Advanced search with Elasticsearch, faceted filters and relevant ranking – customers find the product in a couple of clicks. In a catalogue with thousands of items, search is the most used feature on the site and the one that most affects whether a visit ends in an order.

Why plain search is not enough

A database substring match handles neither typos nor inflection nor synonyms, and cannot rank results by what is relevant to the customer. People type a colloquial term or misspell it and get an empty page even though the product exists. That is lost revenue which is very hard to spot in the statistics.

What we cover

  • Tolerance for typos, inflection and diacritics, including local search habits.
  • Synonyms and alternative product names, including the ones only customers use.
  • Faceted filters that show real counts and never lead to an empty result.
  • Ranking by relevance, availability and margin, not merely alphabetically.
  • Autocomplete offering products and categories while the user types.

Measuring and tuning

Search cannot be configured once and forgotten. We track zero-result queries, queries with high exit rates and the most frequent terms, then add synonyms and adjust weights accordingly. That cycle delivers more than deploying the technology itself.

We can run the new search alongside the existing one and switch over only after comparing results.

Frequently asked questions

Do we need Elasticsearch, or is the database enough?

For a catalogue of a few hundred items, well designed database search is often sufficient. A dedicated search engine pays off from thousands of products upwards, with faceted filters across many attributes, or when you need typo tolerance and synonyms.

Can we influence search commercially?

Yes. Weights can be set so available products, own brands or higher-margin items rank higher. The important part is not to overdo it – once results stop matching the query, both trust and conversion drop.