Catalog search with query analytics

What this solution does

Catalog search module for 1C-Bitrix with query analytics. The system locates products while accounting for typos, various Russian word forms, transliteration, and keyboard layout errors. Store owners receive data on the most frequent queries and searches that returned no results.

Search operates across products and sections. The module builds a custom index for selected information blocks. Searchable fields include product names, descriptions, tags, and properties, as well as section names and trade offer properties. Trade offers are displayed via the parent product card and are not shown as separate results.

Supported features include:

- Russian morphology: words with different endings, such as "drilei" and "dril", are normalized to a common base;

- Automatic typo correction based on the dictionary of the built index;

- Handling of transliteration and incorrect keyboard layouts: for example, "kartridj" converts to "kartridzh", and "gjbcr" converts to "poisk";

- Configurable synonym groups when morphology is enabled;

- Search by SKU without separators and partial word matching if exact matches are not found;

- Search by product ID, including short numbers and numbers with leading zeros. Section search by ID is not supported.

Names of lists, bindings, and properties of high-load directories can be indexed. The current version of the module processes up to 10,000 records per directory; values outside this range use the internal XML_ID. Transliteration does not guarantee recognition of all word forms or arbitrary spelling variations.

When the Trade Catalog module is installed, search results consider product availability, purchase possibility, public price, photo presence, and activity period. Properties and available prices of trade offers are reflected in the parent product card.

For the general index, the minimum price per unit is selected from price types allowed for the 'All Users' group simultaneously for viewing and purchasing. Personal prices, coupons, and cart discounts are not calculated. Rows that cannot be correctly converted to the base currency must not be used as a public price.

It is possible to hide products without a price, those unavailable for purchase, and items from inactive sections. Configuration allows adjusting relevance boosts for availability, photos, and search transitions, as well as defining result sorting.

When search results are empty, the module suggests a corrected query or popular items based on the frequency of clicks from search. These recommendations are not based on purchase statistics.

Analytics

The module collects daily aggregates of valid search queries when statistics are enabled: number of searches, number of zero-result pages, and registered clicks from the results page. Reports display frequent queries, the share of zero-result pages, and the ratio of clicks to searches.

From the report, you can navigate to create a synonym or a display rule. Exports are available in CSV and Excel (SpreadsheetML) formats with the .xls extension. Each export file contains up to 5,000 aggregated rows for the selected filter.

Queries that do not meet search conditions, such as non-numeric queries that are too short, are excluded from statistics. Clicks from dropdown product suggestions in the current version are not counted as clicks from the results page.

Manual Control and SEO

For specific queries, you can pin or hide items on the results page, boost the ranking of items in a category and its subcategories, configure redirects to landing pages, and add custom text suggestions.

Rules for pinning, hiding, and promoting sections apply only to the main search results. Product suggestions are generated separately: a product hidden from the main list may still appear in autocomplete.

You can configure the title, meta description, and SEO text for search queries using the #QUERY# and #COUNT# placeholders. The system prevents indexing of empty result pages and pagination pages.

SEARCH FORM

The form supports dropdown suggestions including browser search history, site-wide popular queries, sections, and products. For products, images, prices, and availability status can be displayed. Voice input is available in browsers that support speech recognition.

User search history is stored locally in the browser. On the server, only query texts and daily aggregated data are saved, excluding IP addresses and session identifiers from the statistics table. The click rate limiter uses a separate cache with a value derived from the IP address.

INDEX CONNECTION AND UPDATE

The package includes the ozhogov:searchstat.form and ozhogov:searchstat.page components, along with templates for the standard bitrix:search.title and bitrix:search.page components.

The search module with query analytics uses the ozhogov.searchstat template when working with standard components. The search scope is configured in the module settings; existing filters and categories of standard components are not migrated automatically. Custom components from ready-made solutions may require adaptation; universal compatibility with Aspro, Intek, and Sotbit solutions is not guaranteed.

After selecting information blocks, an index must be built. Event handlers place changes to elements, sections, prices, and stock levels into a queue. The agent runs every five minutes and processes the queue in batches; completion time depends on the queue size and site agent activity.

Direct property assignment via SetPropertyValues or SetPropertyValuesEx without a standard element update is not covered by default subscriptions. Such changes require manual reindexing. Visibility changes based on activity dates may be reflected with a delay until the output cache expires.

The results page must be excluded from the composite cache. The module executes its own SQL queries directly against the site database; Sphinx and Elasticsearch are not required.

REQUIREMENTS AND TESTED ENVIRONMENT

- PHP 8.2 or newer.

- 1C-Bitrix main module 20.0.0 or newer.

The 'Catalog Search with Query Analytics' module works with Information Blocks. The 'Trading Catalog' is used for prices and availability, while 'Highload Blocks' handle property directories.

The code includes schemas for MySQL and PostgreSQL. Integration scenarios have been tested exclusively on MySQL. Full compatibility verification with PostgreSQL, MariaDB, and all minimum platform versions has not been performed.

Known limitations of the switches in the current version:

- When SHOW_SUGGEST is set to N, the clear button may still open the list of popular queries.

- Disabling price display in suggestions also disables photo loading.

- Setting USE_SEO to N disables meta tag generation but does not affect SEO text or prevent the use of SEO titles when SET_TITLE is set to Y.

For the “Catalog search with query analytics” module, General iT can perform installation, configuration and a functional check in your project.