Signup fraud signals that matter
How disposable emails, VoIP numbers, datacenter and VPN IPs and geo mismatch work as fraud signals, and how to combine them into allow, review or block.
The goal is to stop abuse, not users
Signup abuse takes many forms: fake accounts, repeated free trials, promotion farming and card testing. The signals that point to it are also present in the data of many honest users. People use throwaway emails to avoid newsletters, VPNs for privacy, and internet-based phone numbers for a second line.
That is why no single signal should block a signup. The work is to collect several signals, weigh them together and spend friction only where the evidence adds up. This article covers the four signals that carry the most weight and a simple way to combine them.
Disposable email addresses
A disposable address comes from a service that issues temporary inboxes, often with no registration. It is one of the stronger signals for trial abuse, because creating a new one takes seconds. Email Validation flags disposable domains along with role-based addresses, catch-all domains and mailbox checks, at 1 credit per lookup.
Detection depends on lists of known domains, and new services appear regularly, so a clean result does not prove the address is durable. Privacy relay and alias services are a different case, because they forward to a real inbox the person controls, and blocking them turns away real people.
A common policy is to accept a disposable address for a free browsing account and require a durable address before anything with cost attached, such as a trial with usage limits, a coupon or a payment method.
VoIP phone numbers
A VoIP number runs over the internet. Fixed VoIP numbers are common for business phone systems. Non-fixed numbers can be issued to an app user within minutes, which makes them useful to people who want many accounts. Phone Validation returns line type for 1 credit, and for US and Canadian numbers adds carrier, porting and a risk level.
A VoIP flag also has a practical effect: some VoIP numbers cannot receive text messages, so a one-time code may never arrive. Checking the line type before you send the code saves the message and a confused user.
Two numbers from the same block or carrier signing up minutes apart are worth noticing as well. One such pair means little, but a run of them alongside disposable emails is a pattern that you can describe in a rule.
Datacenter and VPN addresses
People rarely browse from a rented server, so a signup from a datacenter address is often a script or a service relaying the traffic. VPN and proxy exits look similar, and Tor exit nodes are published, so they are easy to recognize. IP Intelligence returns location, ISP and ASN with VPN, proxy and Tor flags for 1 credit.
Weigh these flags carefully. Many legitimate users sit behind a VPN, a corporate gateway or a cloud-based device, and server-side calls from partners come from datacenters by design. Treat the flag as a reason to look closer, not as a verdict.
The network owner helps you decide. An address that belongs to a consumer broadband provider or a mobile carrier reads differently from one that belongs to a hosting company, and the ASN in the result tells you which you are looking at.
Geo mismatch
Geo mismatch compares places that should usually agree: the country of the IP address, the country of the phone number, the billing country and the language or time zone of the browser. A user whose IP is in one country, phone is registered in another and card is billed in a third deserves a second look.
Travel, expatriates and mobile roaming produce many harmless mismatches, and IP geolocation is an estimate that is more reliable for country than for city. Compare countries, not cities, and let the mismatch add weight to other signals instead of deciding the outcome.
Know your own product when you set the rule. A service that sells only in one country can treat any foreign IP as a reason for review. A product used by travelers and remote teams should expect mismatches and give them little weight.
Combining signals into allow, review or block
A workable policy has three outcomes. Allow means the signup proceeds without extra friction. Review means the user gets a step-up check or the account is held for a person to read. Block means you refuse the signup, which is reserved for cases where the signals are strong and agree. The rules below are an example to adjust to your own data.
- Allow: no strong signals, or one weak signal such as a VPN alongside an otherwise normal email and phone.
- Review: two signals that agree, for example a disposable email with a non-fixed VoIP number, or a datacenter IP with a country mismatch.
- Block: three or more strong signals together, or a signal such as a known abusive pattern that you have confirmed in your own data.
Use one call and keep good users moving
Signup Check combines email, IP and an optional phone number into an allow, review or block decision with the reasons, for 5 credits per check. Fraud Detection gives a rule-based score from any mix of email, phone and IP signals for 3 credits. Running Email Validation, Phone Validation and IP Intelligence yourself costs 3 credits and lets you write your own rules.
Whatever you choose, keep the reasons with the decision and review a sample of blocks regularly, so you find rules that catch honest users. Offer a way to appeal, such as a support address, and make the review path light: a code to a verified email, or a short wait, is often enough to separate real people from scripts.
Finally, tune the rules on your own outcomes. Compare the signups you allowed with the accounts that later caused chargebacks, abuse reports or support load, and move the thresholds toward what your data shows, not toward what feels cautious.