The leaks I can see from the public listing
Title, bullets, A+, reviews, brand defense, and obvious ad-real-estate problems. Storefront URL only — no Seller Central login until we agree to work.
Services · Wachstum
Keyword-Intelligence
Volume is vanity. Harvest is how ACOS actually falls.
Best fit: catalogs stuck on the same 10 expensive terms
Loop
Harvest runs every week, not once a quarter
Exact
Converters leave discovery and get a bid of their own
Negate
Losers cannot sneak back into spend
The keywords that print profit are rarely the ones with the highest search volume. Harvesting means reading search-term reports every week, promoting converting queries into exact, and negating the rest.
Auto, broad, and category targeting exist to discover those queries. Exact campaigns exist to scale them. Mixing the two in one campaign is how accounts slowly bleed.
I also reverse-lookup competitor ASINs where your listing actually wins the click — price, reviews, or a clearer use case — so harvest is not only your own search terms.
Who this is for
Brands whose search-term reports are unread, whose exact campaigns are thin, and whose ACOS is a blend of winners hiding next to losers.
Who it is not for
A one-time “keyword list” for a copywriter. Harvest is a weekly loop, not a spreadsheet you buy once.
How the work actually runs
01
Mine
Read search-term reports across auto, broad, and category. Tag sold, wasted, and too-new-to-call.
02
Promote
Anything that converts at an acceptable ACOS gets an exact campaign and a real bid.
03
Negate
Junk is negated at the right level so discovery stays cheap and exact stays clean.
04
Repeat
Same loop next week. A static list is how accounts go stale in a month.
What is in the workstream
- Weekly search-term mining
- Exact-match promotion rules
- Negative keyword architecture
- Competitor ASIN reverse lookups
- Long-tail vs head-term split
- Harvest tables in the weekly log
- Query-level win/loss notes
- Cleanup of duplicate targeting
Week one
I pull search-term history, tag converters vs waste, stand up exact homes for anything that already sells, and build the first negative architecture.
Every week after
Discover → prove → exact → negate. Converters get a campaign. Losers get filed so they cannot crawl back into spend.
On this workstream’s desk
Named seats, not a logo wall. Each page says the job, the rail, and what still has to happen in Seller Central or the Ads Console.
Helium 10
H10, Cerebro, MagnetReverse-ASIN and seed-keyword research (Cerebro, including Magnet) so listings and exact campaigns start from real demand — then the search-term report decides.
Jungle Scout
JS, Keyword ScoutKeyword Scout, product database, and sales estimates as a second research lens — especially when Helium 10 and Jungle Scout disagree on volume.
Claude
Anthropic ClaudeLong-context read of search-term exports, change logs, and playbooks — drafts the founder note. Does not bid, does not log into Seller Central.
Codex
OpenAI CodexScripts that clean Ads Console exports, join Sellerboard P&L to search terms, and build repeatable harvest tables — not a secret bidding API.
FAQ
Is this a one-time keyword list?
No. Harvest is a weekly habit. Amazon’s query set moves. A static list is how ACOS creeps back.
Do you use third-party keyword tools?
Yes — Helium 10 (Cerebro) and Jungle Scout (JS) as research seats. They suggest. The source of truth is still your search-term report: queries you already paid to test. Indexed on the operator stack.
What about branded terms?
Branded exact is defense, not harvest. Those sit in the brand-defense layer so conquesting cannot cheap-out on your name.
Will this replace PPC management?
It is the brain of PPC. Most brands run it inside management. You can hire it as a workstream if the rest of the account is already clean.
This workstream on the free review
You do not need to hire it to find out if it belongs on the account. Send the public store link. I tell you whether this lever is leaking, whether I am the person to run it, and what week one would be.
Whether this desk is a fit
If I cannot show the leaks, I decline. You are not added to a roster. A “no” is a complete answer, not a follow-up sequence.
What week one would be
Harvest, defense, listing pass, or “not yet.” Named in writing — not a 40-slide deck and a 12-month PDF.
Capacity is limited because I work the accounts myself. The review is how we decide if this workstream belongs on yours.
Request a free reviewRelated workstreams
Amazon PPC-Management
Tägliche Gebote, Budgets und Suchbegriffe über Sponsored Products, Brands und Display.
Brand Analytics
Market-Basket und Search-Query-Performance als Desk-Input, nicht als Slide.
Produktlaunches
Strukturierte Launches mit Exit — nicht ewig „aggressiv“.
Listing-SEO
Titel, Bullets und Backend, die bezahlte Queries tragen — nicht nur Suchvolumen.