Most of the brands that call us with a stalled Amazon channel have grown a channel before. They've done it in grocery or in the B2B business that funds the consumer line. Amazon is the one place where adding ad budget or dropping the price can barely move the line, or move it the wrong way, and the reports still look fine while it happens. If that's where you are, you didn't miss something obvious. The channel hides the problem from the people running it.
On sales calls I put it this way: if your bones are set up right, then as you grow organically or do any other marketing, you get interest and sales on Amazon. Get the bones right, make the unit profitable, and after that it's a fairly simple question of how much you reinvest.
To increase Amazon sales without buying them, work in a fixed order: confirm each SKU clears its margin floor, finish the listing so it converts, then run ads that bring in new customers, and reinvest only once organic rank holds. Each step multiplies whatever sits underneath it, broken parts included.
We call that sequence the Amazon Profit System. On our homepage it's three steps (rebuild the math, earn the shelf, spend against profit). On this page I've split the last step in two, because deciding how hard to push ads and deciding when to reinvest the profit are separate calls, and brands get them wrong for different reasons.
Why does the order decide whether more Amazon sales make you money?
Because on Amazon each step amplifies the one before it, and growth stacked on a broken step grows the loss. Scale ad spend before the unit economics work and you're buying more of the sales that lose money. Pay for traffic before the listing converts and you're renting customers you can't keep.
The way I'd describe the mechanism to a CFO: Amazon is a search engine built to answer one question: what do people buy when they search for X? Every sale you make for a term is Amazon's evidence that you answer that question well, so it moves you up in organic rank for that term, and a listing that converts well also pays less for the same click. The better you perform, the further it launches you. That flywheel works in both directions, though. A listing that converts badly sends Amazon the opposite evidence, and ad spend pointed at it buys the clicks at full price and the rank not at all.
So the first thing we check is whether the unit makes money with no advertising at all. We call that number pre-ad gross margin: sale price, minus true landed cost (product, packaging and inbound freight), minus Amazon's fees. For a consumables brand it needs to clear 30%, durable goods need closer to 40%, and under 20% we tell a brand the channel can't carry advertising yet. I've written up where that floor comes from, and what to do if you're under it, in the piece on the 30% floor. Read that first if you don't know your number per SKU, because nothing below this paragraph fixes a unit that loses money.
One P&L we built for a client shows why the order matters so much. Amazon's fees came to roughly 43% of sales, fulfillment was about 30% of that, ads ran just under 15%, and about 20% was left as net. There isn't much room in a stack like that, and a listing that converts poorly spends it on clicks that don't turn into orders.
The order, and the check to pass before each next step
This is the table I'd want a new operator to keep open. Each step has a gate, and the gate is the thing to confirm before the next step gets any budget.
| Step | What you fix | The check before moving on | What skipping it costs |
|---|---|---|---|
| 1. Rebuild the math | Per-SKU pre-ad gross margin: price, true landed cost, Amazon's fees, pack size | Every SKU you'll advertise clears 30% (consumables) or about 40% (durables) | Ad spend multiplies a per-unit loss, and better ROAS can't save it |
| 2. Earn the shelf | Category, title, images, A+ content, backend keywords, Buy Box, pricing against fee thresholds | Indexed for the terms you want to buy, conversion near or above 5%, Buy Box held on hero SKUs | You pay full price for clicks that leave, and the ads can't place on terms you aren't indexed for |
| 3. Build the review floor | Vine, insert cards, review averages per variation | Roughly 15 reviews at 4.3 stars or better on the SKUs you'll advertise | Shoppers skip the listing, and AI shopping answers skip it too |
| 4. Spend against profit | Brand and non-brand campaigns split, TACoS ceiling set, spend staged by budget | New customers outpacing the ones you lose, inside the TACoS ceiling | Ads harvest people who'd have bought anyway, and the channel stops growing |
| 5. Reinvest | Subscribe & Save, repeat orders, long-tail terms first, then category terms | Organic rank holds on the specific terms when spend comes down | You keep buying sales you'd already earned |
Steps 2 and 3 are both inside what our homepage calls earning the shelf. I've broken reviews out because they have their own gate, and because brands that pass the listing check often still fail this one.
