How to Collect Reviews from Wildberries and Ozon Without Bans: A Complete Guide for Sellers

Analyzing competitor reviews is a critically important task for any marketplace seller. Reviews reveal real product issues, help improve listings, and identify untapped niches. However, marketplaces strictly block mass parsing: Wildberries bans IPs after 50-100 requests, Ozon shows a captcha, and Yandex.Market has completely closed access to the API for individuals. In this article, we will discuss how to collect reviews safely, which tools to use, and how to set up proxies to maintain access to data.

Why Marketplaces Ban Review Parsing

Marketplaces protect their data for several reasons. Firstly, mass parsing creates a high load on servers — thousands of requests per minute from parsers can slow down the site for regular customers. Secondly, reviews are valuable data that platforms monetize through paid analytics (for example, Ozon Analytics or WB Analytics cost from 5000 rubles per month). Thirdly, competitors may use reviews for unethical actions: copying successful strategies, inflating reviews, or even finding dissatisfied customers of competitors.

Technically, marketplaces track suspicious activity based on several parameters:

• Request Frequency from One IP: If your IP sends 100+ requests per minute — it is clearly not a human. Wildberries bans such IPs for 24 hours after 50-70 requests.
• Behavior Patterns: Parsers open pages too quickly (less than 1 second per page), do not move the mouse, and do not scroll. Modern anti-bot systems (such as Kasada or DataDome) easily recognize this.
• User-Agent and Browser Fingerprint: If your script sends requests with an old version of Chrome User-Agent or without JavaScript — this is a red flag.
• Absence of Cookies and Sessions: Regular users have browsing history, authentication cookies. Parsers often come "clean".

Ozon uses particularly aggressive protection: after 20-30 requests, it shows a captcha, and upon repeated attempts to bypass it, blocks the IP at the CDN level (Cloudflare). Yandex.Market closed public API access for individuals in 2023 — now access is only for legal entities with a contract. Avito bans not only by IP but also by phone number if it detects automated activity.

What Happens If You Parse Without Protection: Consequences of Bans

The consequences of blocking depend on how exactly you parse. If you use a home IP without a proxy — you will receive a temporary ban for 24 hours. This is not critical for one-time data collection, but it is a problem for regular monitoring. If you parse through a seller account (for example, using a personal WB account for access to extended statistics) — you risk getting your account banned. Restoring it is difficult: you need to write to support, proving that it was not you, but "a virus on your computer." The process takes from a week to a month.

A real case: a seller of children's toys on Wildberries set up automated parsing of competitor reviews through a Python script. The script worked from his work IP, collecting 500 reviews every hour. After 3 days, Wildberries blocked not only the IP but also the seller's account — the algorithm linked the parser's IP with the IP from which he accessed his personal account. Result: loss of access to the store for 2 weeks, lost revenue of about 400,000 rubles (the products were in stock, but managing supplies was impossible).

Another issue is legal. Formally, parsing violates the user agreement of most marketplaces (section "Prohibited Actions"). In practice, platforms rarely go to court with parsers, but there are precedents. In 2022, Ozon filed a lawsuit against an analytics service that collected data on prices and stock levels. The court sided with Ozon, and the service shut down. For individuals, the risk of legal claims is minimal, but for companies that sell parsing as a service, this is a real threat.

3 Ways to Collect Reviews: Manual, API, Automated Parsing

Method 1: Manual Collection (for Small Volumes)

If you need to collect reviews for 5-10 competitor products once a week — you can do it manually. Open the product card on Wildberries or Ozon, scroll through the reviews, and copy the text into Excel. Pros: no technical skills required, zero risk of a ban. Cons: very slow (it takes 30-40 minutes for 100 reviews), impossible to track dynamics, no automation.

This method is suitable for micro-businesses: you sell 2-3 products, want to understand what is being said about competitors to improve your listing. For regular monitoring of dozens of products, manual collection does not scale.

Method 2: Official API (Limited Access)

Some marketplaces provide APIs for partners. Wildberries API for sellers allows you to obtain reviews only for your own products (not competitors). Ozon API similarly — access is only to your own data. Yandex.Market API for content partners requires a legal contract and is only available to companies with a turnover of at least 10 million rubles per year.

