Campaign Roles, Match Types, and Account Structure

Once the seller knows what traffic can cost, the next question is what each campaign is supposed to do.

Chapter 5 established the economic foundation of Merch PPC. The seller now knows that dashboard ACoS is not enough, break-even must be calculated from royalty, CPC has a ceiling, royalty groups change advertising tolerance, and PPC decisions need rules. That foundation matters because a campaign can only be judged properly when the seller understands what the traffic is allowed to cost.

But math alone does not create structure.

A seller can understand break-even and still build a chaotic account. They can launch campaigns with unclear goals, mix discovery and scaling traffic, combine seasonal and evergreen products, place broad and exact targets together, or create campaign names that make future review almost impossible. When that happens, the data becomes difficult to read even if the seller understands the math.

This chapter is about campaign purpose.

A campaign without a job produces data without meaning.

That principle should guide every campaign the seller creates. A campaign is not simply a container for products, keywords, bids, and budgets. It is a decision tool. It should answer a specific question, expose a specific type of traffic, or protect a specific part of the account.

A discovery campaign is not judged the same way as a scaling campaign. A defensive campaign is not judged the same way as a seasonal test. A second-chance campaign is not judged the same way as a winner graduation campaign. The same ACoS, CPC, CTR, or order count can mean different things depending on the campaign's job.

For example, a discovery campaign may be useful even if it is not immediately profitable, provided it reveals relevant search terms, ASIN signals, or products with early buyer response. A scaling campaign has a different standard. If its job is to increase volume on a proven product, then weak economics are more serious because the campaign is no longer only buying information. It is supposed to buy controlled volume.

This is where many Merch accounts become messy. The seller:

• launches a campaign to "test," but later expects it to scale;
• launches a broad campaign to discover search language, but then judges it like an exact campaign;
• adds proven products and untested products into the same structure;
• mixes apparel and mugs even though the royalty math, buyer behavior, and conversion expectations may be different;
• lets seasonal products sit beside evergreen products, then wonders why the data is inconsistent.

The campaign may be spending money, but it is not producing clean decisions.

The default operating model is simpler than the account usually looks:

1. Use discovery campaigns to buy information at controlled cost.
2. Classify what that traffic actually revealed at the product and term level.
3. Give underexposed products a second chance before calling them weak.
4. Move repeated signals into cleaner structures through harvesting and graduation.
5. Scale, defend, or clean up only after the campaign role is clear.

This chapter gives the structural language for that chain. Chapters 7 and 8 turn it into Lottery campaigns, ASIN buckets, harvesting, negatives, and winner graduation.

Before the seller builds more campaigns, they need to understand the basic building blocks of PPC structure.

The first distinction is between search terms and keyword targets. A search term is the language the buyer actually uses. A keyword target is the tool the seller bids on. Confusing those two leads to bad harvesting, bad negatives, and bad optimization decisions.

The second distinction is between discovery and control. Automatic targeting, Broad match, Phrase match, and Exact match are not simply different settings inside the dashboard. They represent different levels of freedom given to Amazon. More discovery gives the seller more possible signals, but also more noise. More control gives the seller cleaner traffic, but less ability to discover what the market is actually typing.

The third distinction is campaign role. A campaign should have a job before it has a budget.

Some campaigns are built to discover. Some are built to scale. Some protect proven products. Some test seasonal opportunities. Some give underexposed products a fair chance. Some isolate cleanup work inside an old or messy account.

A seller who cannot name the campaign's role will not know how to judge the campaign's data.

The fourth distinction is advertising layer. Sponsored Products should form the foundation of the system because they are the cleanest starting point for product testing, search intent, and buyer response.

Sponsored Brands, display-based ads, and Product Targeting can be useful later where they are available and eligible, but they should not replace the foundation.

Advanced layers should usually amplify or protect evidence that already exists, not compensate for products that have not earned attention.

The final distinction is operational structure. Naming conventions and portfolios are not cosmetic.

They help the seller search, filter, audit, pause, compare, and manage campaigns without opening every campaign manually. A good name should tell the seller what the campaign is, what it is trying to do, and what type of products or traffic it contains.

A portfolio can help group campaigns by purpose, season, product type, marketplace, or risk.

The purpose of this chapter is not to build the full PPC machine yet. Lottery campaigns, ASIN buckets, harvesting, negatives, and weekly optimization will come later.

First, the seller needs the structural language of PPC. They need to know what the buyer typed, what the seller bid on, how much freedom the match type gives Amazon, what job the campaign performs, and how the campaign will be organized for future review.

By the end of this chapter, the seller should be able to look at any campaign and answer five questions:

• What traffic is this campaign buying?
• How much control does it have?
• What job is it supposed to perform?
• How should its data be judged?
• How will the seller find it again during an audit?

Only after those questions are clear should the seller start building larger discovery systems. A large catalog does not need more random campaigns. It needs campaigns with jobs, match types that fit those jobs, and a structure that keeps future decisions readable.