Turbo AI PM
Kits / product scenario · outline

How do we get people to the right thing when they only half know what they want?

Search & Discovery. Site search, catalog search, document and knowledge search.

Every decision below, with checkboxes.

The system

Query understandingWhat is the user actually asking for?
→
RetrievalWhat could answer it?
→
RankingWhich result deserves the top slot?
→
Results experienceWhat happens when nothing is good?

Three decisions to make first

This scenario is an outline for now — enough to start from the product problem and find the right kits. A full framework, like the one for Recommendations & Personalization, may follow.

What counts as a successful search when users rarely tell you?

Clicks, reformulations, and zero-result rates are indirect; pick the signal before optimizing.

What should happen when there is no good result?

A confident wrong result costs more trust than an honest "nothing found" with a next step.

Are we optimizing the first result, or the whole page?

Navigational searches need one right answer; exploratory ones need a good set.

Where the kits come in

A scenario is the product problem and the decisions it raises. Kits are reusable system-building workflows — one scenario draws on several, and each kit serves several scenarios.

Existing kits
RAG / knowledge assistantRetrieval quality — Recall@k, Precision@k, MRR — is most of the search problem.
Classifier / triage & moderationQuery intent detection and routing behave like classification.
Existing frameworks & tools

Other product scenarios