Amazon PPC campaign structure is how you organize campaigns, ad groups, and targets so that the
data coming back is readable and the spend is controllable. Get it wrong and you cannot tell which
keyword earned a sale, which is the actual failure mode, because you then optimize on guesses.
One principle carries most of the weight: a campaign is a budget and a strategy, an ad group is a
theme. If you cannot say in one sentence what a campaign is for and what would make you turn it off,
the structure is wrong regardless of how it performs this week.
The hierarchy, and what each level controls
Four levels, and knowing which lever sits where prevents most confusion.
Account. Everything, plus your total spend.
Campaign. This is where budget, campaign type, bidding strategy, and placement adjustments
live. It is the only level with a budget, which is why a campaign should represent a decision you would
be willing to fund or defund as a unit.
Ad group. This is where default bid, targets, and the ASINs being advertised live. Everything in
one ad group shares a default bid and a set of targets, which is why mixing unrelated products inside
one is so damaging.
Target. The individual keyword or product target, with its own bid.
The practical consequence: you cannot set different budgets for different keywords in the same
campaign, and you cannot separate performance cleanly for different products in the same ad group.
Structure is what buys you those abilities.
The one rule most accounts break
Do not put multiple products in one ad group.
It is the most common structural error and the most expensive. When I have audited accounts, it has been present in the large majority of them. When five ASINs share an ad group, they
share the targets and the default bid, and Amazon serves whichever it judges most likely to convert.
Your report then shows a keyword performing well without telling you which product did the work.
One ASIN or one tight variation family per ad group. That single change makes every other decision
diagnosable, and in our experience it is the change that most improves accounts we take over.
An Amazon PPC campaign structure that works, and why
A framework rather than a template, because catalog size changes the right answer.
Start with three campaign roles per product or product family:
1. Discovery. An automatic campaign, or a broad-match manual one, running at a modest bid. Its job
is not profit. Its job is to find search terms you did not think of. Judge it on what it discovers, not
on its ACoS, and I’d resist the urge to optimize it into silence.
2. Performance. Exact-match manual campaigns holding the keywords you have confirmed convert. This
is where the money is made and where bids deserve attention.
3. Defense. In my experience the most argued-about of the three. Targeting your own brand and your own ASINs, to hold the space competitors would
otherwise buy. Usually cheap, frequently high converting, and often mistaken for wasted spend because
it looks like paying for traffic you would have won anyway. Some of it you would. Not all.
Then add these as the catalog justifies:
Product targeting campaigns against specific competitor ASINs and categories, which behave differently enough from keyword targeting to deserve their own campaigns rather than sharing one.
Sponsored Brands once you have Brand Registry and a range worth promoting as a set.
Sponsored Display for retargeting and competitor placement.
Harvesting, which is the engine
The flow that connects the three roles, and the reason discovery campaigns exist.
- Discovery finds a search term that converts.
- You add it as an exact-match keyword in a performance campaign, at a bid you have chosen
deliberately. - You negate it in the discovery campaign, so the two stop competing and you stop paying twice to
learn the same thing. - Repeat weekly.
Step 3 is the one people skip, and skipping it is why accounts develop internal bidding wars where
your own campaigns compete for the same term. I’ve found that single omission explains more wasted
budget than bad bidding does.
Negative keywords are doing structural work here, not just blocking irrelevant traffic. Negatives are
how you route a search term to the campaign you want handling it.
Naming, which sounds trivial and is not
You will have hundreds of campaigns eventually. A naming convention decided now costs nothing and saves
hours later.
Something like Product | Type | MatchType | Goal, giving you WidgetA | SP | Exact | Performance. The specific scheme matters far less than having one, and than never changing it once bulk operations depend on it. When I have inherited an account with no convention, renaming everything was the first job.
I’d apply it to ad groups too. When you are filtering a bulk file at 11pm, readable names are worth more
than they sound.
How structure changes with catalog size
Under 10 products. Keep it simple. One discovery and one performance campaign per product, plus a
single brand-defense campaign covering everything. More structure than that is overhead without benefit, and in our experience it slows the person managing it down more than it helps.
10 to 100 products. The three-role framework per product family, with families grouped by category
rather than by arbitrary batches. Hero products get their own campaigns; the long tail shares.
Over 100 products. Segment by margin and by role. Your hero SKUs, your supporting range, and your
tail need different budgets and different tolerances, and treating them uniformly means either
starving the winners or overfunding the losers. Bulk operations become essential rather than optional.
