Every report is a shadow
Each report flattens a three-dimensional account onto a single plane. Most of the market works from one shadow at a time, which is why two operators can stare at the same ACOS and prescribe opposite fixes.
Amazon advertising, read in three dimensions
Every report is a shadow.
The advertising reports are one shadow. Search Query Performance (SQP) is another. Amazon Marketing Cloud (AMC), where you query the data yourself, is a third. Each counts something different, and a number like ACOS is only a reading off one of them. Look at one and you can only guess the shape. Connect them, and a finite set of data starts to show the real thing.
Connect the reports, connect the AI, until a finite set of data shows the answer's real shape.
Dashboard vs. map
Metrics like ACOS and TACoS are dashboard readings: lagging numbers that tell you something moved. Being matched (eligible to be shown for a search) and the customer cycle are the map: they tell you where you are and which road leads where.
Why one operator can do it now
Joining Amazon Marketing Cloud query results with the standard reports used to take an analytics team. With AI handling the joining, cleaning and first-pass checks, one operator can run the investigation and keep the judgment.
I don't hand out answers.
I share a view that has been tested hard on real accounts and could still be overturned tomorrow. Every claim here states its premises, says whether it comes from my own measurements or is still a hypothesis, and names the result that would prove it wrong.
Discussion, data, logic, verification. In that order, and on repeat.
Ad reports
AMC
SQP
The account
Three ways in
Read, check, then run it yourself.
The free material is meant to be true enough to stand on its own. The paid course is where you get the full steps to reproduce it on your own account.
New to the terms? The glossary sorts every word on this site into metrics, reports and data sources, and our own terms.
NOTES
Short reads
One popular claim or one reading trap per note. Premises laid out, the conclusion, one example and the condition that would overturn it.
Browse notes → LEDGERPublic ledger
Every verdict in one table, including the ones we later reversed. Open questions are listed too, so you can bring data to them.
Open the ledger → COURSEThe full series
Six modules from the shadow problem to designing what you watch. Every unit includes the step-by-step method, queries and checklists.
See the syllabus →
Latest notes
Recently published
- Our own claim: "It's not ACOS, it's TACoS" — does it still hold?⚠️ Partly
The principle holds — judge ads by what they do for the account. The ratio is a gauge; proving ads drove growth takes holdouts, controls, and placebo arms.
- Reading trap: the Business Report doesn't match the number you saved last month— Reading trap
Revised for weeks: units drift down, sessions drift up. Store the pull time; re-pull before comparing; don't assume a ratio is safe — conversion rate, with its two sides moving apart, is hurt most.
- "Lower ACOS is always better"❌ Doesn't hold
Harvest campaigns are judged by cost per order; ACOS is that cost divided by order value, so it works only while prices hold steady. Neither is a goal for every campaign. A low ACOS can mean the ad stopped finding new customers or ran out of budget.
- "Raise the bid and you'll get more impressions"⚠️ Partly
When the budget has room, a higher bid usually buys more impressions, but not reliably: your own campaigns can take volume from each other, and capped budgets turn higher bids into fewer clicks. Bid sets how actively you reach shoppers you can already be matched to, not who they are or how ready they are. Changed from ❌ on 2026-10-13: ❌ claimed more than the evidence shows.
From the public ledger
Popular claims, judged.
✅ holds · ⚠️ partly · ❌ doesn't hold · 🔶 untested · — reading trap (a way the numbers mislead) · 💡 viewpoint (an argument, not tested). Experience means measured on my own accounts; Hypothesis means the framework is formed and still being tested.
Loading the ledger…
Written for
Sellers with real volume and real data, in-house ad leads, and operators who run accounts for brands. People who already know the buttons and want to know why the numbers move.
Not written for
Anyone looking for a setup checklist, a guaranteed result, or a tool to buy. There is nothing here to sell you except the course, and it promises a method, not an outcome.
Get the notes by email
New notes and ledger reversals, nothing else.
No fixed schedule. You hear from me when there is a new note or when a verdict changes.