Guide Cybersecurity

Multi-factor authentication (MFA): what it is and which method to use

Multi-factor authentication (MFA) adds an important layer of protection to business accounts, reducing the risk of unauthorised access when passwords are stolen, reused or guessed. How well it works depends on the method you choose, how many users and applications it covers, and how secure your recovery processes are. This guide compares the main MFA methods and explains their limitations; a companion guide sets out a practical rollout sequence, including recommendations for Microsoft 365 and Microsoft Entra ID.

30 May 2026
Updated 18 November 2026
8 min read

What is multi-factor authentication (MFA)? MFA is a security method that requires more than one authentication factor, combining categories such as something a person knows, has or is. Its purpose is to reduce the chance of unauthorised access when a password is compromised. Methods such as FIDO2 passkeys offer phishing-resistant protection, while SMS and TOTP codes remain exposed to certain social engineering and interception techniques.

Key takeaways
  • A 2023 Microsoft Research study of Azure AD (now Entra ID) accounts estimated that MFA reduced the risk of compromise by 99.22%; real-world effectiveness varies by method and by attack
  • FIDO2 security keys and passkeys are phishing-resistant; TOTP codes, push and SMS can be defeated by real-time adversary-in-the-middle phishing
  • Secure admin accounts and recovery functions first, then cover every employee, contractor and external identity, starting with email
  • Policy exclusions, legacy authentication paths and weak recovery processes are the most common gaps; a high registration rate is not the same as real coverage
  • ISO 27001, Cyber Essentials and NIS2 all expect strong authentication; Cyber Essentials requires MFA on cloud services wherever the service offers it

Why passwords alone keep failing

A password is a single secret. The moment that secret leaves your control, your account is open. Passwords leave your control through channels you have no visibility into.

Credential stuffing is the most common attack vector against business accounts. Attackers buy databases of email and password combinations leaked from breached websites, then run automated tools that test those credentials against Microsoft 365, Google Workspace, VPN gateways, and banking portals. The attack works because people reuse passwords. A credential leaked from a retail breach five years ago still opens accounts today, because the password never changed. HaveIBeenPwned currently indexes over 13 billion breached accounts.

Phishing hands credentials directly to attackers. A convincing login page for Microsoft 365 or your company's VPN portal takes under an hour to build and distribute. The user types their username and password into what looks like a legitimate form. Attackers capture the credentials in real time. No technical vulnerability required.

Password spraying takes the opposite approach to stuffing. Instead of trying many passwords against one account, attackers try one or two common passwords against thousands of accounts. Common choices like Summer2024! or Welcome1 succeed at scale because password policies that require a capital letter and a number produce predictable patterns, and a slow spray stays below lockout thresholds.

MFA stops most of these attempts, because a stolen or guessed password is no longer enough on its own. It is not absolute. Real-time adversary-in-the-middle (AiTM) phishing kits proxy the genuine sign-in page, relay the user's password and one-time code or push approval, and capture the session cookie that results. SMS, TOTP and push-based MFA can all be defeated this way; we cover the techniques and defences in how attackers bypass MFA. For administrators, finance staff and other high-risk accounts, use phishing-resistant methods.

The most cited figure comes from a 2023 Microsoft Research study, How effective is multifactor authentication at deterring cyberattacks?, of commercial Azure AD (now Entra ID) accounts investigated for suspicious activity in 2022. It estimated that MFA reduced the risk of compromise by 99.22% in that population. That is a measured result for that population, not a guarantee for every method or attack.

99.22%
estimated reduction in compromise risk for MFA-enabled accounts (Microsoft Research, 2023)
13B+
breached accounts indexed by HaveIBeenPwned, available for stuffing attacks
81%
of hacking-related breaches involved stolen or weak passwords (Verizon DBIR 2017)

What MFA is and how it works

Multi-factor authentication requires a user to present two or more verification factors from distinct categories before gaining access. The three categories are:

  • Something you know: a password, PIN, or security question answer
  • Something you have: a physical device such as a phone running an authenticator app, a hardware security key, or a smart card
  • Something you are: a biometric factor such as a fingerprint or face scan

Two-factor authentication (2FA) uses exactly two of these. MFA is the broader term covering two or more. In practice most people use the terms interchangeably, and most deployments combine a password with a device-based second factor.

An attacker who steals your password does not have your phone. An attacker who finds your phone does not know your password. Compromising both, against a specific target, at the moment of a login attempt, is far harder than stealing a password from a breached database and running it through an automated tool. That is why attackers now focus on the sign-in itself, and why the choice of method matters.

MFA methods and their level of protection

MFA methods differ most in one property: whether they resist phishing. NIST SP 800-63B-4 treats an authenticator as phishing-resistant only when its output is cryptographically bound to the genuine service or session, which rules out manually entered codes and push approvals. The right choice depends on the risk of the account and the assurance you need, not on a universal ranking.

