Lead generation agencies and marketers who collect company databases from maps inevitably face the same question: how much does it actually "weigh" to gather 10,000 contacts and why accounts and IP addresses start getting banned already at the third thousand. We break down the traffic calculation step by step, compare Google Maps, 2GIS, and Yandex.Maps, and show which proxies can handle such volumes without getting blocked.
What is Lead Generation from Maps and Why Calculate Traffic
Lead generation from maps is the collection of contact data of companies (name, phone, website, address, rating, category of activity) through Google Maps, 2GIS, or Yandex.Maps for subsequent cold calls, email campaigns, or targeted advertising. This method is particularly popular among agencies that sell B2B services: web studios, CRM providers, marketing contractors — they need databases by niches and cities: "dentists in Kazan," "auto services in Novosibirsk," "beauty salons in Yekaterinburg."
Traffic calculation is important for three reasons. First, the volume of data transmitted determines how much proxy traffic you will consume and how much it will cost when scaling to several cities. Second, maps record abnormal activity based on the number of requests, not just by IP — and if you don't understand how many requests are generated by collecting 10,000 cards, it's easy to underestimate the risk of a ban. Third, knowing the exact volume of data makes it easier to choose between parsing the HTML page, using the internal map API, or a ready-made service — each option differs significantly in traffic.
In practice, 10,000 companies is not an abstract number. This volume corresponds, for example, to the complete database of active legal entities in the "equipment repair" niche across 15-20 major cities in Russia, or all food establishments in a large city like Moscow or St. Petersburg. This is a realistic task for an agency that sells leads in packages.
How Much Does One Company Card Weigh
The weight of one company card depends on what exactly you are collecting and how. If you parse only the list of search results (name, short address, rating), the data is transmitted compactly — usually, this is a JSON response from the internal API of the map weighing 3-8 KB per company in the overall list. However, if you need a detailed card — phone, website, working hours, photos, reviews — you will have to open a separate page or make an additional request, and the weight increases to 80-250 KB due to images and extended JSON.
Below is a comparison of the three main maps that lead generation agencies work with in the CIS:
| Map | Weight of Card in List | Weight of Detailed Card | Protection Feature |
|---|---|---|---|
| Google Maps | 5-8 KB | 150-250 KB | Strict rate-limit and captchas for frequent requests from one IP |
| 2GIS | 3-6 KB | 80-150 KB | Blocking based on request frequency to the catalog API |
| Yandex.Maps | 4-7 KB | 100-200 KB | Anti-bot system with behavioral analysis and proxy blacklists |
An important nuance: the weight of the list is not the entire card. To obtain the phone and website of the company, in most cases, a separate request to the detailed page is needed. This doubles or even triples the actual traffic per lead if you are collecting not just names, but full contacts for sales.
Traffic Calculation for 10,000 Companies Step by Step
Let's take a realistic scenario: an agency collects full cards (name, phone, website, address, rating, category) from Google Maps for 10,000 companies in several cities.
Step 1. Search and Listing. Google Maps returns about 20 results on one search results page. To obtain 10,000 entries, approximately 500 search results pages need to be processed. Each page, considering service requests, map loading, and scripts, weighs 600-900 KB. Total: 500 × 750 KB ≈ 375 MB just for the search stage.
Step 2. Detailed Cards. For each of the 10,000 companies, a separate request is needed for the phone, website, and photos. With an average weight of 180 KB per card: 10,000 × 180 KB ≈ 1.8 GB.
Step 3. Repeat Requests and Retries. In real-world parsing, 10-20% of requests do not go through on the first attempt due to rate-limits, captchas, or changes in layout — they have to be repeated. Add another 15% to the volume: (375 MB + 1.8 GB) × 1.15 ≈ 2.5 GB.
The final figure is about 2.5 GB of traffic to collect 10,000 companies from Google Maps when parsing detailed cards. If you only collect lists without phones and websites, the traffic drops to 450-500 MB, but such a database will be poorly suited for sales — without a phone, a lead does not convert into a call.
For 2GIS and Yandex.Maps, the calculation is similar, but the volume is usually 20-30% lower due to a more compact API format — expect 1.7-2.2 GB for the same 10,000 cards.
Why Maps Ban for Mass Data Collection
Maps are a commercial asset of the platforms themselves, and mass automated data collection directly contradicts their business model (selling ads in cards, API access for partners). Therefore, Google Maps, 2GIS, and Yandex.Maps react aggressively to atypical behavior.
The main triggers for blocking: too high a frequency of requests from one IP (more than 1 request every 2-3 seconds), identical User-Agent and headers without variability, absence of cookies and session data, requests without accompanying resources (images, scripts) that a regular browser always loads. When the limit is reached, the service either shows a captcha, temporarily blocks the IP for 15-60 minutes, or permanently bans the IP for repeated violations.
For the calculated volume of 2.5 GB and 10,000+ requests from one "clean" IP, the probability of a ban approaches 100% already at the first thousand cards. That is why any serious lead generation project from maps is built around IP address rotation — without this, the traffic calculation loses its meaning, as it will be physically impossible to collect the planned volume.
