Welcome to the first newsletter on the InternetData Proxy IP databases: Residential Proxy IP, Mobile Proxy IP and Datacenter Proxy IP. Approximately every two weeks we'll share what's changed across the three: coverage shifts, what's new behind the scenes that may affect those shifts, and what we're working on right now.
Each of the three is published in six rolling windows (1d, 7d, 14d, 30d, 60d, 90d) so you can choose the freshness/breadth tradeoff that fits your use case. The numbers below are all from the 90d window, the most complete picture we publish.
Database at a glance
| Metric | May 13 | May 27 | Δ |
|---|---|---|---|
| Residential proxy IPs (90d) | 122,714,368 | 135,342,070 | +12,627,702 (+10.3%) |
| Mobile proxy IPs (90d) | 2,710,351 | 2,723,651 | +13,300 (+0.5%) |
| Datacenter proxy IPs (90d) | 2,267,061 | 2,190,174 | -76,887 (-3.4%) |
| IPv6 share of residential (90d) | 47.5% | 53.0% | +5.5 pp |
| Active providers contributing IPs (90d) | 106 | 105 | -1 |
| Commercial proxy providers cataloged | 309 | 338 | +29 (+9.4%) |
The "active providers" row breaks down as 103 residential, 1 mobile, and 5 datacenter, deduplicated by provider ID since four of them span more than one category. The 338-strong catalog is the new automated master list mentioned below in "What changed behind the scenes" - it's the universe of commercial proxy providers we know about, including ones we don't actively cover yet.
Inside the residential pool
The residential 90d window is rich enough to support a few observations about how the commercial residential proxy ecosystem actually behaves. All numbers are from the May 27 snapshot:
- 47.1% of residential proxy IPs are seen exactly once in 90 days: 28.3% for IPv4 vs 63.7% for IPv6.
- Median lifetime in the pool: 5.0 days for IPv4, 1.0 days for IPv6. And the v4 distribution has a long tail - the 75th percentile is 38.4 days, meaning a quarter of all observed IPv4 residential proxy IPs stay visible for over a month. IPv6 by contrast is nearly all single-day: median and 75th percentile both round to 1.0. Practical implication for risk-scoring: an IPv4 hit in this database is a far more durable signal than an IPv6 hit.
- 40.2% of residential proxy IPs appear across more than one provider simultaneously: 60.8% for IPv4 vs 22.1% for IPv6. IPv4 pools are heavily reshuffled across resellers (the same
/32commonly shows up under three or four provider IDs at once); IPv6 attribution is much cleaner, with most addresses unique to a single provider at any given time.
What changed behind the scenes
- High-coverage providers fixed. A significant fix was made for providers with very large pools. We were facing scaling issues covering them, so our coverage of them was highly inconsistent. The root cause was found and fixed across all providers being covered, and 24-hour coverage is now consistently about 6M+ IPs.
