Seller guide
How should I structure Amazon Sponsored Products auto and manual campaigns?
Turn reviewed product economics and keyword evidence into separate discovery and control drafts without promising traffic or changing a live campaign.
Give each campaign one explicit job
Write the objective before naming a campaign. An automatic campaign can gather query and product-target evidence, while a manual campaign can isolate reviewed keywords or product targets with clearer bids and reporting. Do not call automatic traffic bad or manual traffic efficient by definition. Keep launch, rank, defense, profit, and research objectives separate so later results can be compared with the reason the campaign exists rather than with one account-wide average.
Group products only when their economics are comparable
Products sharing an ad group should have compatible relevance, price, contribution, inventory posture, and conversion context. A cheap accessory and a high-margin core product can make the same bid mean very different risk. Record the advertised SKU or ASIN, selling price, pre-ad contribution, target contribution after ads, and stock constraint beside every draft. Split products when one bid or search-term decision would otherwise hide a materially different economic ceiling.
Keep match types and discovery sources reviewable
Place exact, phrase, broad, product-target, and automatic discovery logic in named structures rather than one undifferentiated ad group. Preserve the source campaign and targeting expression when a search term is harvested. Check whether an exact target already exists before drafting another. A clean structure improves source review, but it does not guarantee traffic routing, auction priority, lower CPC, or better conversion; Amazon still controls eligible auctions and placement.
Design exclusions as a reviewed handoff
Use a negative-exact candidate for one observed query before considering a broader negative phrase. Search all supplied rows for profitable, branded, strategic, or product-accurate queries that the phrase could block. Record proposed scope, evidence window, spend, clicks, attributed orders, and a human-readable reason. Never convert a generated campaign plan into live exclusions automatically, and never infer shopper intent or product incompatibility from query text alone.
Launch from a paused draft and define the next comparison
Create the structure as a paused workbook with seller-reviewed names, products, targets, bid scenarios, budgets, and safeguards. Before activation, an authorized seller verifies current product facts, account settings, marketplace rules, inventory, and economics. Record the intended observation window and attribution basis so the next report can be compared like for like. Friday reads no Ads account, creates no campaign, places no bid, spends nothing, and changes no listing or advertising setting.
Tools that help with this
Prices below are exact x402 per-call prices. Human card checkout uses fixed $3, $9, and $29 bundles; follow a tool page to see its applicable bundle.
- Amazon PPC Launch Kit ($0.50) — Build a paused auto/manual campaign workbook from seller-reviewed inputs
- Amazon Search-Term Spend Waste Miner ($0.50) — Create reviewed negative and harvest queues from a supplied report
Every paid output is a draft for your review before you publish, appeal, or activate anything. Friday does not access your seller account.
More seller guides
- How do I calculate an Amazon PPC starting bid from product margin?
- How do I find Amazon ad search terms spending without orders?
- What Amazon PPC safeguards should I define before launching a campaign?
- When should I harvest an Amazon search term as an exact keyword?
- Competitors rank for terms your listing never mentions.
- Amazon backend search terms: why yours may not be indexing.
- Amazon backend search terms: bytes vs characters, explained.
- Why do my eBay listings get impressions but few views or sales?
- What do Walmart Search Insights rank gaps actually mean?