Residential, Mobile, and Datacenter Proxies for Map Parsing
The choice of proxy type directly affects how many requests you can make before getting blocked and how much traffic will need to be "burned" on retries. For data collection from maps, the reputation of the IP is critical — maps block datacenter ranges much more harshly because the majority of automated parsing comes from there.
| Proxy Type | Trust Level of Maps | When to Use |
|---|---|---|
| Residential Proxies | High — IPs of real users | Main tool for collecting 5,000-50,000 cards with rotation by cities |
| Mobile Proxies | Very high — common IPs of operators | Targeted collection by niches with strict anti-bot protection (Yandex.Maps) |
| Datacenter Proxies | Low-medium — often in blacklists | Draft runs, script tests on small volumes without real output |
In practice, for the calculated 10,000 companies, the optimal combination is residential proxies with rotation by city subnets: you get IPs from the same region as the target audience in the cards, reducing the share of captchas and repeat requests that we accounted for in the traffic calculation above. Datacenter proxies can only be used in this task during the script testing phase on 50-100 cards, to avoid burning the budget of residential traffic on debugging.
Data Collection Setup: Tools and Step-by-Step Scheme
To collect a database of companies from maps, agencies usually use ready-made no-code and low-code tools instead of writing a parser from scratch. Among popular solutions are A-Parser (flexible template settings for Google Maps, 2GIS, Yandex.Maps without coding), Octoparse (visual script builder for collection), and ZennoPoster (automation templates with proxy and anti-detect profile support). For small volumes, cloud parsing services with payment based on the number of cards are suitable.
The step-by-step scheme for launching the collection of 10,000 companies looks like this:
- Define the list of cities and niches for which leads are needed — this sets the number of search queries and allows for a more accurate traffic forecast using the formula from the section above.
- Set up a list of residential proxies in the parser with IP rotation every 5-10 requests — most tools support proxy lists in the format login:password@host:port.
- Enable random delays between requests of 3-7 seconds and rotate User-Agent — this reduces the captcha rate by 30-40% compared to fixed intervals.
- Run a test run on 100-200 cards and record the actual traffic consumption — compare it with the calculated one to adjust the forecast for the full volume.
- Scale the collection to 10,000 cards, breaking the task into batches of 1,000-1,500 entries — this simplifies error control and re-launching in case of failures.
- Export the results to CSV or directly to CRM, removing duplicates by phone number and website — duplicates often occur on maps due to network organizations and branches.
If the collection task is repeated regularly (for example, monthly updates of the niche database), it makes sense to set up an automatic launch on a schedule with a constant pool of residential IPs — this saves time on manual restarts and reduces the risk of bans by evenly distributing the load over time.
How to Reduce Traffic Consumption Without Losing Leads
The calculated 2.5 GB for 10,000 companies can be reduced by 30-40% without losing data quality by applying several practical techniques.
First, disable image loading at the request level — photos of establishments can occupy up to half the weight of a detailed card, but they are not needed for sales. Second, use direct requests to the internal map API instead of full page rendering in a browser — this removes unnecessary scripts, styles, and fonts that do not carry useful data. Third, filter duplicates before parsing detailed cards: if a company has already appeared in another area of the same city, do not make a repeat request to its page — check by name and address at the list stage. Fourth, first collect lightweight lists across all cities, and request detailed cards only for companies that passed the rating or category filter — this is especially useful if you need not all 10,000, but only the top 30% in relevance.
These techniques reduce not only the volume of traffic but also the number of requests to the map servers, thus reducing the risk of blocking. For the agency, this means lower proxy traffic consumption at the same rate and the ability to collect more cities for the same budget.
Common Mistakes When Collecting Leads from Maps
The first mistake is starting the collection from one static IP without rotation. Even with a low request speed, the map records abnormal activity after 200-300 consecutive requests, and further collection becomes impossible without changing the IP.
The second mistake is ignoring the regional correspondence of proxies and the target city. If you collect companies in Kazan through an IP from another country, the map often shows irrelevant results or immediately blocks the request as suspicious.
The third mistake is the lack of traffic calculation before launching. Teams often start collecting "by eye" and hit the proxy traffic limit halfway through the task, losing already collected data due to session interruption.
The fourth mistake is collecting without deduplication on the fly. When parsing 10,000 entries across several overlapping requests (for example, "cafe" and "restaurant" in one area), duplicates can account for 15-25% of the database, artificially inflating the estimated traffic and actual costs.
The fifth mistake is using the same User-Agent and headers for all requests in the session. Anti-bot systems of maps analyze not only the IP but also the consistency of the browser's digital fingerprint, so for large volumes, it is recommended to combine proxy rotation with an anti-detect browser and change the fingerprint along with the IP.
Conclusion
Collecting 10,000 companies from maps is a predictable traffic task: when parsing detailed cards from Google Maps, the consumption is about 2.5 GB considering repeat requests, while for 2GIS and Yandex.Maps, the figure is 20-30% lower. Knowing this number in advance allows for precise budget planning for proxy traffic and avoids situations where collection is interrupted halfway due to exhausted limits.
The main factor for successful collection is not saving on traffic, but the correct choice of proxies. For volumes of several thousand cards per day, we recommend residential proxies with rotation by cities — they reduce the percentage of captchas and repeat requests, which consume up to 15-20% of traffic. If collection is conducted in niches with particularly strict anti-bot protection, it is worth considering mobile IPs as a supplement to the main pool — this increases session stability with minimal cost increase.