How the analysis works
This page explains the analysis at the level you need to judge its output. It is not a description of the implementation.
What it reads
Section titled “What it reads”| Source | What it contributes |
|---|---|
| Product Listings | What each product is |
| Advertising keywords | What you decided it should be found for |
| Search terms | What shoppers actually searched |
| Performance history | How targets behaved over time |
| Change history | What was altered in the account, and when |
Everything comes from the Amazon account you connected. Nothing has to be imported or modelled by hand.
What it looks for
Section titled “What it looks for”Targets that changed, not targets that are bad. A target that has always converted poorly is a different problem from one that stopped converting last month, and only the second gets a brief.
Each search term and targeting is read on its own rather than inside an account-level average, and recent performance is compared with the stretches before it — so a fortnight that broke does not disappear into a six-week average.
Where an ASIN’s own record is too thin to answer the question, the Product Group’s evidence is read as one context. Siblings are never compared one against another. See Decision scope.
When a break is found, the account’s change history around that point is searched for the cause — a bid change, a budget change, a negative — and that cause is attached to the brief rather than left for you to reconstruct.
What comes out
Section titled “What comes out”A Decision Brief: the action, the reason, the evidence behind it, and the period and scope it applies to.
Not the same thing as an Automation Data Window
Section titled “Not the same thing as an Automation Data Window”A rule evaluates thresholds over one Data Window you chose. The Analyst compares consecutive periods to find a change. Do not read one as though it described the other.
What it deliberately does not do
Section titled “What it deliberately does not do”- It does not compare siblings one by one. Different sample sizes across variations produce differences that are noise, and ranking them invites acting on that noise.
- It does not lower its standard of proof to produce output. No brief is a valid answer.
- It does not act. See Preview Changes.
- It does not read Inventory. How much stock you hold neither produces a brief nor suppresses one. What you are about to run out of is a real question and it is answered on its own side of the product, in Inventory and Supply.
Why briefs appear for some targets and not others
Section titled “Why briefs appear for some targets and not others”Usually because nothing broke. A target behaving the way it always has does not need a decision, and a target with too little history has no pattern to break yet.
VendlenPrivacy PolicyTerms of ServiceContact
© 2026 HELODA PTE. LTD. Not affiliated with or endorsed by Amazon.