What should you fix on your Amazon listing before you run ads?
Fix the four things that decide whether a click turns into an order: the category Amazon has filed you in, what the listing says in each element on its own, the backend search terms, and whether you own the Buy Box. Ads can't fix any of them, and all four change what an ad dollar returns.
Josh Grillo, who runs brand management here, explains it with a coin. One face is advertising, and its job is to bring qualified traffic to the listing. The other face is the listing, and its job is to convert that traffic. Ads drive visits, and the listing turns visits into orders. So sending paid traffic to a listing that isn't finished wastes the spend and starves everything downstream of it, because no order means no Subscribe & Save signup and no repeat purchase. That's why we build the listing foundation before we spend a dollar on ads, on every account. And it's why I've told brands on a first call that I wasn't comfortable pitching them an ongoing retainer until that foundation was set. Running ads straight away would teach them faster, but it would cost a lot more than finishing the listing first (which is an odd thing for an agency to say, and still the right call).
There's a harder version of that rule. Amazon indexes you for the words on your listing, front end and back end, and you can't run ads on a term you aren't indexed for. On one call this year I told a brand that their ads wouldn't succeed if their listings stayed the way they were, because they weren't showing up organically for a lot of their own category terms, much less in a position to buy them. For some brands, the ads could be switched off today with no effect on sales, because they were never placing where it mattered.
Where to start on the listing:
- The category. Your browse node is the aisle Amazon has filed you in, and most brands set it once from a dropdown and never check it again. The wrong one affects compliance, your competitive set, automatic ad targeting and best-seller rank. The ten-minute check and how to get it changed are in Amazon Browse Node: How to Find and Fix the Wrong Category.
- Every element carrying the case alone. Shoppers don't assemble your title, bullets, images and A+ content into one argument, so each has to work for someone who reads only that one. Images do most of the selling now, and bullets are often collapsed behind a click. I wrote this up in detail in how we fix an Amazon listing.
- The backend field. It holds roughly 250 characters, and the most common waste is repeating words already on the page, which Amazon indexed the moment you published them. It's for synonyms, spelling variations, Spanish-language searches and terms that don't fit the visible copy. More in why your backend keywords shouldn't repeat your listing.
- Ranking mechanics. What decides where a listing ranks beyond the copy, including sales velocity per keyword, is in Amazon SEO for Food Brands: What Ranks a Listing in 2026.
Two more sit in this step for most of our accounts. The Buy Box: if resellers or unmanaged promotions are taking it, you're advertising for someone else's sale. On one specialty ingredient brand we took over, the hero SKU was winning the Buy Box around 72% of the time and one variant as low as 43%. And price against Amazon's fee thresholds, because a price set just above one changes what you keep on every order. That ingredient brand had a key variant listed just above one, with no Subscribe & Save discount to soften it.
If you'd rather someone else looked first, What an Amazon Audit Should Tell You Before You Hire Anyone covers what a useful audit should hand you, whoever runs it.
What conversion rate should an Amazon listing have before you spend on ads?
Plan around 5% as the floor, and treat beating your category median as no reassurance at all. The median includes a lot of neglected listings, and Amazon converts higher than other marketplaces because shoppers there are further along. When a brand converting around 5% told me they were beating their category median, I told them I'd want to see roughly double that before calling the listing finished.
Two cautions so you don't misread your own number. A newly launched product often converts high and then drifts down as you push non-branded traffic at it, because that traffic is colder. One launch I audited was converting at 9.5%, which is excellent, and planning around it as the new normal would have made the next quarter look like something broke. And conversion is per SKU. A blended number hides the variant that's dragging the rest.
How many reviews do you need before running Amazon ads?
Around 15 reviews at a rating of 4.3 or better on each SKU you'll advertise. Below that, you're paying to send traffic to a listing that can't convert it, and the spend teaches you nothing you couldn't have learned for free.
The 4.3 isn't arbitrary. Between 4.3 and 4.7 Amazon displays a 4.5-star rating, and at 4.2 the display drops to 4 stars. Conversion falls away at that line. The organic route to 15 reviews is slow, because the review rate on Amazon sits around 2%, so a brand waiting on organic reviews needs roughly a hundred sales to earn two. Amazon Vine is the fastest legitimate way off zero: enroll up to 30 units, and on a 27-unit batch plan on roughly 23 or 24 reviews back, from reviewers who are thorough and don't rubber stamp. Josh's full take, including when Vine will hurt you, is in the honest version of Amazon Vine.
