Campaign and ad-group structure — the blueprint AMALYZE deploys.

This is the consolidation episode. Twenty-two earlier episodes covered individual decisions — formats, targeting types, bid strategies, match types, naming. This one assembles them into the structure we actually deploy when we set up an Amazon advertising account from scratch.

The four-campaign skeleton, per ASIN cluster

An ASIN cluster is the unit of structure — a parent ASIN plus its variants, or a tightly-related set of standalone ASINs targeting the same shopper. Every cluster gets four campaigns:

1. SP Auto — Harvesting. Small daily budget. Down-only bidding. Per-bucket bids per episode 15. Purpose: discover converting search terms. Search-term report mined weekly.
2. SP Manual Exact — Profitable. The conversion engine. Holds harvested winners and head terms. Down-only at launch, up-and-down once mature. The campaign that holds most of the long-run spend.
3. SP Manual Broad / Phrase — Coverage. Optional second harvester. Covers modifier and synonym space the auto does not catch. Negative-keyword list maintained weekly.
4. SB / SD — Brand & Defence. SB brand-defence on the brand name itself, SD remarketing on views-not-purchased. Small budget, high ROAS, structural.

Ad-group split rules inside each campaign

Within a manual campaign, split ad groups by:

• Match type. One ad group per match type. Never mix exact and phrase keywords in the same ad group.
• Funnel stage of the keyword. Head terms in one ad group, mid-funnel modifiers in another, long-tail in a third. Bids will differ; lifetime CVR will differ; the split keeps reporting clean.
• Goal cluster. Brand-defence keywords in their own ad group even if they live in a manual campaign — different bid logic.

Conquesting and category targeting — separate campaigns, always

Conquesting keywords (competitor brand names) and product / category targeting campaigns get their own campaigns — never folded into the four-campaign skeleton. They have different bid economics and different reporting needs. Mixing them dilutes both.

Why this structure scales

The four-campaign skeleton replicates per ASIN cluster. Twenty clusters = eighty campaigns. The naming convention from episode 06 lets you grep, filter and roll up reporting at any level — by format, by goal, by product line, by marketplace. The pieces are repeatable; the dashboards are stable across account size.

When to break the structure

Two cases. First, an ASIN with a margin profile so different from the rest of the catalog that it needs its own ACOS targets — give it its own portfolio rather than its own naming exception. Second, a major event push (BFCM, Prime Day) where the campaigns need date-bounded versions; build them as a sibling set under a "BFCM 2026" portfolio rather than rewiring the always-on campaigns.