Your Login Page Is the Bot Target Now. The Numbers Say So.
Bots test stolen username and password pairs against your login page at scale, because password reuse makes a small success rate profitable. Lockout rules never fire, because each pair is tried once. The fix is breach-list checks, bot detection before the password check, and MFA that cannot be fatigued.
The flood, and what it looks like in the logs
Check the traffic on any store login endpoint and the bots outnumber the humans, sometimes by orders of magnitude. Credential stuffing uses stolen username and password pairs from other breaches, betting that people reuse credentials, and the numbers explain why it works. Between September and November 2024, Cloudflare found that 41 percent of successful human logins across sites on its network involved a password that had already leaked in a breach elsewhere, and bots made 95 percent of the login attempts that used leaked passwords. Verizon 2025 DBIR research found credential stuffing made up a median 19 percent of daily authentication attempts in single sign-on provider logs.
In the logs, stuffing looks like a flood: thousands of login attempts in a short window, each with a different username, each with a different password, rotating through IPs and user agents to dodge simple rate limits. The success rate is tiny, well under one percent, but the volume makes it profitable. One thousand attempts at a 0.1 percent success rate is one compromised account, and the botnets run millions of attempts. The tell is credential diversity: a legitimate spike shows the same users retrying, while stuffing shows a thousand unique usernames in ten minutes, because each attempt is a new stolen pair.
Why lockout never fires
The standard defense, lock the account after five failures, never fires against stuffing, because the attack tries each stolen pair once against an account, and uses proxies to hide where the attempts come from. There is no fifth failure on any single account, because the attacker moves on after the first try. Password spraying is quieter still: a few common passwords, Winter2024, Password123, the store name plus 123, tried against many accounts, evenly paced, which is exactly why it works. The signature is password repetition, not failure volume, and the defense that counts failures per account cannot see it.
The cost of a single success is high, because one compromised account means stored payment methods, order history, and loyalty points. In 2024, Roku disclosed that about 576,000 accounts had been accessed through credential stuffing. In January 2025, New York Department of Financial Services announced a 2 million dollar penalty against PayPal over credential stuffing failures. The attack is cheap, automated, and behind most account takeovers, and the login page is where the bet gets placed at scale.
MFA fatigue: the second factor has a hole
Multi-factor authentication is the structural fix, and the OWASP credential stuffing prevention cheat sheet calls it by far the best defense, citing Microsoft analysis that it would have stopped 99.9 percent of account compromises. But not every second factor is equal. A push notification that only asks for approve or deny invites MFA fatigue: the attacker who already holds the password sends dozens of prompts in a row until someone taps approve out of habit or irritation. That is how the 2022 breach at Uber began.
The fix for fatigue is a second factor that cannot be approved by accident: a number-matching prompt, where the user has to type the number shown on the login screen, or a phishing-resistant factor like a hardware key or passkey, which cannot be approved by tapping a notification. Push the strongest factor hardest on accounts with stored payment methods and admin roles, where the payoff of a takeover justifies the friction. And check new passwords against a breach list at registration and password change, as NIST SP 800-63B-4 requires, so reused pairs never become valid on your service in the first place.
The controls that actually change the math
Verify a human before checking the credential. That is the highest-value change, because the attack dies before the password comparison, and the attacker never learns which pairs are valid, and a validated pair is itself a saleable asset. Make attempts cost something: proof of work is negligible for one login and prohibitive across a million. Check credentials against breach corpora at registration and password change. Deploy phishing-resistant MFA on the accounts that matter. And do not rely on account lockout, because it is a denial-of-service tool an attacker can aim at your own users, locking real customers out by failing logins against their accounts on purpose. The bots will keep hammering the login page regardless. The question is whether the hammering can ever succeed, and the controls above are what make the answer no.
Frequently asked questions
What is credential stuffing, exactly?
It is an attack that replays username and password pairs stolen from one breach against other sites, betting that people reuse passwords. Each pair is tried once, so lockout rules never fire, and the volume makes a tiny success rate profitable.
How do I tell stuffing apart from a real traffic spike?
Look at credential diversity. A real spike shows the same users retrying. Stuffing shows a thousand unique usernames in ten minutes, because each attempt is a new stolen pair, rotating through IPs and user agents.
Does MFA stop credential stuffing?
Yes, mostly. Microsoft analysis cited by OWASP says MFA would have stopped 99.9 percent of account compromises. But push-only MFA can be beaten by fatigue attacks, so use number-matching or phishing-resistant factors on high-value accounts.
Why does account lockout not work against stuffing?
Because the attacker tries each pair once and moves on, so no single account accumulates enough failures to trigger the lock. Lockout also becomes a denial-of-service tool, letting an attacker lock real customers out.