Food and beverage brands have an extra problem here. Amazon changed how variations aggregate reviews, and flavor is the variation theme that no longer pools them, so a listing that showed one combined rating across four flavors can now show each separately, one with 7 reviews beside one with 89. If your catalog runs on flavors, check what shoppers see before you plan ad spend around it. The workaround and when not to use it are in Amazon Stopped Combining Your Reviews Across Flavors.
How should you run Amazon ads once the listing is ready?
Split branded search from non-brand and category terms, set a ceiling on total ad spend as a share of sales, and add ad types in stages as the budget grows. The goal of this step is new customers. Defending your own brand name is part of it, and if it's all the ads do, you're paying for people who'd have bought anyway.
Branded campaigns protect the name, and they matter more than brands expect, because competitors bid on your brand terms and branded traffic converts at a far higher rate than anything else you buy. Non-brand and category campaigns are where new customers come from. When the two are blended into one ROAS number, the branded sales flatter the total, and that's the oldest reporting trick in this business. Ask whoever runs your ads for the split.
The ceiling is TACoS, total ad spend divided by total sales including organic. There's no single good number. If the job is protecting margin on loyal customers, a low TACoS is right. If the job is winning new ones, we'd expect something nearer 15% to 25%, and a brand at 9% calling it efficient is often harvesting demand it already had. The test underneath is whether new customers are arriving faster than existing ones drop off, because in consumables that's the only definition of growth that holds.
On staging: Sponsored Products and Sponsored Brands are the minimum. Display comes later, and DSP isn't worth discussing until PPC spend is in five figures a month. We start budgets modestly and build them up, because at a couple of hundred dollars there isn't enough data to conclude anything, and blasting a big budget into a new structure teaches you less than it costs. Stability matters too. Accounts that have changed hands often carry several versions of the same Sponsored Products campaign, launched, paused and relaunched with inconsistent bids, and each copy throws away the history the original had earned. The staging by budget is in The Amazon Ad Funnel, Staged by What You're Spending.
The waste when nobody is minding this step can be large. We've seen single ad groups so far past break-even that the only advice was to turn them off immediately.
Does Amazon's AI shopping assistant change the order?
It raises the stakes on steps 2 and 3, because the assistant reads the listing and the reviews and returns a shortlist. Shoppers using Rufus (which Amazon has renamed Alexa for Shopping) type a situation, something like a hot sauce gift for someone who likes heat but needs low sodium, and the answer comes back as a shortlist of a few products.
From what we can see, the assistant reads what's on Amazon: your title, bullets, A+ content, the text in your images, your reviews and your returns. We haven't seen a brand get credit on Amazon for what only lives on its own website, so don't count on off-Amazon SEO carrying over, and products under a 4.0 average get skipped in its answers. A thin listing used to cost you conversion. Now it can cost you eligibility before a shopper sees the product at all. The detail, including how to mine competitors' customer questions for listing copy, is in Amazon Rufus and the Long Question.
Which numbers tell you the order is working?
Watch conversion per SKU, organic rank on the specific terms you're targeting, the share of sales that come from ads, and new customers against lost ones. Those four tell you whether the listing is converting, whether rank is building, and whether the ads are finding anyone new. ROAS on its own can't tell you any of it.
Brand Analytics, which comes with brand registry, has most of it. The search query reports show how much of the market for each keyword you're getting. On one listing we reviewed, the report showed sales on a small term with only a few hundred searches in a quarter, and no sales at all on a bigger term the brand cared about, even though Amazon already saw them as relevant to it. That's the kind of read that tells you where to put the next dollar of listing work or ad spend. Which reports to read monthly, and in what order, is in Amazon Brand Analytics: The Reports a Food Brand Should Read Monthly.
The share of sales from ads matters more than most dashboards suggest. When nearly all of a brand's sales come from ads, the revenue is rented, and it stops the day the spend stops. When that share falls while sales grow, organic rank and repeat buyers are carrying more of the load, and the brand owns more of its revenue.
