The most common way an Amazon agency report flatters itself has nothing to do with lying. Every number is accurate. The account returns four times what it spends, the chart climbs, and the brand is pleased.
Then you ask which searches the money went to.
Ask any prospective Amazon PPC agency what percentage of your ad budget goes to branded terms versus non-branded terms, because branded search converts at roughly 90% to 95% and spending there buys sales you had already earned. My lead brand manager calls this the biggest manipulation in reporting he's seen in his career, and he isn't describing fraud. He's describing an agency optimizing honestly for the number you asked about.
Why does the branded versus non-branded split matter so much?
Because those two kinds of spend do completely different jobs, and only one of them grows your business.
Someone typing your brand name into Amazon has already chosen you. They saw you in a store, heard you on a podcast, got a recommendation. That shopper converts at around 90% to 95% almost regardless of what your listing does, because the decision happened before they arrived.
Bidding on your own brand name captures those shoppers and books them as ad-attributed revenue. Your return on ad spend looks superb, because it's measuring conversions that were nearly certain. Meanwhile nobody new has discovered you.
Non-branded terms are where growth lives, and they're harder. The shopper doesn't know you, conversion is lower, and the reported return looks worse. An agency judged solely on return on ad spend is being paid to avoid exactly the work that would grow the account.
There's a legitimate version of branded spend, and I want to be fair about it. Defending your own brand terms matters, because competitors will bid against your name and a shopper searching for you should find you. An always-on branded campaign at low spend is good practice. The problem is when that defensive campaign is most of the budget and the report presents it as growth.
What does a good answer to the branded question sound like?
A good agency answers with a number, tells you branded and non-branded campaigns are built and judged separately, and can show you what share of your own brand searches you're winning.
After the number, listen for whether the agency treats branded campaigns as their own category. Branded terms should be cheap to win and post a return that would look absurd anywhere else, and their efficiency targets should be set separately from the ones on category terms. A single blended return across both mixes the easiest sales in your account with the hardest, and the total can't tell you which half is carrying it.
The failure can run the other way too. On audits we often find competitors all over a brand's branded searches, with the brand capturing well under half of the searches for its own name plus its main product, while the previous agency runs a few branded campaigns in keep-the-lights-on mode. Shoppers who typed the brand's own name are being handed to someone else.
So a good answer covers both directions: branded spend that's small, defensive and winning most of its own searches, and most of the budget working on terms where the shopper hasn't picked anyone yet. Branded campaigns are table stakes, and new customer acquisition is what you bring an agency in to do.
You can check the defensive half yourself in about thirty seconds. Search your exact brand name on Amazon and see whose product shows up first. My lead brand manager's view is that an exact brand-name search should convert at around 95%, so if a competitor's product sits above yours on that page, you're giving away about the most certain sale you'll ever get.
What does an Amazon PPC agency do day to day?
An Amazon PPC agency builds and runs your Sponsored Products, Sponsored Brands, Sponsored Display and, at larger budgets, Amazon DSP campaigns, deciding which searches and products to bid on, how much to bid, and which terms to cut, against an efficiency target that should come from your margin.
The way we build an account, automatic campaigns do cheap discovery, showing us which search terms convert that keyword research never would have predicted. Once a term has earned enough conversions, we add it as a negative in the automatic campaign and move it into an exact-match campaign with its own daily budget. The proven terms get funded on purpose, and the discovery campaign keeps looking for the next one. On top of that the account is segmented by job, with separate campaigns for non-branded terms, branded terms, competitor terms, competitor product pages and defense of your own product pages.
When we audit an account someone else has been running, the gaps are usually structural. A typical audit finds no automatic campaigns for discovery on the core products at all, no targeting of competitor products, no Sponsored Display retargeting of shoppers who'd added to cart and left, and no defense on the brand's own product pages. Sometimes we find an ad group so far past break-even that our advice is to turn it off that day.
What an agency should run also depends on what you're spending. Sponsored Products and Sponsored Brands are the minimum. Sponsored Display comes later, and DSP isn't worth discussing until PPC spend is somewhere in five figures a month. We build a modest budget up as the data comes in, and a few hundred dollars of spend doesn't produce enough data to conclude anything either way.
How should an Amazon agency be paid?
Look at the incentive before you look at the price, because the fee model decides what advice you'll get for the next year.
Percentage of ad spend is the one I'd push back on hardest, and we don't use it. If an agency earns more when you spend more, they have a reason to recommend spending more, and you'll never fully know which recommendations came from your account and which came from theirs. A flat fee removes that conflict: the agency has no stake in your budget size, only in keeping you happy long enough to stay.
Percentage of channel revenue is what we use, alongside a monthly retainer on our Standard Model, and I'll defend it on one specific ground. Rebuilding a listing brings in customers who never click an ad, so billing only on ad-attributed sales would undercount the work and push us toward being an ads-only operation. Charging on what the channel produces keeps us responsible for the whole thing.