Pros of the API: stable access, you do not violate platform rules, structured data in JSON. Cons: you cannot obtain competitor reviews (the main task for most sellers), high entry threshold for the official API.

Method 3: Automated Parsing via Proxies (Universal)

This is the main method for professional sellers and agencies. You use special tools (parsers) or write a script that automatically opens product pages, collects reviews, and saves them to a database. To avoid getting banned, all requests go through proxies — each request from a new IP address, the marketplace sees regular users from different cities.

Pros: you can collect reviews for any products (yours and competitors'), full automation, scalability (up to 10,000 products per day). Cons: you need technical skills or paid tools, costs for proxies (from 3000 rubles per month).

Which Proxies Are Suitable for Marketplace Parsing

The choice of proxy type depends on your budget, parsing volumes, and data quality requirements. Let's discuss three main types and their applicability for marketplaces.

For parsing Wildberries and Ozon, we recommend residential proxies. These marketplaces use advanced anti-bot systems that easily identify datacenter IPs (they are located in known ranges of AS numbers of hosting providers). Residential proxies are IPs of regular home internet providers (Rostelecom, MTS, Beeline), and they cannot be distinguished from real buyers.

An important point: for marketplaces, proxies with IP rotation are needed. This means that each of your requests goes from a new IP address. For example, if you parse 1000 reviews — that’s 1000 different IPs from different cities in Russia. The marketplace sees that real users from across the country are accessing the product, with no suspicious activity.

Ready-Made Tools for Collecting Reviews: Service Overview

If you are not a programmer, you do not need to write a parsing script from scratch — there are ready-made solutions. Let's review popular tools for marketplace sellers.

1. Mpstats (mpstats.io) — Analytics for Wildberries and Ozon

The most popular service among Russian sellers. It collects not only reviews but also prices, stock levels, search positions, and sales history. Reviews are updated once a day, and you can export them to Excel. Price: from 3000 rubles per month (the "Start" tariff). Pros: no need to set up proxies yourself (everything works out of the box), user-friendly interface, ready-made reports. Cons: expensive when scaling (if you track 100+ products, the tariff increases to 15,000 rubles), data updates with a delay.

2. Moneyplace (moneyplace.io) — Parser for Ozon

A specialized tool for Ozon. It collects reviews, customer questions, and seller ratings. There is a sentiment analysis feature (automatically identifies negative and positive reviews). Price: from 2500 rubles per month. Pros: deep analytics specifically for Ozon, API for integration with your systems. Cons: works only with Ozon, no Wildberries or other platforms.

3. Parser.ru — Universal Parser Without Code

A cloud service for parsing any websites, including marketplaces. It works on the principle of "specify what to parse — get the result." No programming is needed, everything is set up through a visual interface. Price: from 1500 rubles per month + separate payment for proxies (if you use your own — free). Pros: flexibility (you can set up parsing of any data, not just reviews), works with any marketplaces. Cons: takes time to set up (1-2 hours to learn the interface).

4. Octoparse — Desktop Parser for Windows/Mac

A program for parsing with a visual editor. Install it on your computer, open the marketplace website, click on the elements you need to collect (review text, rating, date). The program remembers the structure and parses automatically. Price: free version (up to 10,000 records per month), paid from $75/month. Pros: no need to know HTML/CSS, there are ready-made templates for popular sites. Cons: works only on your computer (if you turn it off — parsing stops), English interface.

5. Custom Python Script (for the Technically Savvy)

If you have basic programming skills, you can write a script in Python using libraries like Selenium (browser emulation) or Scrapy (fast parsing). Pros: full control, zero subscription fees (you only pay for proxies), you can parse any data in any format. Cons: takes time to develop (from 5 hours for a simple parser), you need to maintain the code (marketplaces change the site structure — the script breaks).

For most sellers, we recommend starting with ready-made services like Mpstats or Moneyplace. They pay off if you earn from 200,000 rubles per month on marketplaces. If the budget is limited or you need non-standard parsing logic — consider Octoparse or a custom script.

