Search and catalog
Search and filters that actually find the product
On-site search is your best salesperson or your worst. If "samsunk phone" returns nothing, the customer leaves. I set up Elasticsearch so it understands shoppers and knows about your business: stock, margin, season.
Signs search is losing money
- Lots of zero-result queries, and nobody reads them.
- Filters take over a second.
- Out-of-stock items rank above available ones.
- Synonyms, typos and part numbers with and without spaces are not matched.
- 3,000 results for a broad query with no sensible order.
When you don't need Elasticsearch
If the catalog is small and SQL copes, a separate search cluster is just more to maintain. Sometimes a SaaS search is the better call, and I'll compare the options honestly.
When Elasticsearch is overkillWhat I do
Search log review
What people look for, what they don't find, where they drop off.
Mappings and analyzers
Morphology, synonyms, typos and keyboard layout.
Facets and aggregations
Answers in milliseconds on tens of thousands of SKUs.
Ranking
Accounts for stock, margin and merchandiser rules.
Autocomplete and suggestions
The shopper sees the product before finishing the query.
Zero-downtime reindexing
With aliases, and sync with Bitrix or the ERP.
How the work runs
- 1
Log and 20 queries
What people search for and where search answers badly.
- 2
Review
On its own or as part of the architecture audit.
- 3
Implementation
A separate fixed-price stage with zero-downtime reindexing.
- 4
Follow-up
Zero-result rate and filter speed compared before and after.
Timing and price
- Timing
- The review fits into the audit, implementation is timed after it
- Price
- A search review can be added to the architecture audit. Implementation is priced as a separate fixed stage.
Proof
Proof
A 28,000-SKU catalog: Next.js on top of 1C-Bitrix
The PHP backend stayed in place with orders, pricing and integrations. The storefront and search moved to Next.js and Elasticsearch, and the SEO structure carried over.
Read the case studyQuestions
- Can you add Elasticsearch to our current Bitrix without going headless?
- Yes. Search and filters can move on their own while the storefront stays.
- Who maintains the cluster?
- We decide upfront: your hosting, a managed service or my monthly engagement. A cluster with no owner is a bad idea.
You work with me
Ivan Ponamarev, senior full-stack engineer, 10+ years in production. You talk to the person who makes the technical decisions and writes the critical code.
Send me 20 queries your search handles badly
I'll tell you what to do about them.