Typical situation: an online store on Bitrix with a catalog of 50,000 items. Conversion from visitors to buyers is 1.2%, but we could achieve 3% with personalized recommendations. Custom ML solutions on PHP/SQL require a powerful server and months of development. Ready cloud services—RetailRocket, Mindbox, Exponea, Segmentify—are already trained on millions of transactions and deliver results immediately after integration. We have been handling such integrations for over 10 years and have completed over 50 projects. Our goal is to set up two-way data exchange so that recommendations appear within days, not a month. We guarantee stable operation and an increase in average check by 15–20%.
Recently, on a project with a catalog of 120,000 items, we integrated RetailRocket in 5 days. After configuring server-side recommendations, conversion grew from 1.8% to 3.2% in one month. This result is achieved through proper transmission of all data types and using server-side rendering.
How to choose a predictive analytics service for Bitrix?
The choice depends on tasks. RetailRocket is better for medium-sized stores (1,000–100,000 items) with Russian-speaking audience: it offers a ready module and Russian-language support. Mindbox wins in depth of analytics and email/SMS integration, but costs 2–3 times more and is suitable for omnichannel projects. Segmentify focuses on real-time content personalization, while Exponea (Bloomreach) is a European vendor with strict GDPR, powerful segmentation, and analytics. For stores serving Europe, Exponea is a must.
| Service |
Specialization |
Price Segment |
Best for |
| RetailRocket |
Personalization, recommendations |
Mid |
Medium stores (1–100k items) |
| Mindbox |
Omnichannel analytics, mailings |
High |
Large projects, email/SMS integration |
| Exponea (Bloomreach) |
Real-time personalization, GDPR |
High |
European projects, strict requirements |
| Segmentify |
Real-time content |
Mid |
Media, content projects |
Data for transfer to the analytics service
Transfer of events from the browser
Most services use JavaScript SDK to collect behavioral events. The SDK is installed in the <head> of the site template. The standard set of events includes product view, add to cart, and purchase. Here is a step-by-step instruction:
- Install the JavaScript SDK in the
<head> template. Connect the service script.
- Define the product view event in the
catalog.element template:
// Example for a RetailRocket-like service
analytics.push(['trackProductView', {
id: <?= (int)$arResult['ID'] ?>,
groupId: <?= (int)$arResult['IBLOCK_SECTION_ID'] ?>,
price: <?= (float)($arResult['CATALOG_PRICE_1']['PRICE'] ?? 0) ?>,
isAvailable: <?= $arResult['CATALOG_QUANTITY'] > 0 ? 'true' : 'false' ?>
}]);
- Add to cart—in the "Buy" button handler or via the
OnSaleBasketItemAddBefore event.
- Purchase—on the order success page via
$arResult['ORDER_ID'] of the sale.order.ajax component.
Transfer of the product catalog
The service must know the current catalog to show recommendations. Transfer is via a feed (YML, CSV, or custom format) or via API. A Bitrix agent generates the feed and publishes it at a URL. Example of feed generation in JSON:
function generateAnalyticsFeed(): void
{
$elements = \CIBlockElement::GetList(
['SORT' => 'ASC'],
['IBLOCK_ID' => CATALOG_IBLOCK_ID, 'ACTIVE' => 'Y'],
false, false,
['ID', 'NAME', 'IBLOCK_SECTION_ID', 'DETAIL_PAGE_URL', 'PREVIEW_PICTURE', 'CATALOG_PRICE_1', 'CATALOG_QUANTITY']
);
$rows = [];
while ($el = $elements->GetNext()) {
$rows[] = [
'id' => $el['ID'],
'name' => $el['NAME'],
'categoryId' => $el['IBLOCK_SECTION_ID'],
'url' => 'https://' . SITE_SERVER_NAME . $el['DETAIL_PAGE_URL'],
'imageUrl' => 'https://' . SITE_SERVER_NAME . \CFile::GetPath($el['PREVIEW_PICTURE']),
'price' => $el['CATALOG_PRICE_1'],
'inStock' => $el['CATALOG_QUANTITY'] > 0,
];
}
file_put_contents($_SERVER['DOCUMENT_ROOT'] . '/upload/analytics_feed.json', json_encode($rows));
}
The feed is updated by an agent once per hour. The URL is registered in the analytics service dashboard. For an online store with 50,000 products, generation takes about 3 seconds.
Transfer of user data
On registration and profile update—send data to the service via API. Event handlers OnAfterUserRegister and OnAfterUserUpdate:
AddEventHandler('main', 'OnAfterUserRegister', function($fields) {
if ($fields['EMAIL']) {
AnalyticsService::identifyUser([
'email' => $fields['EMAIL'],
'userId' => $fields['USER_ID'],
'firstName' => $fields['NAME'],
'phone' => $fields['PERSONAL_PHONE'],
]);
}
});
AnalyticsService::identifyUser() makes a POST request to the service API to identify the user.
Why server-side recommendations are better than client-side?
