Last updated: November 2026
Quick answer: To stop fake signups, promo abuse and multi-accounting, combine checks at the point of registration with limits on the reward itself. At signup, score the email (disposable domains, new or risky addresses), the phone number (virtual and VoIP numbers), the network (VPNs, proxies, hosting IPs) and the device (fingerprints linked to other accounts), and add bot detection. Then make rewards harder to farm: tie bonuses to verified actions such as a first paid order, limit one offer per device and payment method, and delay payouts until fraud checks finish.
Free trials, signup bonuses, referral rewards and first-order discounts all work, which is exactly why fraudsters farm them. One person with a script, a list of disposable emails and a residential proxy network can create hundreds of “new customers” in an afternoon. The cost shows up as wasted incentives, skewed growth metrics, inflated marketing spend and, in some sectors, money laundering through new accounts.
| Signal | What to look for |
|---|---|
| Disposable domains, plus-addressing and dot tricks on the same inbox, brand-new addresses with no history | |
| Phone | Virtual and VoIP numbers, numbers recently used on other accounts |
| Network | VPNs, proxies, hosting providers, many signups from one IP range |
| Device | One device fingerprint linked to several accounts, emulators, automation traces |
| Behavior | Instant form fills, pasted values, identical signup flows, odd hours |
| Payment | The same card or wallet across multiple accounts, prepaid cards |
Automated registrations are the cheapest abuse to stop. Put bot detection or a modern challenge on the signup form, and rate limit registrations by IP, device and network. Our guide to how websites detect bots covers the methods, and our comparison of reCAPTCHA, hCaptcha and Turnstile helps pick a challenge.
Email and phone intelligence APIs flag disposable domains, recently created inboxes, VoIP numbers and addresses already seen in fraud. Normalize email addresses (remove plus tags and, for Gmail, dots) before checking for duplicates.
Many abusers hide behind proxies and VPNs to look like separate people. Flagging anonymized networks at signup, rather than blocking them outright, gives your rules a strong input. See how to block VPN users on your website.
Device fingerprinting is the single most useful tool against multi-accounting, because it links many accounts back to one machine even when emails, IPs and names change. Read what device fingerprinting is for how it works and the consent rules involved.
Fraud controls work better when the incentive is harder to farm:
Regulated businesses and high-value programs may need document and selfie verification. Use it as a step-up for risky signups rather than for everyone. Our list of KYC and KYB compliance platforms covers the main vendors.
Abuse rings show up as clusters: accounts sharing devices, cards, addresses or referral chains. Reviewing clusters, rather than single accounts, catches organized farming and avoids punishing one innocent user at a time.
Every check adds friction, so apply them by risk. Most genuine users should sign up without seeing any extra step. Reserve phone verification, challenges and identity checks for signups that already look suspicious, and measure your signup conversion rate before and after each change.
Link accounts through shared signals: device fingerprint, payment method, phone number, normalized email, shipping address and network. Device fingerprinting is usually the strongest single link because it survives changes to email and IP.
For offers and trials, yes, or at least require extra verification. For newsletters and low-risk content, blocking them may cost more legitimate signups than it saves.
It raises the cost, but abusers can buy virtual numbers. Pair phone verification with a check that flags VoIP and recently reused numbers.
Promo abuse is the repeated, improper use of promotions, such as claiming a one-per-customer welcome discount many times through fake or duplicate accounts.