The mistake at every size is the same: building the structure you will need at ten times your current
size. Build for the catalog you have, and restructure when it hurts.
The mistakes that cost the most
Multiple ASINs per ad group. Covered above, and worth repeating because it undoes everything else.
No negatives. Your campaigns bid against each other, and you pay for the privilege. In our experience this is the cheapest thing to fix and the most frequently left alone.
Judging discovery campaigns on ACoS. They exist to find terms. Optimizing them for efficiency turns
off the thing you built them for.
Restructuring constantly. Every restructure resets the performance history that Amazon’s delivery
uses. I’d change structure deliberately and rarely, and change bids often instead.
Budget caps that bind at noon. A campaign that exhausts its budget early is not being controlled, it
is being throttled, and you never learn what the afternoon would have delivered.
Ignoring placement adjustments. Top-of-search converts differently from rest-of-search, and the
adjustment sits at campaign level, which is another reason campaigns should be coherent units.
A note on benchmarks. You will see ACoS targets quoted as though they were universal. They are not.
The only ACoS that matters is the one your margin supports, which means working out your landed cost,
referral fee, FBA fee, and returns rate before you copy anybody’s target. I’d distrust any published
figure that does not ask what you sell.
Auditing a structure you inherited
If you have taken over an account rather than built one, five checks tell you most of what you need
inside an hour.
Count ASINs per ad group. Anything above one is a flag. In our experience this alone explains most
inherited accounts that “used to work”.
Download the search term report and look for duplicates across campaigns. The same term appearing
in several campaigns means missing negatives and internal competition.
Check which campaigns hit their budget cap and when. A campaign spent out by midday is not a
performing campaign, it is a constrained one, and its apparent efficiency is an artifact.
Look for campaigns with no negatives at all. Common, and usually the cheapest fix available.
Read the campaign names. If you cannot tell what a campaign is for from its name, nobody could
manage it deliberately, and I’d treat that as the real finding rather than a cosmetic one.
Two things worth resisting. Do not rebuild everything at once, because you lose the performance history
that Amazon’s delivery leans on and you lose your own baseline. And do not turn off the worst-performing
campaign before checking whether it is the discovery campaign doing its job.
Bids sit on top of structure, not instead of it
Worth separating, because the two get conflated.
Structure decides what you can see and control. Bidding decides what you pay within it. A good bidding
strategy on a bad structure produces confident decisions based on unreadable data, which is worse than
cautious decisions on clean data.
Get the structure right first, then tune. When I have done it the other way round, the tuning had to be redone. Our guide to
Amazon bid types and strategies covers the
bidding layer, and if you would rather have the whole thing built and managed, that sits inside our
Amazon PPC service. We are an Amazon Ads partner
and an Amazon SPN Verified Partner.
FAQ
What is Amazon PPC campaign structure?
The way campaigns, ad groups, and targets are organized so performance data is readable and spend is
controllable. Campaigns hold budget, bidding strategy, and placement adjustments; ad groups hold the
default bid, the targets, and the advertised ASINs.
Should I put multiple products in one ad group?
No. Products in one ad group share targets and a default bid, so your reports cannot tell you which
product a keyword actually sold. One ASIN or one tight variation family per ad group is the single most
valuable structural rule.
How many campaigns should I have per product?
For a small catalog, two or three: a discovery campaign to find search terms, a performance campaign
holding confirmed exact-match keywords, and brand defense covering the range. Add product targeting,
Sponsored Brands, and Sponsored Display as the catalog justifies.
What is keyword harvesting in Amazon PPC?
Moving search terms that convert out of discovery campaigns and into exact-match performance campaigns,
then negating them in the discovery campaign so the two stop competing. The negation step is the one
most often skipped, and skipping it creates internal bidding wars.
Why do my own Amazon campaigns compete with each other?
Almost always missing negative keywords. Without negatives, the same search term can trigger several of
your campaigns, so you pay repeatedly to learn the same thing. Negatives route each term to the campaign
you want handling it.
How often should I restructure my Amazon PPC account?
Rarely and deliberately. Every restructure resets the performance history Amazon’s delivery relies on.
Change bids frequently, structure infrequently, and restructure when the current setup genuinely
prevents a decision you need to make.
Last updated: August 31, 2026. Amazon’s campaign types, placement controls, and bidding options change
over time. Seller Central and the Amazon Ads console carry the current feature set for your marketplace
and account.
