Aglet

Learn from search boost conflicts and ordering tradeoffs

A boost dispute often reveals a missing contract between business placement and search relevance. Preserve the original query and policy, compare later behavior, and turn the conflict into a repeatable review question. Learning should improve the decision process without declaring one ordering signal universally correct.

Keep the lesson for the next incident

  1. Keep the original tradeoff

    Record the query, result order, active rule, expected job, and what the displaced result offered. State what was unknown. This lets later reviewers distinguish a deliberate campaign choice from an accidental ranking side effect.

  2. Compare later ordering

    Review new query examples after the rule or copy changes. Look for unrelated phrases receiving the same treatment, useful results displaced, or confusion reduced by a label. Preserve exceptions and avoid using one improved query as proof the rule is safe.

  3. Add one tuning question

    Adopt a prompt such as “which user job does this rule protect?” or require an out-of-scope query in every ranking review. Assign the practice and a revisit date. Define the observation that would show the added question prevented another conflict.

What to carry forward

The lesson should connect a real ordering conflict to one repeatable ranking-review adjustment. Stop when another reviewer can test the question on a new boost and see its boundary. Keep uncertainty where later behavior cannot separate rule quality from changing query mix.

Technical background: Elastic documentation.

Keep the decision with the work.

Use a Work Item in Aglet to record the problem, the evidence you have, and the next decision. Add an owner and priority, then keep updates in the discussion so the next person can follow the reasoning.

Create an account See the product workflow