There is no shortage of Amazon agencies in the US. There is a shortage of ways to tell them apart.
Open forty agency sites and you get the same service list, the same partner badge, and the same growth percentage with no baseline attached to it.
So this is a procedure rather than a ranked list. Seven steps to run before you sign anything:
- How to work out what is actually broken in your account before you take a single sales call
- Which claims on an agency website can be checked independently, and exactly where to check them
- What a US-based provider buys you, and which parts of that are worth paying extra for
- What to put in writing before onboarding, and how to judge the engagement at ninety days
Full disclosure before you read on: Xneeti is an Amazon account management platform. The steps below are written to be run against us too.
What Amazon account management actually includes
Account management is the least standardized phrase in this industry. Two providers using it can be selling jobs that barely overlap.
The distinction that matters most is this. An ads agency manages campaigns. An account management agency manages the account those campaigns run inside. If campaigns really are the only gap, Amazon Sponsored Ads handled well is a narrower and cheaper purchase than everything below.
Take that table into every sales call and make the provider mark each row as owned, shared, advisory, excluded or chargeable. Two proposals are not comparable until both have been marked up this way.
Step 1: work out what is actually broken before you look at anyone
Most brands start this search after a bad quarter. Then they shop for a provider who sounds impressive rather than one who fixes the specific thing that went wrong.
The symptom is rarely the cause. Here is how to trace one back to the other.
Before you conclude the first row is an agency problem, check your numbers against normal Amazon Ads Cost benchmarks for your category. A 32% ACoS is a crisis in one category and a good month in another.
The last row is the one people misdiagnose most. A brand that has hit a catalog ceiling does not need better management of what it already sells, it needs more surface area, which is closer to the problem Amazon Ads for Scaling Brands is built to solve.
Any provider is genuinely strong at two or three of these rows, not all six. Knowing which two you need turns forty options into six.
Write your top two constraints down before the first call. Then do not let a pitch move them.
Step 2: decide which model you are actually buying
Most buyers compare agencies to other agencies. The more useful comparison is between six commercial models that behave very differently once money changes hands.
If advertising is genuinely the whole problem, read our guide to Amazon PPC Ads first and then look at the specialists. Our breakdown of Amazon Product Ads Management Companies covers the ones worth a call at that level.
The pairing almost nobody prices is Amazon's own. Strategic Account Services covers catalog, compliance and program access but not advertising, so it sits alongside a provider rather than replacing one. Price it before you take an agency call, because it changes what the agency fee is actually buying.
Pick the model first. Then compare providers only inside it.
Step 3: what "in the USA" actually buys you
A US-registered agency and a US-delivered service are two different things, and the website will not tell you which one you are looking at.
Where the people sit decides your response time
Suppressed listings, Buy Box losses and Seller Central cases are hour-level problems. A team working an inverted clock will still fix them, but the first reply lands the next morning. Ask which hours your account is actively watched, stated in your time zone rather than theirs.
Who attends the call is not always who does the work
The common structure is US strategy and account management with offshore execution behind it. That model works, and it lowers your cost. It only fails when nobody tells you upfront.
Contracts, entity and recourse
A US entity gives you an enforceable agreement and a jurisdiction if something goes wrong. Check that the legal entity on the contract matches the brand on the website. They are not always the same company.
Category compliance is a US-specific job
Supplements, cosmetics, children's products and anything carrying a safety claim bring US regulatory exposure that lands on your listing as a documentation request. A provider without US category experience treats those as Amazon tickets rather than regulatory ones, and the difference shows up in how long the listing stays down.
Data handling and account access
Ask how Seller Central access is granted and removed, and where your account data is stored and processed.
Offshore delivery is not a red flag on its own. Undisclosed offshore delivery is, and so is paying a US premium for work being done somewhere else.
Step 4: verify the claims that can be verified
Most of an agency website cannot be checked. A small part of it can. Start there.
Managed revenue, client counts and growth percentages are not lies. They just carry no cohort, no baseline and no period, which makes them impossible to compare between two providers.
One question resolves most of it. Ask for three named clients in your category and revenue range, each with a metric and a timeframe attached. A provider with real results sends them quickly. A provider without them offers you an aggregate instead.
Then ask to speak to one current client rather than read a testimonial. How fast that gets arranged tells you as much as the call itself.
Step 5: read the pricing model, not the price
Two providers quoting the same monthly number can be selling completely different incentives.
The fee is also not the cost. Ask which of these sit outside it before you compare two numbers:
- Creative production: images, video, A+ modules and brand store builds, usually quoted per asset
- Software subscriptions, sometimes passed through at cost and sometimes marked up
- Reimbursement recovery, often charged as a percentage of whatever comes back
- Additional marketplaces such as Walmart, priced as a separate engagement
- The internal hours you still carry: approvals, supplier coordination, answering their questions
Creative is the line item that catches people out. Video in particular gets quoted per asset, so agree what a Sponsored Brands video costs before you plan any Amazon Video Ads work, not after the campaign is approved.