Method and main limitation
Phishing
✓
Device-bound passkey or FIDO2 security key
YubiKey, Google Titan Key, and platform authenticators such as Windows Hello for Business. The private key stays on the device. Limitation: enrolment, lost-key recovery and device management need planning.
Phishing-resistant
✓
Synced passkey
Passkeys stored in an Apple, Google or password manager account and synced across devices. Limitation: protection depends on the security of that credential provider account and on how sync and recovery work, so assurance is lower than a device-bound key.
Phishing-resistant
×
Authenticator app code (TOTP)
Microsoft Authenticator, Google Authenticator and similar. Six-digit code that changes every 30 seconds and works offline. Limitation: a phishing proxy can relay the code in real time.
Not phishing-resistant
×
Push notification with number matching
Microsoft Authenticator, Duo. The user enters the number shown at sign-in into the app. Limitation: it stops accidental approvals, but a user can still be talked into approving, and AiTM proxies relay the approval.
Not phishing-resistant
×
SMS or voice code
Code sent by text message or phone call. Limitation: SIM swap, number porting fraud and interception. NIST classes it as a restricted authenticator. Better than no MFA; not suitable for high-value accounts.
Not phishing-resistant

FIDO2 security keys and passkeys resist phishing because each credential is scoped to the relying party ID, the domain of the genuine service, and the browser checks the origin of the page before it lets the authenticator respond. A look-alike phishing domain cannot obtain a valid response for the real service, even if the user is fooled. TOTP codes and push approvals carry no such binding: an adversary-in-the-middle proxy can capture the password and the code, or trigger the push, and replay them within seconds.

Number matching, which Microsoft now applies to all Microsoft Authenticator push notifications, makes the user type the number shown at sign-in. That stops MFA fatigue attacks, where repeated prompts are sent until a tired user taps "Approve". Push remains open to social engineering, such as a caller posing as the helpdesk.

Do not rely on SMS for business-critical accounts

SIM swap fraud is well documented: an attacker persuades a mobile operator to move your number to a SIM they control, and every SMS code follows it. Operator identity checks vary and are outside your control. Move business-critical accounts to phishing-resistant methods, or at minimum to an authenticator app, and remove SMS as a fallback option for those accounts.

How to roll out MFA

Our companion guide on rolling out MFA sets out the deployment sequence, starting with admin accounts and recovery functions, along with the four mistakes that leave the largest gaps, the Microsoft 365 and Entra ID settings to configure, and what ISO 27001, Cyber Essentials, NIS2 and GDPR expect.

Frequently asked questions

Is SMS-based MFA worth using?

SMS MFA is far stronger than no MFA at all, because a stolen password alone is no longer enough for automated attacks. The problem is that SMS is vulnerable to SIM-swapping, where an attacker convinces a mobile operator to transfer your number to a SIM they control, to interception on telecoms networks, and to real-time phishing proxies that simply relay the code. For low-risk accounts with no better option available, SMS MFA is acceptable. For email, VPN, finance systems, or any admin account, use an authenticator app with number matching at minimum, and phishing-resistant methods such as FIDO2 security keys or passkeys for high-risk accounts. The goal is to move users off SMS over time, not to leave accounts unprotected while you wait for the perfect solution.

What is the difference between TOTP and FIDO2?

TOTP (Time-based One-Time Password) generates a six-digit code in an authenticator app every 30 seconds, based on a shared secret and the current time. It works offline and is supported by almost every platform. The weakness is that a phishing page can capture both the password and the TOTP code in real time and replay them before the code expires. FIDO2 is the standard behind hardware security keys like YubiKey and behind passkeys on modern devices. Instead of a code, the authenticator signs a challenge with a private key that is scoped to the relying party ID, the domain of the genuine service, and the browser checks the page origin before the authenticator responds. A phishing page on a different domain cannot obtain a valid response for the real service, which makes FIDO2 phishing-resistant in a way TOTP is not. For accounts that are high-value targets, FIDO2 is the right choice.

How do we handle MFA for shared accounts?

Shared accounts create a real problem for MFA because the second factor is registered to one person, blocking everyone else, and nobody is individually accountable. The correct fix is to eliminate shared accounts wherever possible. Most platforms support shared mailboxes, role-based access, or delegated permissions that team members reach through their own individual credentials. For shared admin accounts, give each administrator an individual account and use Microsoft Entra Privileged Identity Management (PIM) for just-in-time role activation. Where a shared account is truly unavoidable, a PAM (privileged access management) tool can broker access to it after individual authentication. Some organisations use a shared authenticator on a dedicated device stored in a secure location. None of these is as clean as individual accounts, but all are better than one person holding the MFA token for an account the whole team uses.

Ryland Deakin
About the author
Lead Consultant, Cyvra · CISM · CompTIA Security+ · MCP

Ryland has delivered cybersecurity, compliance, and IT management programmes for regulated organisations across the UK and the Netherlands for over 20 years. View full profile

Talk to Cyvra

Get MFA deployed correctly across your business

We configure MFA and Conditional Access for Microsoft 365 environments across the UK and EU. Contact us to scope the work.

Disclaimer: This article is for general informational purposes only and does not constitute legal, regulatory, or professional advice. Cyvra makes no warranty as to the accuracy or completeness of this content, which may not reflect the most current regulatory developments. Readers should seek independent legal and regulatory advice appropriate to their specific circumstances. Cyvra accepts no liability for any loss arising from reliance on this content.