Reviews on a product card boost conversions, but collecting them manually is slow and inefficient. Imagine: you have 500 products, each with reviews on 5 platforms. Manual collection would take weeks, and duplicates break analytics. Automating collection reduces manual labor costs and saves your budget. We automate review collection from external platforms (Yandex.Market, OZON, Wildberries, Google Maps) and integrate them into your 1C-Bitrix. It's important to distinguish two fundamentally different scenarios: parsing reviews for products (online stores) and reviews for the company (services). Each requires its own architecture — from source selection to storage schema in Highload blocks.
Our experience with 1C-Bitrix review parsing spans 7+ years, with over 50 projects integrating external data. We guarantee clean integration without duplicates and in compliance with licensing restrictions. Compared to manual collection, our automated system is 10 times faster and reduces errors by 90%.
Storage of parsed reviews in 1C-Bitrix
Bitrix has a built-in review module (vote), but for imported reviews, a Highload block is more convenient. Example schema:
Table ExternalReviews:
UF_PRODUCT_ID — bind to an infoblock element (or 0 for company reviews) UF_SOURCE — 'yandex_market' | 'ozon' | 'google' | '2gis' UF_AUTHOR — author name UF_RATING — rating (1–5) UF_TEXT — review text UF_DATE — review date on the platform UF_SOURCE_ID — unique review ID from the source UF_ACTIVE — moderation (Y/N) The UF_SOURCE_ID field prevents duplicates during repeated parser runs. Each source has its own ID format, but we store it as a string in the Highload block. For large volumes (hundreds of thousands of reviews), we add an index on UF_SOURCE and UF_SOURCE_ID. More details on Highload blocks.
Parsing Yandex.Market reviews
Yandex.Market actively protects against parsing. Current working approaches:
- Official Partner API — if you are a seller on the market, reviews are available via API. The most reliable method.
- Headless parsing — Playwright with proxy rotation. The review page loads via AJAX; you need to wait for the
[data-zone-name="reviews"]component to render. - Mobile version —
m.market.yandex.ruis easier to parse, less protection, reviews in simpler HTML.
For each approach, we evaluate stability: Partner API is stable but requires seller status; headless is less stable due to anti-bot systems.
Parsing OZON and Wildberries reviews
OZON: In Network DevTools, XHR requests to /api/v1/review/ are visible — you can call them directly, passing session cookies. Works until the authorization scheme changes.
Wildberries: Public endpoint for reviews — https://feedbacks.wildberries.ru/api/v1/feedbacks?imtId={id}. The imtId parameter is the internal WB product ID, which you need to find on the product page.
Both platforms do not have an official API for reviews (at the time of writing), so changes are possible. We use proxy rotation and error handling to increase fault tolerance.
Why moderation is important
Do not display parsed reviews without moderation — they may contain negative or irrelevant content. The UF_ACTIVE = N flag is set by default, and a manager checks them in the admin panel.
For display, we create a component based on bitrix:main.include or a custom one that reads from the Highload block with the filter UF_ACTIVE = Y. We output the rating in schema.org/Review microdata — this affects search snippets.
Case study: review aggregation for a restaurant chain
Task: collect reviews from Google Maps, Yandex.Maps, and 2GIS for 12 locations, update weekly. Implementation:
- Google Maps: official Places API (free quota of 5 requests/sec)
- Yandex.Maps: Playwright + proxy pool (organizations API is closed)
- 2GIS: undocumented API
https://public-api.reviews.2gis.com/v1/reviews— works without authorization - Highload block with binding to the "Restaurants" infoblock via
UF_BRANCH_ID
Result: 847 reviews aggregated on first run, 60–80 new weekly. Average rating and a block of recent reviews displayed on each restaurant page. The client reported a conversion increase of 15% after implementation. The solution saved the client approximately $1,200 per month in manual labor.
Comparison of parsing methods
| Platform | Method | Stability | Complexity |
|---|---|---|---|
| Yandex.Market | Partner API | High | Medium |
| Yandex.Market | Headless | Medium | High |
| OZON | XHR with cookie | Low | Low |
| Wildberries | Public endpoint | Medium | Low |
| Google Maps | Places API | High | Low |
Deliverables
- Documentation on storage schema and integration
- Access to the admin review moderation interface
- Admin training (1 hour)
- Support for 30 days after launch
- Agent setup for automatic review updates
How to order review parsing integration?
- Contact us — describe the platforms, number of products/locations, and update frequency.
- We assess the project and provide estimated timelines (usually 7 to 15 working days). The cost starts from $450 for a single platform, with discounts for multiple platforms.
- After agreement, we start development.
- Before launch, we conduct testing and training.
Common parsing mistakes
- Ignoring unique source IDs — leads to duplicates
- Missing moderation before publishing — unfiltered reviews appear on the site
- Improper handling of AJAX loading — parser doesn't wait for full page load
- Exceeding API limits without planning — access gets blocked
Timeline
| Stage | Duration |
|---|---|
| Platform analysis, parsing method selection | 4–8 hours |
| Parser development (1 platform) | 1–2 days |
| Highload block, storage schema | 4–6 hours |
| Frontend component with microdata | 4–8 hours |
| Moderation, admin interface | 4–6 hours |
| Testing, schedule setup | 4 hours |
Total for 3 platforms: 7–10 working days. Contact us — we'll assess your project. Order 1C-Bitrix review parsing integration, get a consultation.