> ## Documentation Index > Fetch the complete documentation index at: https://developers.soax.com/llms.txt > Use this file to discover all available pages before exploring further. # IP Quality & Fraud Score > What fraud score is, why it matters for proxy use, how different proxy types compare, and how SOAX maintains pool quality. Fraud score is one of the most misunderstood concepts in proxy infrastructure. This page explains what it is, what affects it, and what it means for your use case. ## What is a fraud score? Fraud score is a reputation metric that anti-fraud systems assign to an IP address. It reflects how likely that IP is to be associated with malicious or automated activity — spam, credential stuffing, ad fraud, scraping, and so on. Several services maintain these scores: IPQualityScore, Scamalytics, IPInfo, MaxMind, and others. Websites and platforms use them (often without telling you) to decide whether to serve content, trigger a CAPTCHA, flag an account, or block a request entirely. A low score means the IP looks clean. A high score means it's been flagged — either because it was previously used for something suspicious, or because it belongs to a subnet with a bad reputation. ## How proxy types compare Not all IPs are treated equally by fraud scoring systems. The IP's origin is a major factor. **Residential IPs** come from real home connections assigned by ISPs to real subscribers. Fraud systems treat them as legitimate user traffic by default. They have the lowest fraud scores of any proxy type — unless a specific IP has been abused previously. **Mobile IPs** come from carrier connections. Because carriers use NAT to share a single IP across hundreds of real users, fraud systems are reluctant to flag mobile IPs aggressively — blocking one could mean blocking thousands of legitimate users. Mobile IPs tend to have very low fraud scores. **ISP proxies** (static residential) are hosted in datacenters but registered to ISPs rather than cloud providers. They score better than datacenter IPs but not as well as genuine residential connections. **Datacenter IPs** are the easiest to identify. ASN lookups immediately reveal the hosting provider (AWS, GCP, DigitalOcean, etc.). Most fraud scoring systems assign datacenter IPs a high baseline score regardless of their history. This doesn't mean they're useless — many targets don't check fraud scores — but for anything that does, datacenter IPs are the most likely to be blocked. ## What else affects an IP's fraud score **Usage history.** If an IP was previously used for spam or credential stuffing, that history follows it. This is why IP pool quality matters — a provider that doesn't monitor for abuse will accumulate IPs with bad histories. **Subnet reputation.** Fraud systems don't just look at individual IPs — they look at the entire subnet. If many IPs in a /24 block have been flagged, all IPs in that range get a higher baseline score. **IP age and stability.** IPs that have been assigned to the same subscriber for a long time tend to score better than freshly allocated IPs. Residential IPs are more likely to have stable histories. **Concurrent usage patterns.** An IP that's handling hundreds of requests per minute looks different from a typical home connection. Some fraud systems factor in behavioral signals in addition to static reputation. ## How SOAX maintains pool quality SOAX actively monitors the proxy pool and removes nodes that show signs of degraded reputation. This includes: * Checking IPs against major fraud scoring databases. * Monitoring for abuse patterns within the network. * Removing nodes that are generating unusual error rates or getting blocked at abnormally high rates. This is ongoing work, not a one-time check. The pool changes constantly as new nodes join and underperforming nodes are removed. That said, no pool is perfect. If you're targeting sites that do aggressive fraud scoring, you may still encounter IPs with elevated scores — especially in smaller geos where the available pool is limited. The practical approach is to treat elevated block rates as a signal to rotate to a fresh IP rather than expecting every IP to be pristine. ## What this means for your setup **If you're scraping public data at scale:** fraud score matters less. Most scraping targets block based on request patterns, not IP reputation. Focus on rotation strategy and realistic headers. **If you're doing ad verification or brand protection:** fraud score matters more. The target systems are specifically designed to detect non-human traffic, and they use fraud score as one signal. See [Choosing the right proxy type](/getting-started/choosing-proxy-type) to pick the best fit for your use case. **If you're working with platforms that create or manage accounts:** fraud score is critical. Platforms that tie accounts to IPs (social media, e-commerce, marketplaces) use fraud scoring to flag suspicious activity. ## Next steps How proxies affect account-level bans and how to reduce that risk. What to do when the target site is blocking your requests. What affects response times and how to optimize for speed. Compare residential, mobile, ISP, and datacenter proxies.