Amazon PPC Strategy: Campaign Structure That Scales Past 1M

An Amazon PPC strategy that scales past $1M in annual revenue is mostly an architecture problem, not a bidding problem. Once an account carries a few hundred targets, growth is capped by campaigns competing with each other for the same query, budget absorbed by traffic nobody chose to buy, and a reporting surface too tangled to read. The structure below is what a starter setup usually has to be rebuilt into: brand separated from non-brand, three tiers with a defined flow between them, and negatives doing the plumbing.

What breaks when a starter setup gets big

A structure that worked at $20k/month tends to fail in three predictable ways at $80k+.

Budget collision. An auto campaign, a broad manual campaign and an exact campaign all match the same converting query. You pay to compete with yourself, and the query's true cost is smeared across three reports.

Brand traffic hiding the truth. Branded search converts at a much higher rate than category search, so when both live in one campaign the blended ACoS looks fine while non-brand acquisition quietly loses money.

Budget behaving like a cap instead of a decision. Campaigns that hit their daily budget by early afternoon go dark, and the highest-intent hours get spent by whichever campaign happened to burn fastest. If the format-level split underneath is still unsettled, Amazon PPC explained covers how Sponsored Products, Brands and Display divide the work before structure enters the picture.

Rule one: brand and non-brand never share a campaign

Split the account at the top level into brand defense and non-brand acquisition, and never let them share a budget, a bid strategy or a target.

Brand campaigns are cheap, convert well, and mostly buy traffic that would have found you anyway — their job is defending the branded results page from competitor conquesting, not growth. They typically run on exact match with fixed bids, because you don't want Amazon raising or lowering a bid on a query you already own. Amazon's options are fixed, dynamic down only, dynamic up and down, plus rule-based bidding; the trade-offs are worked through in bidding strategies and budgets.

Non-brand campaigns are where incremental revenue comes from and where efficiency targets should be genuinely looser. Every brand keyword and misspelling belongs as a negative exact in every non-brand campaign — otherwise a broad target eats your own branded query at a fraction of the CPC and inflates the whole account's apparent performance.

Competitor conquesting is a third bucket, not a subset of non-brand: lower conversion rate, higher CPC, defensible only against lifetime value. Own campaigns, own efficiency ceiling.

The three tiers: discovery → harvest → performance

Inside non-brand, the structure that holds up under scale is a one-way flow. Queries enter at the top and graduate downward; nothing flows back up.

Tier 1 — discovery

Auto and broad manual campaigns on modest budgets, there to find search terms and ASINs you haven't thought of. Amazon's automatic targeting creates four targeting groups with individually settable bids — close match, loose match, substitutes and complements — and one blended default bid across all four is the most common way to make discovery data unreadable. Bid them separately: close match and substitutes behave nothing alike.

Discovery is a research line item. Judge it on qualified terms produced per month, not on its ACoS.

Tier 2 — harvest

Phrase and exact campaigns holding terms that cleared a promotion threshold. Set that threshold once and apply it mechanically — a fixed number of orders (not clicks) at an acceptable cost over a defined window — so promotion is a rule, not a mood. Broad match on Amazon now serves on semantically related queries, so a broad target without a negative layer is a discovery campaign wearing a manual campaign's name.

Tier 3 — performance

Exact match campaigns for your proven revenue drivers, with the account's real budget and the tightest efficiency target. This is the tier that gets protected during Q4, gets placement modifiers, and gets watched daily. Everything above it exists to feed it.

Negatives are the plumbing, not the cleanup

In a tiered structure, negatives do two different jobs and it's worth keeping them separate.

Graduation negatives stop tiers from bidding against each other. When a term moves from discovery to harvest, it goes into the discovery campaign as a negative exact the same day; when it moves from harvest to performance, the same happens one tier up. Skip this and the three tiers become three bidders on one query — the collision problem, rebuilt on purpose.

Waste negatives block queries that will never convert: wrong size, wrong use case, research intent, DIY variants. Amazon supports negative exact and negative phrase at both campaign and ad group level. Look for recurring modifiers rather than individual terms — one negative phrase usually kills what fifty negative exacts would.

The source for both is the search term report in the Amazon Ads console under Campaign Manager → Reports: weekly for young campaigns, roughly monthly for stable ones.

Single-keyword vs themed ad groups

Single-keyword campaigns give you clean bid control and unambiguous attribution: one target, one budget, one number to read. The cost is administrative — 40 ASINs across five keyword themes becomes a structure nobody maintains, and each campaign gets too little traffic to learn from.

The practical line: single-keyword campaigns for tier 3 only, where a handful of head terms carry most of the revenue. Everywhere else, use themed ad groups of 5–15 closely related targets that share an intent and a rough CPC band. Mixing a $0.60 long-tail term with a $4.50 head term in one ad group means one shared bid serving two economics — one of them will always be wrong.

Naming conventions and budget ceilings

Naming is not cosmetics. At a few hundred campaigns, a consistent name is what makes bulk-file edits, filtering and reporting possible, and it's among the first things we rebuild in an Amazon PPC management engagement. Pick a delimiter, keep the field order fixed, and never improvise:

SP_NONBRAND_PERF_MAGNESIUM-GUMMY_EXACT tells you what it is, what it's for and where it sits in the flow without opening it.

Budget ceilings then attach to groups, not guesses. Amazon portfolios accept an optional budget cap — recurring monthly or date-range — and when it's met, the campaigns inside pause. The caveat in Amazon's own documentation is that a portfolio cap does not pace spend across the period: one campaign can consume the whole thing. So keep discovery and launches in their own capped portfolio, keep tier 3 generously capped at campaign level, and never put an experiment in the same portfolio as a mature performer.

Restructuring this way is what moves efficiency without cutting volume — in one outdoor brand's account, ACoS went from 44% to 21% while sales grew 68%, largely because the tiers stopped bidding against each other.

FAQ

How many campaigns should an Amazon PPC account have?

As many as your reporting needs, and no more. A $1M+ brand typically runs 20–60 Sponsored Products campaigns: brand defense, competitor conquesting, and a discovery/harvest/performance set per product group. If you can't state what a campaign is for in one sentence, it should be merged or paused.

Should brand keywords be in a separate Amazon PPC campaign?

Yes. Branded search converts far better than category search, so mixing the two hides the real cost of non-brand acquisition behind a flattering blended ACoS. Keep brand in its own campaigns with their own budget and efficiency target, and add your brand terms as negative exact across all non-brand campaigns.

What is keyword harvesting in Amazon PPC?

Harvesting is moving a search term that has proven itself in a discovery campaign into a dedicated phrase or exact campaign with its own bid. The step people skip is negating that term back in the discovery campaign — without it, both campaigns bid on the same query and you pay twice for one click.

How often should I restructure my Amazon PPC campaigns?

Rarely, and never mid-Q4. A well-built structure should last a year or more; the routine work is bid, budget and negative management inside it. Restructure when the account outgrows its logic — a new product line, a new market, or a merger of two accounts — not because performance dipped for two weeks.

Where to start

If your account sits between a starter setup and the structure above, don't rebuild everything at once:

1. Split brand from non-brand and negate brand terms across non-brand campaigns — this alone changes every number you're reading.
2. Label existing campaigns by tier (discovery / harvest / performance) and find where two tiers bid on the same term.
3. Run a full pass before you move budget — with the free Amazon audit, or by auditing it yourself in an afternoon.