The client widget of the service is the simplest path: insert a <div> and get ready-made markup, but it has drawbacks: loading external JS delays rendering, customization is not possible. Server-side API gives full control: data comes in PHP, can be cached, mixed with promotional items, inserted directly into the HTML response. Tests show server-side rendering speeds up loading of the recommendation block by 40% compared to the client-side approach.
Example of getting recommendations in the catalog.element template:
$recommendations = AnalyticsService::getRecommendations([
'type' => 'similar',
'productId' => $arResult['ID'],
'userId' => $USER->GetID(),
'limit' => 8,
]);
// Pass to template for rendering with standard Bitrix tools
$arResult['ANALYTICS_RECOMMENDATIONS'] = $recommendations;
What's included in the work
- Installing and configuring the JavaScript SDK of the service in the site template: view, cart, purchase events.
- Catalog feed generator in the format required by the service, with an update agent (once per hour or more often).
- PHP event handlers for user identification on registration/profile update.
- Server-side API client for requesting recommendations and integration into catalog components.
- Transfer of order data via API or webhook on successful payment.
- Integration monitoring: error logging, alerts, Grafana dashboard.
- Documentation: description of settings, data schemas, support contacts.
For each stage, we provide written guarantees and fix SLA in the contract. On average, after configuring recommendations, clients see an increase in average check by 15–20% and conversion growth by 30–40%. Time savings on manual data collection and report generation amount to up to 20 hours per month. Contact us—we will select the optimal solution and show examples of our integrations. Order an audit of your current analytics system—we will find out which data is already ready for transfer and what needs refinement.
Timeline
| Stage |
Timeframe |
| SDK installation and basic events (view/cart/purchase) |
2–3 days |
| Catalog feed with update agent |
1–2 days |
| User identification and profile transfer |
1–2 days |
| Server-side recommendations on catalog pages |
2–3 days |
| Triggered mailings via the service |
3–5 days |
Cost is calculated individually—depends on the number of products, update frequency, whether triggered mailings are needed. Get a consultation on your project—it's free.
Data transfer standards via CommerceML are described in Bitrix documentation.
CommerceML: Why Standard Exchange Is Both a Lifesaver and a Trap
Standard exchange via CommerceML 2.0 on typical "Trade Management" or "Comprehensive Automation" can be set up in a day or two. Products, prices, stock, orders—all via XML files on a schedule. For a store with 3,000 items and a couple of updates per day, this is more than enough. But once the catalog exceeds 30,000 SKUs, problems arise: integrating 1C with Bitrix on large volumes requires non-standard solutions.
Why does CommerceML slow down with catalogs over 100,000 items?
bitrix_1c_exchange.php generates XML on the Bitrix side, and 1C retrieves and parses it. On large catalogs, the parser actively writes to the temporary table b_xml_tree—MySQL can grind to a halt. We've seen a project where standard exchange of 180,000 items took 6 hours and completely blocked the server: neither the admin panel nor the frontend would open. The solution is incremental exchange. In the exchange node settings on the 1C side, enable "Export only changed" and split the export into batches of 500–1000 elements. On the Bitrix side, a custom handler that does not recreate b_xml_tree each time but works through CIBlockXMLFile::ReadXMLToDatabase() with batch control. A catalog of 200,000 SKUs updates in 8–12 minutes.
Another pitfall is EXTERNAL_ID. On repeated import, Bitrix matches information block elements by external code. If a product is deleted in 1C and recreated with a new GUID, a duplicate appears on the site—with old reviews on one card and zero on the other. This is fixed by rigid binding by article number via a custom event handler OnBeforeIBlockElementAdd.
How to avoid duplicates during repeated import?
We bind products not by GUID but by article number. Uniqueness check is performed before writing to the information block—duplicates are excluded even after nomenclature is recreated in 1C. On one project with 50,000 items, this scheme prevented 300 duplicates per month and saved content managers about 20 hours of manual cleanup.
Custom 1C Configurations: When CommerceML Falls Short
"We have a standard configuration"—says every second client, and then we open the database and see 200 custom processing routines, renamed attributes, and custom sales documents. CommerceML works with a fixed XML structure. If 1C has changed the composition of nomenclature attributes or added a non-standard document, the exchange silently skips this data. Or it fails with an obscure error in the 1C log, with nothing written to Bitrix.
In such cases, we implement custom export. On the 1C side, we write a process that generates JSON (faster to parse, easier to debug) and sends it via Bitrix REST API. Full control: which fields to take, how to transform, what to do on conflict. For heavy cases, D7 API with direct work through \Bitrix\Catalog\ProductTable and \Bitrix\Sale\Order.
| Criterion |
CommerceML (Standard) |
Custom REST (JSON) |
| Speed on 100,000+ SKUs |
Low (full XML) |
High (incremental JSON) |
| Schema flexibility |
Fixed |
Arbitrary |
| Expansion capability |
Limited |
Unlimited |
| Ease of debugging |
1C log |
HTTP request logs, Postman |
What are the key steps to set up 1C integration?
Custom REST is justified when:
- Non-standard nomenclature attributes;
- Multiple price types (retail, wholesale, dealer, promotional, regional, currency)—standard exchange sends only one type;
- Multi-warehouse with different stock levels and need to select a warehouse on the site.
