Amazon PPC campaign structure: open a new campaign only when it needs its own budget, bid strategy or placement setting.

Amazon PPC campaign structure should follow the settings Amazon puts on the campaign: open a new campaign only when part of your ads needs its own budget, bidding strategy or placement setting. Every other split spreads the same clicks thinner. Most guides on Amazon PPC campaign structure give you a diagram with a dozen campaigns per product and a naming convention. None of them asks whether your budget can feed that many. This post shows what Amazon actually sets at each level, a three-campaign base that fits most sellers, how many campaigns your orders can support, and why November is the wrong month to rebuild.

What Amazon sets per campaign, and what per ad group

Structure is not about tidiness. It decides which settings you can change separately. In Sponsored Products, four things belong to the campaign:

• Targeting type. A campaign is automatic or manual. Amazon's targeting guide says you choose it when you create the campaign, and once the campaign is live you can't change it.
• Daily budget. Amazon's budget guide sets it when you set up the campaign, as the most the campaign spends.
• Bidding strategy. Fixed, down only, or up and down. Amazon's dynamic bidding guide puts the choice in campaign setup and the campaign settings page.
• Placement adjustments. Up to 900% extra for top of search, rest of search or product pages, per Amazon's Sponsored Products best practices. PPC Land's Sponsored Products explainer (September 2026) notes they are applied at campaign level.

The ad group holds the rest. Amazon's targeting guide says keywords, targets and bids apply to every product in an ad group, and tells you to group products that are closely related. Negative keywords can sit on the whole campaign or on chosen ad groups. So the test for a new campaign is simple: does this group of ads need a different budget, a different bidding strategy or a different placement setting? If not, it is an ad group, or nothing at all.

The Amazon PPC campaign structure that covers most sellers

For each product family, meaning one product and its sizes and colours, start with three Sponsored Products campaigns. Each has one job, and each job wants different settings.

• Auto: find. Amazon matches your ads to searches it picks. Small budget, low bids, fixed or down only, no placement boost. Its output is the search term report.
• Broad and phrase: test. The search terms auto found, plus your own research, at a price each one can afford. Small budget, down only. It proves which terms sell.
• Exact: earn. Only terms that already sold, each with its own bid. Most of the budget goes here, and if top of search pays for these terms, the placement adjustment goes here too.

This is not a structure we invented. Amazon's targeting guide says automatic targeting is "a great supplement to manual campaigns" and that advertisers often run both, but it also says to avoid mixing targeting strategies in the same campaign and ad group. Three campaigns does both. Our post on auto vs manual campaigns covers what each type is good at, and the one on match types covers broad, phrase and exact.

What makes the three work together is the negatives. When a term moves into the exact campaign, add it as a negative exact in the auto and the broad and phrase campaigns. Now only one campaign bids on that search, at the price you set for it, and the other two keep looking for new terms instead of paying for old ones.

The weekly job of moving terms along is search term harvesting, and the terms that fail become negative keywords.

How many campaigns your budget can feed

This is the question page one skips. A campaign is only useful if you can read its numbers, and you can only read them if it gets enough orders. Below about 30 orders a month, luck runs a campaign's ACoS more than your bids do.

The arithmetic is plain. Order counts swing by chance by about one over the square root of the count. At 30 orders a month that is about 18% either way. At 7.5 orders it is about 37%, so a campaign whose real ACoS is 30% can read anywhere from about 22% to 48% on luck alone. You would cut a good campaign and fund a bad one, month after month.

Now count what your budget buys. Take an example product family with $30 a day of ad budget at $1.00 a click. That is 30 clicks a day, or 900 clicks in 30 days. At a 10% conversion rate, 900 clicks bring 90 orders a month. Divide by 30 and you get three campaigns: exactly the base structure, and no more.

Build the twelve-campaign diagram on the same budget and the same 90 orders spread to 7.5 a campaign. Six campaigns get 15 each. In practice the spread is worse than even: the exact campaign takes most of the orders and the rest get a handful each.

