Tune Algolia search relevance with AI
Review searchable fields, filters and ranking signals against representative queries. Plan a change you can test and decide how to roll it back.
Tune Algolia search relevance with AI when a familiar query keeps returning an unexpected first result. A spare shade might outsell the desk lamp it fits, yet a shopper typing the lamp's full model name probably expects the lamp. Start with the query, the returned records and a clear account of what seems wrong.
The algolia-index-configuration skill reviews field priorities, ranking signals, filters and the reason for each setting. It separates constraints that every result must meet from preferences that merely influence position. Comparing a small set of representative searches makes it easier to spot a gain for one group that costs another.
What it does
- Check whether a title match competes with long descriptions
- Distinguish mandatory filters from optional preferences
- Review synonyms against genuine language equivalence
- Compare expected and observed results for common queries
- Plan one change with a baseline and rollback notes
Install and use
- Install the skill in a compatible agent. Provide a permitted settings export and sample queries. A live inspection or settings change needs authorised Algolia tooling.
- Ask, “Accessory sales are pushing spare shades ahead of an exact lamp model. Review these queries and suggest one change we can test.”
- Repeat the query set and record both improvements and regressions. Agree what would justify keeping the change.
FAQ
Source
Made by Algolia. Original skill: algolia-index-configuration, published on GitHub under the MIT licence.
View on GitHub