Commission only, with no floor, sounds like the safest option for you and is usually the worst. No good agency will go underwater to work with you, so the ones who accept commission-only terms staff your account with junior people they can afford to lose if it doesn't work. If your reason for wanting it is that you've been burned before, that's a reasonable instinct pointed at the wrong solution. Ask for a shorter initial term instead.
Our own Performance Model is a straight percentage of Seller Central sales with no monthly retainer, and I know that can look like the arrangement I just warned you off. The difference is who it's offered to. We reserve it for established brands already doing enough volume for a straight percentage to pay for senior people, and it gets the same team as our Standard Model.
What should I ask about pricing?
Ask for the number, and ask what you keep after it.
Nearly nobody in this category publishes pricing. We looked across the agencies we track and found almost none of them stating a figure anywhere public, which means every buyer runs the same frustrating process of booking calls to get a number. We decided to publish ours, and both of our fee models, with what each costs at five different channel sizes, are in our guide to hiring an Amazon agency.
So ask directly, early, and ask a second question most brands don't think of: does the profitability you're showing me include your fee?
That question matters more than it sounds. A client worked this out on a call with us and said it better than I would have. Her Amazon channel was running at roughly 20% net margin, our fee came off the bottom of that, and at around 15% she said the channel started to look like a liquidation channel rather than a profit one. She was right, and we were the ones who'd shown her the P&L that let her see it.
Any agency that can't tell you what the channel contributes after its own fee is either not modeling it or not sharing it, and neither answer is good.
Should ACoS or TACoS decide whether a PPC agency is doing well?
Neither one decides it alone: the ACoS ceiling should be worked out from your margin before any ads run, and TACoS has to be judged against what the account is trying to do, because a low TACoS can mean an account is harvesting demand rather than growing it.
Start with ACoS (ad spend divided by the sales those ads produced), because it's the number most agencies report. The question to ask is where their target came from. We set an efficiency ceiling for every account, and we can't set it until we've had the margin conversation. That's why we ask for your product costs before we run a single ad. Your pre-ad gross margin tells you how much ad spend a sale can carry before it starts losing money. If an agency gives you a target ACoS without ever asking what your product costs, the target came from a benchmark, and benchmarks don't know your landed cost.
The opposite problem is no ceiling at all. Founders come to us after firing an agency that ran large ad budgets with no efficiency targets of any kind, plus high fees, and lost money chasing more sales. Asking for the break-even math before committing to anything is the right instinct.
TACoS (total ad spend as a share of total sales, ad-driven and organic together) is the better health number, and it's easy to misread. One brand we spoke with ran TACoS around 9% and treated it as proof of efficiency. Both sides agreed on that call that winning new customers in its category would take something nearer 15% to 25%. At 9%, the account was collecting demand the brand had already built, and very little was going toward finding the next customer. On a different account, a declining brand leaning on ads, our advice was to push TACoS up toward 15% to 20%, spending more in the places with room to grow and cutting the waste.
So when an agency reports either number, ask what it's being compared to. A good answer names your break-even, names the job (protecting margin, or buying trial), and explains why the number is where it is.
Is there a test I can run on my own account?
Yes, and it's the practical version of the branded-terms question.
Find a non-branded search term where you hold strong market share. Amazon's search query performance report will show you impressions, click share, add-to-cart share and purchase share by keyword, so you can identify one where you're already winning a decent slice of the category.
Then deliberately pull ad spend back on that term, and watch what happens to your share.
If share holds, that spend was buying sales you were getting anyway, and it can go somewhere useful. If share drops, the spend was doing real work and you've confirmed it. Either answer is worth having, and it costs you a few weeks of a controlled experiment rather than an argument with your agency about whose interpretation is right.
One honest complication: this is harder when you're already number one in your category, because there's less room to cut without surrendering impression share you'd struggle to win back. The test is most useful in the middle of the pack, which is where most brands are.
How can I tell whether my current PPC agency is doing the job?
Look for signs that the ads are bringing in people who didn't already know you: new-to-brand orders growing, branded search volume rising over time, organic rank improving on the terms you pay for, and ad spend that returns more than it costs on the terms that matter.
The hardest part for most founders is that they can't see the work, and plenty of them tell us they don't know how to double-check what their agency is doing. When we look, the findings are often basic: back-end keyword fields left blank on one SKU and filled with words already on the front of the listing on another, a research tool that's supposedly in use with no way to confirm it, and ads spending several dollars for every dollar of sales they're credited with. In a case like that, our advice is to switch the ads off, put the budget into rebuilding the listings first, and turn advertising back on once the page can convert the traffic.
You don't need to read every campaign to check an agency, because a few signals tell you most of it.
Amazon Ads reports new-to-brand orders, meaning purchases from shoppers who haven't bought from your brand in the previous 12 months. If that number isn't growing, the ads are mostly reaching your existing customers, and the agency is getting credit for repeat purchases you'd likely have had anyway.
Branded search volume is a downstream signal. When the non-branded work is doing its job, more people end up searching for you by name. On a specialty ingredient brand we took on after two previous agencies, branded search volume grew about 27% and branded purchases about 35%, and the brand's share of purchases on its most important category search rose from about 48% to 55%, in a category that was shrinking at the time.
