Amazon PPC Campaign Structure at Scale (2026)

There is no correct number of Amazon PPC campaigns. There is only a structure you can still reason about six months from now, and one you can't.

Most accounts fail the second test. They start clean, accumulate campaigns as products launch, and end up somewhere nobody can explain — three campaigns bidding on the same keyword, budgets set in a hurry two years ago, and no way to tell which change caused which result.

This is about structure that survives growth.

What a campaign actually controls

Worth being precise, because half of structure decisions come down to this.

At campaign level you set: daily budget, bidding strategy, placement adjustments, start and end dates.

At ad group level: the default bid, and which products are advertised.

At keyword or target level: the individual bid.

So the question "should this be its own campaign?" really means: does this need its own budget, its own placement adjustments, or its own bidding strategy? If the answer is no, it belongs in an ad group.

That single test resolves most structure arguments.

The three-layer structure

For almost any catalogue, this holds up:

Layer 1 — Discovery. One automatic campaign per product (or per tight product group). Amazon chooses the targeting. Its job is to find search terms you would not have guessed, and it is genuinely good at that. Layer 2 — Control. Manual campaigns holding the search terms discovery proved. You set the bid per keyword, so this is where profit is actually managed. Layer 3 — Defence. Campaigns targeting your own listings and close competitors by ASIN. Optional, and only worth it once layers 1 and 2 are stable.

The flow between layer 1 and layer 2 is the engine of the whole account: discover in auto, promote winners to manual, and negate the promoted term in the auto campaign so the two stop covering the same ground.

When a product needs its own campaign

Give a product its own campaign when it needs its own budget. That's the practical trigger.

If a strong seller shares a campaign with four weak ones, the weak ones consume budget the strong one could have used, and you cannot fix it with bids because budget is set a level higher. That's a real cost, and it's the main argument for splitting.

For Merch and print-on-demand the argument goes further, down to ad group level: bids are set per ad group, so several designs sharing an ad group share a bid. A design converting at 15% and one converting at 2% get treated identically, indefinitely. One ASIN per ad group is the fix, and we covered the mechanics in the lottery campaign setup.

The counter-argument is real too: more campaigns means each one accumulates data more slowly, and a campaign with four clicks a week never tells you anything. Below roughly 40 clicks per product per month, consolidation beats granularity.

Naming, which matters more than it sounds

At fifty campaigns you can hold the account in your head. At five hundred you cannot, and the only thing standing between you and chaos is a naming convention you can sort and filter.

Something like:

SP_Auto_B0B38RKXRM SP_Manual_Exact_B0B38RKXRM SP_Manual_Broad_B0B38RKXRM SP_PT_B0B38RKXRM

Type, match, ASIN — in a fixed order, every time. It means you can sort the bulk file alphabetically and every campaign for a product sits together, and you can filter to _Auto_ across the whole account in one step.

Two constraints worth knowing before you commit:

Campaign names must be unique across the account. Amazon will reject a duplicate, so the ASIN (or another unique token) has to be in the name.
Renaming is safe; restructuring is not. Changing a campaign's name keeps its ID and its history. Moving products between campaigns does not.

We built a free campaign naming generator that emits a consistent scheme if you'd rather not design one.

The cost of restructuring

This is the part that catches people, and it's worth being blunt about.

When you move a converting product into a new campaign, its conversion history does not move with it. Amazon prices your clicks partly on how well an ad has performed, and a new campaign starts with none of that. A better structure will underperform the messy one it replaced for roughly three to four weeks.

The failure mode is predictable: seller restructures, performance drops, seller concludes the new structure is wrong, restructures again, and pays the cold-start cost twice.

So: restructure rarely, deliberately, and then leave it alone for a month. If your old campaigns are only paused rather than deleted, re-enabling them is far cheaper than rebuilding them, because the IDs and history are intact.

A structure that scales

For a growing catalogue:

One auto campaign per product, one ASIN per ad group.

One manual campaign per product per match type — exact and phrase to start; add broad only once exact is profitable.

Budget at campaign level so your best products cannot be starved by your worst.

A naming convention from day one, because retrofitting one across 400 campaigns is a bad afternoon.

Negatives flowing auto → manual every time you promote a term.

Placement adjustments set per campaign once you have enough data to justify them — not copied from a default.

At scale, none of this is manageable in the ads console. The bulk file is the only practical interface, and building the manual campaigns from auto-campaign winners is mechanical enough to automate — which is what our auto-to-manual builder does: it reads the bulk file, finds the converting terms, and emits the manual campaigns with negatives already written back into the auto campaign.

Amazon PPC Campaigns: A Structure That Survives 500 of Them