The Amazon PPC Campaign Structure We Use to Scale Brands Past $500K/Month

Most agencies run one campaign template across every client. It is efficient for the agency and a ceiling for the brand. The standard template, one campaign per product type split into auto, broad, and exact, works fine to roughly $500,000 a month. Past that it collapses under its own budget. Spend concentrates on a handful of head terms, the long tail starves, and TACoS creeps because every dollar competes with itself inside the same campaign. We inherited a brand stuck at $480,000 a month for 14 months on exactly this template, running a 19% TACoS it could not move. The structure, not the bids, was the cap. Bids tune a campaign. Structure decides the ceiling.

The unit of structure should be contribution margin, not the catalog. This is the single change that moves an account into its next phase of scale. Group ASINs into 3 margin tiers. Tier A is the high-margin engine, products above 35% contribution that can absorb aggressive spend and still print profit. Tier B is the 20% to 35% middle that scales carefully. Tier C is sub-20%, defended and harvested but never pushed. Each tier gets its own budget logic and its own TACoS ceiling, so a 12% TACoS that is criminal on a Tier C product is correct on a Tier A one. When we rebuilt the stuck brand around 3 margin tiers instead of 9 product lines, the same ad budget produced 27% more contribution within 2 months, because spend finally flowed to the products that could carry it. Margin decides the budget. The catalog does not.

Inside each margin tier, the account runs in 4 layers, and every campaign has exactly one job. Layer 1 is discovery: broad match and auto, capped near 15% of tier budget, whose only purpose is to find converting queries. Layer 2 is performance: the promoted winners from discovery, in tight ad groups, carrying the bulk of spend and judged on TACoS. Layer 3 is defense: exact-match branded and top-converting terms held at high impression share so competitors cannot rent your position. Layer 4 is expansion: DSP and Sponsored Display retargeting glance-view audiences from the first 3 layers. A query is born in Layer 1, earns its way to Layer 2, gets protected in Layer 3, and is re-engaged in Layer 4. That flow is the entire system.

Structure fails in practice when campaigns cannibalize each other. Two pieces of plumbing prevent it. Naming first. Every campaign carries a coded name: tier, layer, match type, ASIN group, for example A-L2-EX-hydration. It is not cosmetic. It is what lets one operator run a 7-figure account in 45 minutes a week instead of getting lost in 80 unlabeled campaigns. Negatives second. Discovery campaigns negative-out every term promoted to performance, and branded defense is negated everywhere else, so a single query is never bought twice in the same account. In one rebuild, adding disciplined negative architecture alone cut wasted spend by 22% in the first 30 days with zero change to bids. Plumbing is boring. Skipping it is expensive.

Budget should move on rules, not feelings. The accounts that scale past $1M a month route money weekly against a simple decision tree. If a Layer 2 cluster beats its tier TACoS ceiling by more than 3 points, it gets a 20% budget increase the next week. If it breaches the ceiling 2 weeks running, it gets cut 25% and the difference flows to the best-performing Tier A cluster. Discovery budget is fixed as a percentage and never raided to cover a shortfall, because starving discovery is borrowing next quarter's growth to patch this week. A brand we moved onto this logic scaled from $480,000 to $1.1M a month in 7 months while TACoS fell from 19% to 11.6%, and the bids were barely touched. Routing did the work.

Layer 4 is where structure compounds, and it is the layer most 7-figure brands run worst. They treat retargeting as an afterthought instead of the close. The discipline is sequencing. Display should only retarget audiences the first 3 layers already warmed: glance-viewers who did not buy, subscribe-and-save lapses, and competitor-ASIN viewers. Broadcasting Display to cold audiences at a 9% conversion rate is how brands convince themselves the channel does not work. Run it as the closer and the math changes. On a personal-care account, sequenced Display retargeting returned a 6.1x ROAS against a 1.4x on the untargeted campaigns it replaced, and it added 9% incremental revenue that the search layers were leaving on the table. The layer is not optional at scale. It is the part of the funnel that pays for the discovery you run up top.

You cannot rebuild a live 7-figure account overnight without paying for it in lost rank. Migration runs over 6 weeks. Weeks 1 to 2, build the new structure in parallel and tag every ASIN to a margin tier. No traffic moved yet. Weeks 3 to 4, shift 30% of budget into the new map, starting with Tier A where the upside is clearest and the risk is lowest. Weeks 5 to 6, move the remaining budget as the new clusters prove out, retiring old campaigns only after their replacements hold rank for 10 straight days. Expect a 1 to 2 week dip near the 50% mark. It is the model relearning, not a failure, and accounts that panic and revert at the dip are the ones that stay stuck at $500K. Structure is a decision. Make it once, correctly. The TACoS-versus-ACoS guide explains the ceilings each tier should run to, and the intent-signals piece covers the bidding layer that sits on top of this map.