Step-by-Step Guide to Parsing via Proxies

Let's discuss the setup using the combination of Octoparse (no-code parser) + residential proxies. This method is suitable for beginners and does not require programming.

Step 1: Obtain Access to Residential Proxies

Register with a residential proxy provider (for example, ProxyCove). After registration, you will receive connection details:

Proxy Server: gate.proxycove.com
Port: 8080
Username: user_your_id
Password: your_password
Type: HTTP/HTTPS (with IP rotation)

Important: choose a plan with request rotation (each HTTP request = new IP). This is critical for parsing marketplaces.

Step 2: Install and Configure Octoparse

Download Octoparse from the official website (octoparse.com), install it on your computer. Upon the first launch, the program will ask you to create an account — use the free version for testing (limit of 10,000 records per month, enough for 50-100 products).

Open settings (Settings → Proxy Settings) and add your proxies:

• Proxy Type: HTTP
• Server: gate.proxycove.com
• Port: 8080
• Username: your username
• Password: your password
• Check the box "Use proxy for all tasks"

Click "Test Proxy" — the program will check the connection. If everything is set up correctly, you will see the message "Proxy is working."

Step 3: Create a Review Parsing Task for Wildberries

Click "New Task" → "Advanced Mode". Paste the product card URL on Wildberries (for example, https://www.wildberries.ru/catalog/12345678/detail.aspx). Octoparse will open a built-in browser and load the page via the proxy.

Now teach the program what exactly to parse:

1. Scroll down to the reviews section.
2. Click on the text of the first review → select "Extract text" → the program will highlight all reviews on the page.
3. Click on the rating (stars) → "Extract text" → the program will remember the rating.
4. Click on the review date → "Extract text".
5. Find the "Show more" button (pagination) → right-click → "Click to paginate" → the program will automatically scroll through the pages.

Set limits: in the "Pagination" section, specify a maximum of 50 pages (about 500 reviews). This is a safe volume for one session.

Step 4: Set Delays Between Requests

This is critically important! If the parser scrolls through pages too quickly, Wildberries will suspect a bot even through a proxy. Open "Task Settings" → "Speed" and set:

• Page loading delay: 3-5 seconds (delay between pages)
• Action delay: 1-2 seconds (delay between clicks)
• Enable random delay: enable (adds randomness, simulates a human)

Step 5: Start Parsing and Export Data

Click "Run" → the program will start parsing. You will see data being collected in real-time. After completion (usually 10-15 minutes for 500 reviews), click "Export" → choose Excel or CSV format. Done — you have a table with all reviews, ratings, and dates.

Safe Parsing Rules: Limits, Delays, Rotation

Even with proxies, you can get banned if you violate "rules of decency." Marketplaces analyze not only IPs but also behavioral patterns. Here’s a checklist for safe parsing:

1. Observe Request Limits

Safe limits for Russian marketplaces:

• Wildberries: no more than 100 requests per hour from one parsing task. If you parse 10 products — take a 30-40 second break between products.
• Ozon: no more than 50 requests per hour (they have more aggressive protection). Pause between products — 1 minute.
• Yandex.Market: no more than 30 requests per hour. After every 10 requests, take a 5-minute break.

These limits are calculated for parsing through residential proxies with rotation. If you use datacenters — divide the limits by 2.

2. Use Random Delays

Do not make delays fixed (for example, exactly 5 seconds between requests). This looks suspicious. Set random delays: from 3 to 7 seconds. Most parsers (Octoparse, Scrapy) support this function out of the box.

3. Change User-Agent

User-Agent is a string that tells the site which browser you are using. Parsers often send outdated User-Agents (for example, Chrome 90, while the current version is 120). Set up User-Agent rotation: each request from different browsers (Chrome, Firefox, Safari) and different versions.

