Build a useful investigation brief
Build the ranking trace
For each representative query, record matched fields, relevance score if available, boost or pin, freshness, filters, sort, and final position. Include a query without the suspected rule. Keep the trace structural and avoid copying sensitive payloads.
Vary one input
Hold the query constant while removing or narrowing the boost, then hold the rule constant while changing sort or category context. Compare the displaced result and the top result. Negative changes matter because they show which promise the rule protects.
Write the ordering brief
State the first input that changes order, the user-visible consequence, and competing explanations. Recommend a scoped test with a pass condition. Do not claim that score alone explains usefulness; link the result to the search task it supports.
What to carry forward
A useful brief names the ranking input that changes order, the affected query context, and the tradeoff between business placement and textual match. Stop when the next experiment has a single variable and visible pass condition, or record why the order cannot yet be explained.
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