Do the same with tooling. If you already pay for Amazon Ads Software yourself, ask whether the retainer replaces those licences or sits on top of them.
Compare total monthly cost of ownership. Not headline retainers.
Step 6: the first call, and what a good answer sounds like
The questions matter less than the shape of the answers you get back.
The TACoS question is the one that separates operators from campaign managers. ACoS measures ad cost against ad-driven sales only, so it can look healthy while organic rank erodes underneath and total spend climbs to cover the gap.
The report question is worth pushing on too. If what arrives is a screenshot of the Amazon Ads Dashboard with no commentary attached, you are buying reporting rather than analysis, and you can pull those numbers yourself for free.
The failure question has a purpose as well. A provider who cannot name one has either not run enough accounts or is not being straight with you.
Last thing. Ask directly whether the senior person on this call stays on the account after signing, and watch how long the answer takes.
Step 7: test before you commit
You are being asked to commit twelve months on a judgement formed over two calls and a proposal deck. That asymmetry is fixable.
Buy an audit first
A paid audit is the cheapest test available. You are buying a sample of their thinking, and you keep the findings whether or not you hire them.
What a real audit contains
- Named ASINs and campaigns, not category-level observations that would apply to any account
- Problems ranked by revenue impact, with an estimate of what each one is costing you now
- At least one finding you did not already know about
- A sequenced plan for the first ninety days, rather than a list of services
Structure the contract for an exit
Ninety-day initial terms with rolling thirty-day notice afterwards is standard and fair to both sides. A twelve-month lock with no performance clause and no exit benefits one party, and it is not you.
Settle what happens to your creative, campaign structures and reporting history if the relationship ends. That conversation is much harder to have later.
Red flags worth walking away from
These are patterns rather than isolated incidents. Most sellers who got burned saw at least one of them before signing.
- Guaranteed sales figures or ranking positions offered before anyone has looked inside your account
- Pricing that cannot be explained clearly, or a proposal where the scope and the fee do not match
- No answer to who specifically works on your account, or a different name every time you ask
- Reporting that arrives as green arrows and percentages with no baselines and no action items
- More ad spend offered as the fix for every problem, including problems caused by ad spend
- A twelve-month lock-in with no performance clause, no notice period and no exit terms
- Partner badges on the site that do not appear anywhere in Amazon's own directory
- The senior person who ran the pitch is unavailable for any question once the contract is signed
- Reluctance to connect you with a current client, or testimonials that trace back to no real company
One flag is a question worth asking. Three is a pattern, and that pattern has usually already been described by someone in a seller community, which is worth twenty minutes of searching before you sign anything.
What to agree in writing before onboarding
Most agency relationships fail on ambiguity rather than competence, and the ambiguity is usually visible in the original contract.
Agree the starting numbers before onboarding, in writing. TACoS, contribution margin, organic rank on your top ASINs, and your current reimbursement position. Without a baseline, every future report is an assertion rather than a measurement.
How to judge the engagement at 30, 60 and 90 days
Different parts of an Amazon account move on very different clocks, so set the expectation before the first invoice.
Advertising responds in weeks. Listing and SEO work needs a full quarter to index and accumulate ranking signals, and any provider promising otherwise is describing advertising.
At ninety days, judge against the baseline you agreed and against contribution margin rather than revenue. If they cannot explain what changed and why, the next ninety days will look the same.
One more question to ask in 2026
Amazon's shopping assistant, renamed Alexa for Shopping, now sits between a lot of shoppers and the results page. It reads a listing differently from the way the search algorithm does.
Bullet points written to match keywords are now being read by something that answers questions instead.
The second change is speed. The advertising auction moves hourly, which makes a weekly optimization cadence a structural disadvantage rather than a service-level choice.
Three questions that separate providers who have adapted from providers who have not:
- How have you changed listing copy in response to AI-assisted discovery, and can you show a before and after?
- How often are bids and placements actually adjusted, and is that automated or a person making changes?
- What are you doing differently this year from what you did in 2024?
A provider with no answer to the last one has not changed anything.
Why sellers choose Xneeti
Xneeti was built for US brands that have already been through an agency, know exactly what the failure looked like, and are not willing to repeat it.
It is not a tool and not a traditional agency. Natively built AI runs the account hourly across ads, listings, inventory and payouts, with a dedicated strategist reviewing everything it does.
Against the questions above: continuous optimization instead of weekly cycles, listings written for both A10 and Amazon's shopping assistant, and payout reconciliation that closes the margin gap most accounts never audit. Account-to-manager ratios sit at roughly half the industry average, so your account is not one of twenty.
US-based, Amazon Ads Verified Partner and Amazon SPN Partner, built by ex-Amazon category managers and ex-Google engineers. Portfolio averages across 80+ managed accounts are 50% lower TACoS and 30% revenue growth. Rated 4.8 on Google.
The honest boundary is Vendor Central. Brands with heavy 1P exposure should test hybrid scenarios during scoping before committing.
If you want to see what the AI finds in your account before committing to anything, book a demo and bring your last 90 days of advertising data.
