asking what your conversion rate did.
The research question that separates them
Ask what they analyze when they look at a category, and specifically what they look at beyond review count and sales volume. Those two numbers are the easiest to pull and the least informative, because everyone has them.
Our own product and category work runs on 90+ data points, and the ones that change decisions are usually the unglamorous ones: market size, growth trajectory, return rate for the segment, how concentrated the top sellers are, and the rating gap between the leaders and everyone else. Differentiation gets built from what competitors' negative reviews complain about repeatedly, not from invention, which is a much duller and much more reliable method than brainstorming.
You do not need a candidate to use our number. You need them to have a list of their own and to explain what each item changes. Anyone whose research process is a screenshot of a keyword tool is executing, not deciding.
Five checks before you decide
- Read-only access, then 48 hours. We deliver a free written audit inside two days covering listing quality, primary image click-through, conversion rate, ad performance, traffic channel activation, pricing, and return rate. Ask everyone for their version and compare specificity.
- The named person. Ask who the manager will actually be, how many brands they carry, and whether you can speak to them before signing. Pitch teams are not delivery teams.
- Coverage. Confirm who writes copy, who produces images, who handles Seller Support cases, and who watches inventory. Gaps here become your evenings.
- Access hygiene. Work happens through your own account via granted user permissions you can revoke. Nobody needs your login credentials.
- The stop condition. Ask what evidence would make them recommend discontinuing a product. An account manager with no brake is a spending plan.
What separates a manager from a task-taker
Both answer emails quickly. The difference shows up on the week when the numbers are bad. A task-taker reports what happened. A manager arrives with a cause, two options, a recommendation, and the cost of doing nothing.
You can test for it before hiring by describing a decision you already made and asking what they would have done differently. Watch whether they ask what your margin was, what the return rate is, and what the rating trend looked like, or whether they answer immediately. The ones who ask first are the ones you want.
What most agencies will not tell you
Account management is capacity work, and capacity is the number nobody publishes. The same person is either reading your search term report weekly or is not, and no framework compensates for a book of brands that is too large. Ask the number, then ask what happens to it after they sign their next three clients.
The second omission: reporting is often designed to be reassuring rather than useful. Insist that your weekly update contains profit direction, conversion rate, rating trend, and inventory position alongside spend and ACoS. We send a written update each week and run a live review every two weeks, and the point of the written one is that it can be compared to last month's without anyone's memory involved.
Related answers
- Questions to ask before hiring an Amazon agency
- What does a good Amazon account audit include
- KPIs an Amazon agency should report weekly
- Best Amazon account managers for small brands
- Build vs buy for your Amazon channel: the complete guide
Test our diagnosis first with the free written audit at Flapen.
