NevTan Sign
NevTan Sign
guide

How to Verify a Signer’s Identity: Email Codes, SMS OTP and ID Checks Compared

How to Verify a Signer’s Identity: Email Codes, SMS OTP and ID Checks Compared
NS 9 min read

Most e-signature platforms list “identity verification” options that don’t verify identity. An SMS code proves someone controls a phone number. An email link proves someone can open an inbox. Neither establishes who the person is.

That distinction — authentication versus identity proofing — is the whole subject. Get it right and the rest of the choices follow naturally.

Key takeaways

  • Authentication proves control of a credential. Identity proofing establishes who someone is. Most “verification” options do the first.

  • NIST separates these as AAL (authentication) and IAL (identity proofing). They’re independent choices.

  • Knowledge-based verification is no longer acceptable for identity proofing under the current NIST standard.

  • SMS is the only authenticator NIST classes as restricted — permitted but conditioned.

  • Every added step costs completions. Match assurance to risk rather than applying one level everywhere.

  • Some regulators mandate a level regardless of your preference.

The distinction that makes this simple

Identity proofing happens once: establishing that a real person with this name exists and that the person in front of you is them. Checking a passport is identity proofing.

Authentication happens each time: confirming that whoever is acting now controls the credential established earlier. Entering a code is authentication.

The current standard splits these explicitly. Identity Assurance Level (IAL) refers to the identity proofing process; Authentication Assurance Level (AAL) refers to the authentication process. Organizations select a level for each function separately. nistnist

Why this matters for signing: you can have strong authentication of a weakly proofed identity. An SMS code sent to a phone number the signer typed in themselves authenticates that number perfectly — and tells you nothing about who they are. If the number was never tied to a verified identity, the chain starts nowhere.

The methods compared

Method

What it proves

What it doesn’t

Friction

Email link

Access to an inbox

Who the person is

None

Access code (sender-supplied)

Knowledge of a shared secret

Identity; only as strong as how you shared it

Low

SMS OTP

Control of a phone number

Identity; vulnerable to SIM swap

Low

Knowledge-based (KBA)

Ability to answer record-derived questions

Increasingly little — see below

Medium, high failure

ID document check

A valid-looking document was presented

That the presenter is its holder

Medium

ID + liveness biometric

Document holder is present

Subject to injection and deepfake attack

Medium-high

Qualified certificate / eID

Identity proofed by a trust service

—

High

Email link only

The default on nearly every platform. The document goes to an address; whoever opens that inbox can sign.

Reasonable for: internal documents, established relationships, routine low-value agreements, anything where the relationship itself is the context.

Weak where: the email account may be shared or compromised, the address was self-supplied, or you’d need to prove which individual signed.

Worth noting: under the current NIST authentication volume, email OTP is no longer an acceptable authenticator. That’s a statement about authentication standards rather than about signature validity — an email-delivered signature remains legally valid under ESIGN — but it tells you how little assurance to attribute to it.

Access codes

The sender sets a code and shares it through a different channel — a phone call, an in-person handover, an existing portal.

This is genuinely useful and underused. Its strength is entirely the channel you used. A code read over the phone to someone whose voice you know is meaningful. A code sent in the same email as the signing link is theatre.

Good for: adding assurance without vendor cost, where you already have a trusted side channel.

SMS OTP

The most commonly offered “identity verification,” and the most commonly misunderstood.

What it proves: control of a phone number at that moment.

Known weaknesses: SIM swap, SS7 network interception, number recycling after disconnection, and carrier social engineering.

Current standing: SMS and voice codes over the phone network are the only authenticator NIST classes as restricted. Restricted doesn’t mean prohibited — a service can use SMS at AAL2, but only if it offers a non-restricted alternative, warns users of the risks, covers SMS in its risk assessment, and maintains a migration plan. SMS OTP is not permitted at AAL3. NIST is explicit that this is a risk acceptance rather than a security recommendation, and revision 4 does not rehabilitate it.

Honest assessment: SMS meaningfully raises the bar against casual misuse and costs little friction. It is not identity verification and shouldn’t be described as such. Use it where the marginal improvement over an email link is worth the cost, and don’t rely on it where identity is genuinely contested.

Knowledge-based authentication, and a real problem

KBA asks multiple-choice questions generated from credit and public records — former addresses, loan amounts, vehicles.

It is no longer acceptable for identity proofing. Under the current standard, knowledge-based verification SHALL NOT be used for identity verification, although it may support fraud management. It has been removed as an acceptable identity proofing mechanism across all assurance levels. nhimgauthlayer

The reasoning is straightforward: the underlying data has been breached repeatedly and is purchasable. Questions an attacker can look up aren’t a secret.

It also fails legitimate signers at meaningful rates — thin credit files, recent movers, young adults and elderly signers all struggle, and a single wrong answer typically fails the whole set.

