Web scraping over a French mobile IP

Why datacenter IPs get blocked

Site operators have nothing against you personally: they are defending against robots. Lists of hosting addresses are public and free. When a request arrives from an OVH, Scaleway, AWS or Hetzner range, it carries a "server" label before any content is served. The application firewall then applies a different policy: CAPTCHA, slow-down, empty page.

A French mobile address carries no such label. It is assigned dynamically by the carrier to one subscriber among thousands, exactly like the legitimate visitors the site is trying to serve. That is why the success rate changes so dramatically on the same targets.

Configure your scraper in 30 seconds

With requests and an HTTP proxy, pass the proxy URL to the proxies argument of your request call. With Playwright, set the SOCKS5 server with your username and password in the launch options. For a quick check from the command line, run: curl -x http://USER:PASSWORD@proxy.proxymobile.fr:28080 https://api.ipify.org

The response should show the French mobile address instead of your own.

Five practices that make the difference

1. Respect the target. Read robots.txt, stay on public data, do not defeat paywalls and do not collect personal data. That is a legal obligation under the GDPR as well as a technical survival question.
2. Space out your requests. A random delay of 1 to 3 seconds between pages beats a burst of 50 requests per second: your traffic looks human and the site does not raise its rate limiter.
3. Keep headers consistent. A Chrome user agent, an Accept-Language of fr-FR and a French mobile IP are coherent. A bot user agent behind a mobile IP is not.
4. Use rotation wisely. Automatic rotation every 5 to 20 minutes covers most collections. Do not force a rotation per request: it burns cellular reconnection time and does not improve the success rate.
5. Watch your volume. One gigabyte is roughly 5,000 to 20,000 HTML pages depending on their weight. The customer portal shows consumption to the gigabyte, and you can add more volume as top-ups.

What the proxy will not do

It does not break CAPTCHAs, it does not defeat advanced behavioural protection, and it does not exempt you from the terms of use of a site. If a platform explicitly forbids automated collection, a mobile IP does not change the nature of the activity, only the probability of being filtered technically. The judgement call is yours.

What it costs to run a scraper

A scraper that runs once a day over 500 pages consumes roughly 0.5 to 2 GB per month, which fits the monthly plan at $14.99. For several scrapers in parallel or heavy page captures, a dedicated phone at $79 brings 100 GB and 256 simultaneous streams.