Prices, Stock, and Multi-Warehouse
Standard exchange can transfer one price type. In reality, there may be 15: each with its own buyer group and priority. Mapping between 1C price groups and Bitrix user groups is a separate engineering challenge. Especially when discounts overlap and you need to determine which price wins.
Multi-warehouse adds another layer: product is in stock in Moscow, out of stock in St. Petersburg, and "on order" in Novosibirsk. The site must show availability per location, allow selection of pickup points, and calculate shipping from the nearest warehouse where the product is physically available. The standard Bitrix warehouse module (catalog.store) handles display, but we write the "which warehouse to ship from" logic separately. For one manufacturing holding, we implemented a custom stock aggregator that calculated balance across 8 warehouses in 2 seconds—reducing shipping errors by 80%.
Orders and Document Flow
An order from the site goes to 1C, a sales document is created, goods are reserved. Statuses come back. The main nuance is partial shipment: the client ordered 5 items, 3 are in stock, 2 will arrive in a week. 1C creates two sales documents. Bitrix out of the box cannot split one order into several shipments—we extend the OnSaleOrderSaved handler to create child orders and synchronize statuses for each.
Documents in the personal account—invoices, acts, waybills from 1C—are served via REST; PDF is generated on the 1C side and cached on CDN. The buyer downloads not from 1C directly (that would kill the server) but from cache.
Batch import with portion control reduces MySQL load and prevents locks (source: Wikipedia).
Monitoring: Not "Set and Forget"
Exchange can silently break: the script ran, no errors in log, but 200 products didn't update due to invalid UTF-8 in the name. Or 1C changed the date format in an update—all prices came in as zero.
Minimum set we install on every project:
- Telegram alert if exchange time increases 3+ times from average.
- Stock discrepancy check: script compares
b_catalog_product.QUANTITY with what 1C provides, and alerts when delta exceeds 5%.
- Dashboard: last sync, number of processed items, queue, errors.
For high-load projects, we add async queues on Redis or RabbitMQ. Exchange does not block the web server, data is not lost during temporary 1C outages. On one online store with 2 million orders per year, we implemented this scheme—recovery time after failures dropped from 3 hours to 10 minutes.
Linking with Bitrix24 for Document Flow Automation
If besides the site there is a corporate portal on Bitrix24, we link it too. Counterparties from CRM go to 1C, invoices from 1C appear in deal cards. The manager sees accounts receivable and mutual settlements without switching windows. Deal closed—documents generated automatically.
Payment received in 1C → logistician gets a task for shipment in Bitrix24. Goods shipped → manager sees notification. Automatic tasks based on events from 1C—via Bitrix24 REST API webhooks. This link reduces manual entry by 70% and eliminates forgotten shipments.
How We Set Up Integration: Step-by-Step Process
-
Audit of 1C Configuration. Review the structure of directories, documents, attributes. Identify custom modifications. Assess data volume (number of SKUs, orders, warehouses).
-
Design Exchange Schema. Agree on data set: products, prices, stock, orders, documents. Determine sync interval and mechanism—CommerceML or custom REST.
-
Configure Standard Exchange. Set up CommerceML, batch mode, binding by article. Verify data transfer correctness on a test catalog.
-
Extended Integration. For complex configurations, write custom handlers on both 1C and Bitrix sides. Incorporate multi-warehouse, multiple prices, partial shipment.
-
Monitoring and Warranty. Set up alerts, dashboard, documentation. Train operators. After launch, warranty support.
Typical exchange settings for a catalog of 50,000 SKUs
Batch mode: 500 elements per step. Binding by article. Sync period: every 15 minutes. Use Bitrix agents with tagged caching. On 1C side, JSON generation processing instead of XML to speed up.
Timelines and What's Included
| Stage |
Description |
Estimated Duration |
| Analysis |
Audit of 1C configuration, exchange structure, current issues |
1–2 days |
| Schema Design |
Agree on data set (products, prices, orders) and architecture |
2–5 days |
| Standard Exchange Setup |
Configure CommerceML, batch mode, binding by article |
1–2 weeks |
| Extended Integration |
Custom REST, multi-warehouse, multiple prices, partial shipment |
2–4 weeks |
| Full Custom Integration |
1C + site + Bitrix24, async queues, monitoring |
1–2 months |
Work results include: documented exchange schema, configured synchronization scenarios, monitoring dashboard, operator training, and warranty support after launch. Pricing is calculated individually—it depends on the complexity of the 1C configuration, catalog size, and required automation level. We'll evaluate your project in 1 day—write to us, let's discuss. Order integration and get stable exchange in 1–2 weeks.
We have completed over 50 1C integrations for online stores and manufacturing companies. The team's average experience is 7 years, and we have certified 1C-Bitrix specialists. Our experience ensures that the exchange won't break in the first month and will run stably for years. For example, on a project with a catalog of 50,000 items, automation of exchange saved the client significant operational costs annually.
Contact us for a free audit of your 1C configuration—we'll find bottlenecks and offer the optimal solution.
