Project - Team Search

Overview

As a member of the Search Team, my primary responsibility was developing and improving search functionalities. Beslist.nl attracts 500,000 to 600,000 unique visitors per day, so performance and accuracy were critical.

Every change was A/B tested to minimize financial risk and verify positive results.

Key Projects

Search Optimization

The core search experience was continuously refined to handle the scale and diversity of a 25-million-product catalog. I implemented automatic suggestions, faceted search, and "similar products" recommendations, each validated through A/B testing to ensure measurable improvements in click-through and conversion rates.

Search Management Tooling

To give the editorial and product teams direct control over search quality, I built internal tools for managing stop words, synonyms, and manual redirects. This reduced the team's dependency on engineering for day-to-day search tuning and allowed faster responses to emerging trends and seasonal shifts.

Visual Query Analyzer

I developed a visual debugging tool for ElasticSearch that made it possible to trace exactly what happens when a search term is processed. It replayed every normalizer, filter, and tokenizer in the ES analysis chain step by step, showing exactly which terms the search engine was actually working with at each stage.

• A search term like "PS4" might be tokenized into "PS" and "4", then "PS" gets discarded for being too short — leaving the engine with nothing meaningful to match. Without the tool this was a mystery; with it, the cause was obvious at a glance.
• Once a problem was identified, our data analyst could immediately resolve it by adding a synonym (e.g. mapping "PS" to "Playstation") through the search management tooling — no developer involvement needed.
• This shifted search debugging from an engineering bottleneck to a self-service workflow, letting the team diagnose and fix issues in minutes instead of days.

Turnover Prediction & Smart Bidding

Managing a catalog of 25 million products required intelligent automation to ensure profitability. I helped develop logic that analyzed historical data to predict the turnover potential of individual products. This insight enabled the Marketing team to determine precise budget caps before launching campaigns, ensuring that ad spend was focused on profitable items and preventing overspending on low-performing stock.