The conflict worth knowing about. IRS Publication 1345 requires identity verification meeting NIST IAL2 for remote signing of Form 8879, and KBA has been the standard route to satisfying it. The current NIST standard now excludes KBV from identity verification. That leaves tax practitioners caught between a regulation pointing at a standard and a standard that has moved. Watch for IRS guidance, and in the meantime follow your platform’s and your counsel’s reading. Our guide for accountants covers the 8879 requirements and the wet-signature fallback, which avoids the question entirely.

ID document verification

The signer photographs a government ID. The system checks security features, fonts, data consistency and tampering indicators.

Stronger, because it’s actual identity evidence rather than a proxy for it.

Its weakness is presentation. Document checks confirm a document looks genuine; they don’t confirm the presenter is its holder. Which is why the next step exists.

ID plus liveness

The signer photographs their ID, then takes a selfie or short video. The system matches face to document and checks for liveness.

This is the current practical standard for remote identity proofing at meaningful assurance. It establishes identity evidence and ties the person present to it.

The honest caveat: generative tools and injection attacks are improving against liveness detection. Vendors are responding, but it’s an active contest rather than a settled problem. Ask vendors specifically how they handle injection attacks — the answer tells you how seriously they take it.

Where supervision is required, remote online notarization adds a notary to the session: how remote online notarization works.

Qualified certificates and national eID

The highest tier: identity proofed once by a qualified trust service or national scheme, then a certificate used for signing.

Under eIDAS, qualified electronic signatures carry the legal equivalence of handwritten signatures by default. Types of electronic signatures covers the levels.

Friction is high and coverage is uneven, which is why this is usually reserved for transactions that require it. European signers may already hold credentials through national schemes; US signers generally won’t.

Matching assurance to risk

Don’t apply one level to everything. Tier it:

Document

Suggested

Internal acknowledgments, routine NDAs

Email link

Standard client agreements

Email link, or access code

Employment and HR documents

Access code or SMS

Significant financial commitments

ID verification

Property, lending, regulated transactions

ID plus liveness, or as mandated

Anything a regulator specifies

Whatever they specify

Two inputs: the cost if the wrong person signs, and the likelihood of dispute. High on either moves you up.

Regulatory requirements override your judgment. Several regimes specify a level — IRS rules for certain tax forms, healthcare settings for sensitive authorizations (clinics and therapists), lending and property rules by jurisdiction. Check before deciding.

The completion cost is real

Every verification step loses signers. Not a reason to skip verification — a reason to apply it deliberately.

What helps:

  • Tell signers in advance. An unexpected ID request mid-flow reads as a scam and people abandon.

  • Explain why. “Because this authorizes a payment” converts better than an unexplained demand.

  • Have a fallback. Verification fails for legitimate people. Know your alternate route before you need it.

  • Don’t stack unnecessarily. SMS plus KBA plus ID is three failure points for assurance that one good method delivers.

Personalised signing experiences and forms and fields that get signed cover reducing friction elsewhere, which buys you room for verification where it matters.

The thing verification doesn’t fix

If an attacker impersonates you and sends a fake signing request, your verification settings are irrelevant — the signer is authenticating into the attacker’s system.

Defences are separate: domain authentication, consistent tooling, and telling counterparties what genuine requests look like. See spotting fake signature requests and phishing protection. And enable MFA on your own signing accounts — a compromised sender account lets someone send genuine requests in your name.

Also note: clickwrap identifies an account, not a person. If identity matters, clickwrap isn’t the mechanism.

Common mistakes

  1. Calling SMS “identity verification.” It’s authentication of a number.

  2. Sending the access code in the same email as the link.

  3. Relying on KBA for identity proofing, which the current standard excludes.

  4. One level for everything, usually too low where it matters and too high where it doesn’t.

  5. Surprising signers with verification they weren’t warned about.

  6. No fallback when verification legitimately fails.

  7. Stacking methods instead of choosing a good one.

  8. Treating verification as fraud protection, when the common attack targets the sender.

FAQ

What’s the difference between authentication and identity verification?
Identity proofing establishes who someone is, usually once. Authentication confirms control of a credential each time. NIST treats them as separate assurance levels — IAL and AAL — chosen independently.

Is SMS OTP secure enough for signing?
It raises the bar cheaply but proves only control of a phone number. NIST classes SMS as its only restricted authenticator — usable at AAL2 under conditions including offering an alternative and warning users, and not permitted at AAL3.

Is knowledge-based authentication still acceptable?
Not for identity verification. The current NIST standard states KBV shall not be used for identity verification, though it may support fraud management. Breached data has made record-derived questions unreliable.

What’s the strongest practical remote method?
ID document verification combined with a liveness biometric check, for most commercial purposes. Qualified certificates are stronger where available and required.

Does an email link verify identity?
No. It confirms access to an inbox. Legally valid for signing, but weak evidence if identity is contested.

How do I choose a level?
Weigh the cost of the wrong person signing against the likelihood of dispute, then check whether a regulator mandates a level. Tier by document type rather than applying one setting.

Will verification hurt completion rates?
Yes, somewhat. Warn signers in advance, explain the reason, and keep a fallback route for legitimate failures.

Does verification stop signature fraud?
Not the common kind. Attackers impersonate the sender, so domain authentication and recipient awareness matter more than how you verify signers.