Problems with Bitrix Search: Slow and Irrelevant
Out-of-the-box search on Bitrix often performs poorly. Typical symptoms: index not built — nothing found; all pages found, including administrative ones; on a catalog of 50,000 items, search freezes for 30 seconds. We have configured dozens of projects with varying data volumes — from online stores to corporate portals. Each situation requires its own approach. Let's break down the key mechanisms and techniques that can speed up search by 4 times and improve relevance by 60%.
Understanding how the search module builds the index is critical for diagnosing problems. The index is stored in tables b_search_content (content), b_search_content_stem (stem index for morphology), and b_search_tag (tags). When saving an information block element, the OnSearchIndex event fires. If the index is not updated automatically, check agent settings and the event log for errors.
The consequences of slow or irrelevant search are critical for an online store: customers cannot find the desired product and leave to competitors. For a corporate portal, it reduces employee productivity. We have seen projects where after basic search setup, conversion increased by 15% and search time dropped from 10 seconds to 0.5. Server power savings reach 30% due to reduced database load.
How to Configure Indexing Sources?
Configure sources (Settings → Search → Module Settings):
- Save search history — useful for analytics
- Morphology — enable for word-form search (buy, bought, buying)
- Characters in index — minimum word length for indexing (recommended 3)
Select indexed information blocks (Settings → Search → Reindexing):
- Choose information blocks to index
- Specify which properties to index
- Exclude technical pages and administrative content
Search component:
$APPLICATION->IncludeComponent('bitrix:search.page', '', [ 'SITE_ID' => 's1', 'SEARCH_CACHE_TIME' => 3600, 'DEFAULT_WORD' => '', 'MINIMAL_QUERY_LEN' => 3, 'USE_SUGGEST' => 'Y', 'arrFILTER' => [ ['MODULE_ID' => 'iblock', 'PARAM1' => 'catalog', 'PARAM2' => 5], ['MODULE_ID' => 'iblock', 'PARAM1' => 'content'], ], ]); Why Indexing Fails on Large Catalogs?
For large catalogs (over 50,000 items), a full reindex can take hours. The problem is compounded if all information block properties are indexed indiscriminately. The solution is incremental indexing via agents. We configure agents that process 500 items per step. For a project with 200,000 items, full reindex time dropped from 3 hours to 15 minutes. Additionally, we enable tagged caching of the search component — response speed increases by 4 times.
Relevance Optimization: Weight Configuration
Bitrix uses built-in stemming for the Russian language. With morphology enabled, a query for "red dresses" will find "red dress" and "red dresses". The morphological index is built in b_search_content_stem. Result relevance is controlled by weights: a word in the title — high weight, in the body — medium, in tags — medium. Additional weights can be configured via b_search_content.PARAM1 parameters.
Proper weight configuration is key. A word in the title should weigh 5 times more than in the body. Values are set in module settings or in component code. Also important is to exclude irrelevant pages: personal account, cart, administrative sections. For catalogs with 100,000 items, we use indexing profiles for different content types.
Comparison of Standard Search and Elasticsearch
| Criterion | Standard Search | Elasticsearch |
|---|---|---|
| Indexing speed | Slow for >50k items | Up to 10x faster |
| Relevance | Configurable weights | Flexible settings + machine learning |
| Morphology | Built-in stemming | Dictionaries + language analyzers |
| Database load | High | Offloaded to a separate cluster |
| Implementation complexity | Basic setup 2–4 hours | 2–5 days + development |
How to Speed Up Indexing?
| Catalog size | Standard indexing | Incremental (agents) | Elasticsearch |
|---|---|---|---|
| 10,000 items | 15 min | 5 min | 2 min |
| 100,000 items | 2 h | 20 min | 5 min |
| 500,000 items | 10 h | 1 h | 15 min |
What Is Included in Turnkey Search Setup?
We offer a full cycle of work, backed by 50+ projects and a guarantee of results:
- Audit of the current state: check index, morphology, weights, error logs.
- Architecture design based on content volume and structure.
- Source configuration, garbage exclusion, performance optimization.
- Integration with Elasticsearch for catalogs over 200,000 items.
- Documentation of settings and administrator training.
- Post-implementation support (2 weeks free).
Timeline: basic setup — from 2 hours, fine-tuning with morphology — from 4 hours, Elasticsearch integration — from 2 days. Cost is calculated individually based on data volume and complexity. On average, server resource savings pay for the investment within 3 months.
Common Setup Mistakes
- Forgetting to exclude personal account and admin panel — they appear in search results.
- Not enabling morphology — search does not find word forms.
- Indexing all information block properties — index bloats, speed drops.
- Not using component caching — every request hits the database.
For an in-depth understanding, I recommend reading the article on full-text search on Wikipedia and the official documentation of the search module.
Where to Order Bitrix Search Setup?
Contact us for a free assessment of your project. Our specialists are certified in 1C-Bitrix, with 10+ years of experience. We guarantee that search will find the right products in seconds and reduce server load by 2-3 times. Request a consultation — we will evaluate your project within a day and offer the optimal solution.