Organic rank on the terms you're paying for should improve. Over time, that means you're buying less of your own demand. If rank never moves and spend never comes down, ask the agency what its plan is for that to change.
And the account itself should look maintained. Ask for a walkthrough of the campaign structure on a screen share. If there's no discovery, no separation between branded and non-branded, or a pile of near-identical campaigns, you've learned something without needing a second opinion.
What else should an agency be teaching me?
Ask what you'll learn besides whether you made money, because marketing should be producing answers and sales are only one of them.
A good engagement should be telling you which claims actually convert, which pack size sells best and at what price, which competitors take share from you and on which terms, whether your product has real demand or you've been buying it, and where the category is heading. Those are answers you keep even if the relationship ends.
Ask about campaign hygiene too. Accounts that have changed hands often carry several versions of the same sponsored product campaign, launched, paused and relaunched with inconsistent bids. That churn costs you, because campaigns accumulate historical relevance and a duplicate never performs like the one it replaced. Copying a campaign and killing the original throws away something you paid to build.
And ask what they'd do before running ads. If the answer doesn't include the listing, be careful. Ads buy qualified traffic and the listing decides whether it converts, so an agency happy to spend your money against an unoptimized page is selling you the expensive half of a two-part job.
Should I hire a PPC-only agency or a full-service Amazon agency?
Hire a PPC-only agency when your listings, pricing and inventory are already in good shape and you need someone to run the ads, and hire a full-service agency when the ads are being asked to make up for problems on the page or in the margin.
A PPC-only agency can be the right call. If your listing converts well, your margin can carry ad spend, and someone on your team owns pricing, inventory and reviews, then paying for ads management alone is a sensible way to buy one skill.
The trouble starts when the ads agency is the only outside help and everything else lands on the founder's desk. A natural snack brand that came to us from a PPC-only agency was in exactly that spot. The previous agency was enthusiastic and was running ads at about a one-to-one return, while the listings, pricing, inventory and reviews sat with the founder. Nobody was looking at profit per SKU, and without it the ads were being optimized against a number that couldn't show whether the channel made money. Once branded and non-branded search were split and the rest of the channel was rebuilt alongside them, return on ad spend moved to about 3.5x over roughly six months.
The other thing a full-service agency can do is make you less dependent on ads over time. On a better-for-you beverage brand we relaunched, advertising went from carrying nearly all of the brand's sales to about 46% of them, because the listing, A+ content and storefront work lifted organic and branded demand underneath the ads. That shift came from work a PPC-only agency isn't hired to do.
How long before a PPC agency shows results?
Wasted spend can come out in the first few weeks, but Amazon advertising takes closer to 90 days to compound, and any change in profit worth reporting to a parent company takes months.
The quick wins come from cutting. Turning off an ad group that's losing money, adding negative keywords and moving budget away from campaigns that never convert can all happen early, and on a messy account they show up in the numbers fast.
Growth is slower because campaigns need time to breathe. Amazon rewards a campaign's history, the same reason the campaign churn I described above costs you, and an agency that keeps rebuilding campaigns every few weeks resets that clock each time. Our own onboarding runs the first 90 days on a much tighter call cadence than the monthly rhythm clients settle into afterward, because that's when the account is being rebuilt.
Ask any agency you're considering to forecast. An operations leader once pushed back on me for refusing to give one, and he was right. He wasn't trying to hold me to a number. He wanted someone to make assumptions and show them. The useful version is a range: what the channel is leaving on the table today, and then what 20%, 50%, 100% and 200% growth would each look like for your channel. If an agency can't build that for you, it hasn't looked closely enough at your account to know.
When is an agency not worth it yet?
When the math says so, and a good agency will tell you.
Run the rough calculation. At a healthy pre-ad gross margin, with a typical management fee and a modest ad budget, most brands need to be doing somewhere around five figures a month on Amazon before an agency clears its own cost. Below that, you're paying for management out of a channel too small to carry it.
I've turned work away on this. One prospect's monthly Amazon sales were small enough that our minimum would have been over a third of their revenue, which is absurd and I said so. The right advice for that brand was to fix their listings and pricing themselves first, and I'd rather give it than take the retainer.
The other version is category size. I've declined to pitch a retainer to a brand whose category search volume was too thin for the channel to ever cover the fee, however well we ran it. We set up a small always-on branded campaign to protect their name and told them to come back when the groundwork elsewhere was done.
If an agency has never once told you that you're not ready, ask yourself whether they'd know.
The bottom line
The uncomfortable truth about this category is that a competent agency and a coasting one produce reports that look almost identical, and the difference only shows up in questions most brands don't know to ask.
So ask the branded-terms question on your next call, whoever you're talking to. If the answer is confident and specific, you've learned something good. If it's vague, you've learned something more useful.
And if you want a second read on what your current account is doing, send us your ASINs and we'll tell you what we see, whether or not there's a conversation after it. We've never regretted giving away the analysis.
You're not being difficult by asking hard questions about your own money. You're doing the job.






