Testing Konga's new in-house search: 15 of 15 recommendations implemented
Konga built its own search and merchandising platform (Konga Search, or KSS) to replace Algolia. I tested it as the person who would use it every day, and turned the gaps into prioritised recommendations for the tech team.
- Role
- Product evaluator (business and merchandising lead)
- Company
- Konga
- Timeframe
- Sep 2025
- Search
- Product evaluation
- Requirements
- Usability testing
- 15 of 15recommendations implemented
- 2 roundsof staging tests
- 6product areas assessed
Context
Konga Search Service (KSS) is an internal search and merchandising platform built to replace Algolia. It covers product discovery, search relevance, merchandising rules and search performance analysis. I had been running Konga’s search on Algolia, so I knew exactly what the team would need from the new tool on day one.
The problem
A replacement only works if the people using it can do their jobs at least as well as before. Missing features in a search tool don’t just slow the team down. They put the wrong products in front of customers.
My role
I evaluated KSS from the business and user side: I tested the staging environment, mapped it against our real merchandising workflows, documented gaps and recommended fixes to the tech team.
What I did
- Tested the core areas end to end against our Algolia workflows: search analytics, merchandising rules, synonyms and plurals, date-range behaviour, team permissions, campaigns and navigation.
- Compared each workflow with how we did it on Algolia, so every gap came with a clear “this is what we need” reference.
- Documented 15 prioritised recommendations, each with the issue, its business impact and the suggested fix.
- Defined the must-have requirements for merchandising parity with Algolia.
- Ran a second round of staging tests after the first fixes, tracked what was resolved and pushed on what was still blocking.
- Proposed search-quality metrics for launch: “searches without clicks” and a weekly search-analytics report.
Key decisions and trade-offs
- Focused on the Rules section first, because that’s where merchandisers spend most of their time.
- Separated must-haves from nice-to-haves. Multiple query conditions per rule; plural, synonym and typo handling; adjustable pinned-product positions; and bulk rule import (CSV/JSON) were critical. A tidier action dropdown was flagged as polish, not a blocker.
- Asked for visual previews of pinned products, because publishing a rule that puts a microwave under a “bag” search is the kind of error customers notice straight away.
Outcome
The tech team implemented all 15 recommendations. The feedback gave them a prioritised list grounded in real merchandising work, and a clear bar for when KSS would be ready to replace Algolia. KSS is now live.
What I’d do next
Now that KSS is live, track those metrics, plus zero-result queries, week by week, so the team can see whether relevance is improving.