Example list of User-Agents for rotation:

Mozilla/5.0 (Windows NT 10.0; Win64; x64) Chrome/120.0.0.0
Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) Safari/605.1.15
Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:121.0) Firefox/121.0
Mozilla/5.0 (X11; Linux x86_64) Chrome/119.0.0.0

4. Do Not Parse During Peak Hours

Marketplaces strengthen anti-bot protection during peak traffic hours (usually from 6 PM to 11 PM Moscow time — when buyers return from work). During this time, the risk of getting a captcha or ban is higher. The optimal time for parsing: from 2 AM to 8 AM or from 10 AM to 4 PM.

5. Rotate Proxies by Sessions

If you are parsing one product (for example, scrolling through 50 pages of reviews), use one IP for the entire session. If you change IP on each page — the marketplace may suspect something strange (one "user" jumps from an IP in Moscow to Vladivostok in a second). But when you move to the next product — change the IP.

6. Simulate Human Behavior

Advanced parsers (for example, Selenium) can emulate human actions: mouse movement, scrolling, clicking on random elements. This increases the "humanity" of the parser. If you use a simple HTTP parser (without a browser) — this is not necessary, but then strictly adhere to request limits.

How to Analyze Collected Reviews for Sales Growth

Collecting reviews is only half the battle. The main thing is to analyze them correctly and apply insights. Here are practical use cases.

1. Finding Competitors' Product Flaws

Open Excel with collected reviews, filter out negative ones (rating 1-2 stars). Look at what customers complain about: size does not match, poor packaging, long delivery, low material quality. These complaints are your competitive advantage. If competitors receive 30% negative reviews due to incorrect sizing — specify a detailed size chart with a measuring tape photo in your listing. If they complain about packaging — include a nice box with the product and mention it in the description.

A real case: a seller of sports leggings collected 500 reviews on the top 5 competitors. He found that 40% of negative reviews were related to the fabric being see-through during squats. He ordered a denser fabric from the supplier, filmed a video test "leggings do not show through," and added it to the listing. Result: conversion increased from 3% to 7%, sales doubled in a month.

2. Improving Product Listings Based on Customer Questions

Besides reviews, parse customer questions (they are available on Wildberries and Ozon). Questions show what information is lacking in the listing. If 20 people ask "Is this 100% cotton or synthetic?" — it means the composition is unclear. Add a prominent block in the description "Composition: 95% cotton, 5% elastane" with an icon. If they ask about compatibility with iPhone 15 — add a list of compatible models.

3. Finding Unmet Needs (New Niches)

Analyze positive reviews of competitors. Look for phrases like "great, but I wish..." or "too bad there is no...". These are signals of unmet needs. For example, in reviews of children's backpacks, people often write "good backpack, but no reflective elements." You can launch a backpack with reflective elements and capture this audience.

4. Monitoring Review Dynamics (Early Problem Detection)

Set up automatic parsing of your products once a day. If the number of negative reviews sharply increases (for example, 10 reviews with complaints about defects came in a week) — this is a signal to check the batch with the supplier. Perhaps he sent defective products. The sooner you detect the problem, the less you will lose in ranking.

5. Collecting Semantics for SEO within the Marketplace

Customer reviews are a source of "live" keywords. People write how they search for a product: "bought for the dacha," "looking for a gift for my husband on February 23," "needed a thermos for the car." These phrases can be added to the title and description of your listing — this will increase visibility in the marketplace search.

Conclusion

Collecting reviews from marketplaces is not just a technical task, but a strategic tool for sales growth. Properly collected and analyzed competitor reviews give you a competitive advantage: you know what is lacking in the market, which problems to solve, and how to improve your listing. The main thing is to do this safely to avoid getting banned.

Key takeaways from the article: use residential proxies for parsing Wildberries and Ozon (they do not get banned, unlike datacenters), observe request limits (no more than 100 requests per hour on Wildberries), set random delays between requests (3-7 seconds), do not parse from the same IP from which you access your seller account. For beginners, we recommend starting with ready-made tools like Mpstats or Octoparse — they pay off if your turnover is from 200,000 rubles per month.

If you plan to regularly collect data from marketplaces, we recommend trying residential proxies — they provide a high level of anonymity, real IPs from Russian providers, and minimal risk of blocks. This is an investment that pays off by maintaining access to critically important data for your business.