When should you reinvest, and where should the money go?
Reinvest once organic rank holds on the specific terms you've been buying, and put the money into repeat purchase first and broader terms second. The test is simple: bring spend down on a term you've been advertising and watch whether the rank stays. If it does, the sales on that term are earned, and you can move budget to the next one.
Rank on a term moves on daily sales velocity, so a specific long-tail term needs a number of daily sales you can produce, and a broad category term needs a volume you can't yet. Josh's rule is to take the specific terms to the top three first and contest the category term from a position you can hold. In one niche we sized this summer, holding page one meant about 40 units a day.
Repeat purchase is the other half. The most successful brands we work with don't buy all their sales. They compound repeat orders and Subscribe & Save until velocity earns organic rank, and the paid cost of finding customers falls. A subscriber can be worth several times a one-time buyer over their lifetime, and it's why we stood up Subscribe & Save on both the snack brand and the ingredient brand below. The full argument is in The Flywheel: Stop Buying the Sales You Already Earned.
What does it look like when the order is right?
Three brands, three different starting points, and the same order underneath each one. None of them needed more ad spend first.
| Brand | Where it started | What we fixed, in order | Result |
|---|---|---|---|
| Natural snack brand | Ads at roughly 1:1 return, more than half of sales from ads, single-packs only | Per-SKU margin floors, then the catalog and listings, then ads split by brand and non-brand | ROAS from break-even to about 3.5x, conversion near 11.5%, and the founder reported profit at 10x what it was with the previous agency |
| Specialty ingredient brand | Flat for two years after two agencies, Buy Box in the low 60s, pricing broken | Landed-cost sheet, Buy Box recovery, listings and A+, then ads restructured | Conversion from 16% to 26%, Buy Box to 88%, +60% year over year in a category that shrank more than 20% |
| Better-for-you beverage brand | Almost all sales bought with ads, no storefront, no campaign structure | Positioning, full listing and A+ rebuild, storefront, then ads rebuilt for new customers | Monthly revenue up more than 10x, ads down to about 46% of sales, 79% of ad sales from non-brand and category terms |
The snack brand's founder put it in one line after checking their own profitability dashboard: "This is 10x the profit we were getting with the previous agency." The full write-ups are the snack brand that went from break-even ads to profit, the ingredient brand that was flat for two years, and the beverage brand that stopped buying its sales. The rest are on our case studies page.
The clearest test of listing-first I know of came from a coffee brand that asked us to rework the catalog before they'd trust us with ads. Sales grew in the first month after the listing work went live, and again in the two months after, with no ads running, and when we added ads they roughly doubled it. It's rare to see listing work isolated from ad spend like that, and it's the result I'd point to if you doubt the order.
Who is this for?
This is for a brand that's been on Amazon a year or more, grew, and then stalled, and can't get a straight answer on whether the channel makes money. Usually that's a natural or better-for-you brand, or a manufacturer with a consumer line, doing enough volume that the ad data shows a pattern.
It isn't for a brand starting from zero on Amazon, because without six months of data there's nothing to read yet. And it isn't for a brand that wants to keep margins under 20% at any cost. We politely decline those, because at a 10% margin I can deliver five-to-one on ad spend and the brand still loses money.
If you're reading this because the channel stalled on your watch, you're in good company. Most of the brands we work with came in under the margin floor, with a listing that undersold the product and ads that looked healthy in the dashboard. Every one of those is fixable, and you're already asking the right question.
Where should you start?
Start with step 1 on your own: pull your pre-ad gross margin for every SKU you advertise. If they all clear the floor, check the listing against the table above before you add a dollar of ad budget.
If you want us to look, there are two ways in. The free trial is for brands whose listing is in decent shape and whose ads are the bottleneck: our ads team runs your account for four weeks and you pay us nothing. We read your ads data before we say yes, and if the listing is what's holding the channel back, we'll tell you that instead of taking the account. If you'd rather start with the full picture, the Category Analysis is our paid diagnostic: it rebuilds your unit economics SKU by SKU and maps where the money is leaking, and it's credited in full against your first month if you go ahead within 30 days. You can ask about it on our contact page.





