Learn
The decisions that cost the most.
Short pieces on the defaults that quietly cost money in an Amazon business, from the ads account to the reorder, written from running one. Each ends where the behaviour is specified: the documentation.
- What these are
- One costly mistake each, with the arithmetic
- What they are not
- A substitute for the docs. Named comparisons live in Compare
- Figures
- Worked examples, never a measured account
Articles
Profit
Neither ACOS nor TACOS knows what the product costs you
What ACOS and TACOS each measure, why they move apart, and how a product can look efficient on both and still lose money.
7 min readSpecified in Profit Intelligence
Amazon DSP
A 100% delivery rate can hide an order about to run dry
What Amazon DSP delivery rate compares, why an order spending too fast reads the same 100% as one on pace, and the arithmetic for both.
6 min readSpecified in DSP Delivery rate
Amazon DSP
Last week's DSP results are not finished yet
Why Amazon DSP sales, ROAS and ACOS for the last 14 days keep changing, what judging a line item on them costs, and which cost to divide by.
6 min readSpecified in How fresh DSP numbers are
Inventory
A reorder has to last until the next one lands
The reorder formula worked line by line: the weeks it has to cover, the stock and orders it subtracts, and the rounding that comes last.
7 min readSpecified in Order Planning
Inventory
Days of supply counts the shipment before it lands
Why a days-of-supply figure that includes inbound can hide a stockout, how to find the gap week by week, and what to compare cover against.
7 min readSpecified in The weekly stock projection
Inventory
A sales rate measured while you were out of stock measured the shelf
How to calculate a sales rate that still means demand when the window held a stockout or a promotion, and why one window is not enough.
7 min readSpecified in Selling rate and buffer weeks
Inventory
Short of the factory minimum, add the units that sell soonest
The two minimums a factory order has to clear, and how to choose the extra units by how long they would wait to sell.
6 min readSpecified in Purchasing and Production
Search terms
A keyword's ACOS is an average of searches you never chose
What separates a keyword from the searches it buys, how one keyword on target can hide a search that never sells, and why the fix is per search, not per bid.
5 min readSpecified in Search terms workspace
Search terms
Twenty clicks without an order proves less than you think
The chance arithmetic behind a click threshold, the spend it implies, and why a quiet fortnight is not a verdict on a search that sold for a year.
6 min readSpecified in Search terms workspace
Search terms
A negative phrase also blocks the searches you have not seen yet
What each negative match type blocks, what a one-word phrase costs in searches that sell, and why word order can mean two phrases instead of one.
6 min readSpecified in Harvest, negate and unnegate
Search terms
Harvesting a search term adds a keyword. It moves nothing.
What harvesting changes and what it leaves running, how many orders a term needs first, and how to price its opening bid.
6 min readSpecified in Import rule
Bidding
A bid formula is only as good as the orders behind it
Where the target ACOS bid comes from, how far one order moves it, and the two brakes that stop a wrong estimate moving the bid too far.
6 min readSpecified in Bidding rule
Bidding
A bid rule needs a range where it does nothing
Why a bid rule needs a do-nothing zone, why its lines should move with the order count, and why a high-ACOS keyword is cut before it is paused.
5 min readSpecified in Bidding rule
Bidding
Judge dayparting by blocks of hours, not single hours
How few orders an hour of the week holds, why an hour above target can still make money, and how to cut hours only while the campaign runs expensive.
5 min readSpecified in Dayparting rule
Search terms
Your negative keyword is not blocking what you think it is
Why a term you blocked keeps taking spend, how to read the evidence before you block again, and what unnegation is for.
6 min readSpecified in Negation and unnegate rule
Bidding
An ACOS threshold cannot tell $8 from $800
What a fixed ACOS rule actually measures, why it fires hardest on the targets that matter least, and the arithmetic that fixes it.
7 min readSpecified in Loss Budget Guard
Evidence
Ninety days will hide the fortnight that broke
How averaging a window buries a target that stopped converting, and when to ask a question instead of setting a rule.
6 min readSpecified in Evidence and periods
Every figure in these articles is a worked example written to explain a calculation. None of them is measured performance from any account, ours or anyone else's.
The reasoning is the product.
Read a Decision Brief against the account you already run, and apply it when you agree with it.