The rule: open a campaign only when it needs its own budget, bidding strategy or placement setting, and only when it will earn about 30 orders a month. Monthly ad orders ÷ 30 is the most campaigns a product family can feed.

When a split earns its own campaign

Some splits pass both tests. These are the common ones, in the order most accounts need them:

• A product with a different margin. Two products with different break-even ACoS need different bids and different budgets. Put them in separate campaigns, not one ad group. Our break-even ACoS guide shows how to find each number.
• Your brand name. Branded searches convert differently and need a different bid. Whether they are worth paying for at all is in our post on brand defense campaigns.
• Competitor products. Product targeting gets its own campaign once the orders allow it, because its pages convert at their own rate. See Amazon product targeting.
• A term that pays at top of search. If one exact keyword earns 30 orders a month and needs a placement boost the others don't, it can have its own campaign. Our top of search guide shows how to price that boost.
• Each marketplace. Build the structure again in each country, from that country's own search terms and margins. Copying it is why German campaigns often underperform the UK ones they came from.

And some splits don't pass. Colours and sizes of one product belong in one ad group. A campaign per match type beyond the base three only divides the same clicks. A campaign per keyword makes sense only for the rare keyword with 30 orders a month of its own.

Q4: add a campaign, don't rebuild them

Black Friday is 27 November 2026 and Cyber Monday is 30 November. If your structure is a mess, this is still not the month to fix it. Moving keywords into new campaigns now means your best weeks of the year land in campaigns with no history of their own, and you lose the clean before-and-after comparison just as clicks get expensive.

Do two smaller things instead:

• One deal-week campaign per product family, with its own budget and an end date, for gift and deal searches. Its spend can't drain your exact campaign, and it stops by itself. Our keyword research post covers how to find those terms.
• Check the caps above your campaigns. A portfolio budget cap overrides every campaign budget under it, and a November cap can switch your ads off before Cyber Monday, as our post on Amazon PPC portfolios shows.

Rebuild in January, with Q4's search term data in hand and the peak behind you.

What to do this week

1. List your campaigns by product family and write next to each its orders in the last 30 days.
2. Mark every campaign under 30 orders. Ask what setting it has that its neighbours don't. If the answer is none, plan to merge it in January.
3. Check the negatives: every exact keyword should be a negative exact in that product family's auto and broad and phrase campaigns. This one you can fix now.
4. Add one deal-week campaign per product family before 27 November, with its own budget and an end date.
5. For a new product, start with the three-campaign base and add campaigns only when monthly orders ÷ 30 says you can.

If you would rather not run this by hand, Mirox builds its own campaigns for the products you pick and bids only in those, inside the daily budget you set, which Amazon enforces. Your existing campaigns keep running exactly as they are. Every bid change is saved with the numbers behind it, as how it works describes. The 30-day free trial runs live on up to 50 of your parent ASINs in 1 marketplace.

Common questions

How should I structure my Amazon PPC campaigns?

Start each product family with three Sponsored Products campaigns: an automatic campaign to find search terms, a manual broad and phrase campaign to test them, and a manual exact campaign for the terms that sell. Give each its own budget and bidding strategy, and add every exact keyword as a negative exact in the other two.

How many campaigns should I have per product on Amazon?

As many as your ad orders can feed. A campaign needs about 30 orders a month before its ACoS means much, so divide a product family’s monthly ad orders by 30. At 90 orders a month that is three campaigns.

Should I put one keyword per campaign on Amazon?

Only for a keyword that earns about 30 orders a month on its own and needs a budget or placement setting the rest do not. For everything else, single-keyword campaigns split the same clicks into pieces too small to judge.

Can I mix automatic and manual targeting in one Amazon campaign?

No. Amazon sets the targeting type when you create a campaign and does not let you change it once the campaign is live. It recommends running automatic and manual campaigns side by side, in separate campaigns.

The one-line version

Three campaigns per product family, one job each, and a new campaign only when it needs its own settings and can earn 30 orders a month.

Want to see the numbers behind a real bid? Read how transparency works, then check what your budget buys in Amazon campaign out of budget.