Decision Brief

A Decision Brief is the AI Analyst’s output. It is a written recommendation rather than a score, and it is structured so you can disagree with it on specific grounds.

The four parts

The action

What the brief proposes to do, written as a sentence. Where the action needs one, it carries a match type as well — creating a Negative Exact is a different proposal from creating a Negative Phrase.

The reason

Why the Analyst reached that conclusion, in plain language. This is the part you argue with. If the reason does not hold for your business, the brief is wrong, and you should not apply it.

The evidence

The rows behind the reason. Not a summary of them — the actual records the conclusion rests on, so you can check whether the sample is thick enough to carry the claim.

The period and scope

The period the evidence came from, and the decision scope it was read at. A brief without these two would be unfalsifiable, which is why they are part of the brief rather than a setting somewhere else.

What a brief is not

A brief is not a change. Nothing in it has reached your Amazon account. See Preview Changes for what stands between the two.

A brief is also not a prediction of a result. It states what the evidence supports, not what will happen if you follow it.

Reading one critically

Three questions worth asking every time:

1. Is the period right for this decision? A product you launched six weeks ago has very little history to break a pattern against.
2. Is the scope right? Group evidence is the wrong evidence for an ASIN with its own shopper.
3. Is the sample thick enough? Four observations can produce a confident-looking reason.