What it means
A customer opens an account online using a government identity document, and the service checks the document and asks for evidence that the applicant is the rightful holder. Passing a password test later does not retrospectively prove the original enrolment was correct.
Define the purpose and risk first: a low-stakes newsletter may need only an email check, while a regulated financial account may require stronger identity proofing under applicable rules, and sensitive documents should not be collected without a valid need. NIST's digital identity guidance separates identity resolution, validation and verification, and verification confirms the person presenting evidence is associated with the identity it was issued to.
The full process and assurance level depend on context. NIST's identity-proofing requirements discuss risk-based controls, options for applicants and exception handling, but they are guidance within a particular standards framework, so local legal requirements and business risks still need review.
Choose evidence appropriate to the population, because a document, trusted account or other verified attribute may be available but not everyone has the same device or ID. Offer legitimate alternatives where the service and rules permit.
Check authenticity and validity as well, since a clear image of a card is not enough if it is forged, expired or belongs to someone else, and sources and methods need testing against fraud patterns. Link the applicant to the evidence using a supervised review, comparison or trusted digital assertion, depending on risk and policy; a selfie alone is not a universal identity proof.
Protect against presentation attacks, because a photograph, replayed video or stolen credential can imitate a person. Liveness controls may help, but their error rates and accessibility must be evaluated.
An illustrative verification completion rate is applicants who complete the defined proofing steps divided by applicants who start them. If 700 of 1,000 complete, the rate is 70%, which says nothing by itself about false acceptance or exclusion.
Do not treat a vendor score as certainty: false positives can allow impostors, false negatives can exclude genuine applicants with poor cameras or name variations, so test performance across relevant groups and give unusual cases an accessible human review route. Separate verification from authentication: verification at enrolment asks whether the claimed identity and applicant match, while authentication later asks whether the person accessing an existing account controls an approved credential.
Keep data minimal and record which checks were used and why any exception was approved, without retaining unnecessary raw biometric or document data. For an owner, the right process proves enough for the task and explains failures, and a verified identity does not automatically establish creditworthiness, product eligibility or consent to use data for marketing.
In practice
Real-world examples.
Example
An online lender asks an applicant to upload a government identity document. The service checks the document's authenticity and then uses a holder-linking step to confirm the applicant is the person it was issued to. Only after both steps does the account become active.
Example
A genuine user fails an automated camera check because of poor lighting on an older phone. The service offers an alternative supervised review path rather than rejecting the applicant outright. The reviewer records the reason for the exception and the evidence relied on.
Example
A bank verifies a customer's identity at enrolment, and a sales team later assumes this means the customer qualifies for a premium credit product. Risk staff correct the assumption, because a verified identity is not automatic credit approval. Eligibility and consent are separate decisions with their own checks.
Formula
Calculation
Illustrative completion rate = applicants finishing the defined proofing steps / applicants starting x 100. If 1,000 applicants start and 700 finish, the rate is 700 / 1,000 x 100 = 70%. Suppose a further review shows that 14 of the 700 completed profiles were later confirmed as fraudulent, which is 14 / 700 x 100 = 2%. The completion rate and the fraud rate answer different questions, so accuracy and exclusion need separate measures.Case study
Seen in the real world.
In this entirely fictional example, Cedar Finance tests its online enrolment. Some genuine applicants fail due to poor document images, so the team adds a supervised review route and limits document retention. It still checks fraud outcomes separately, comparing the share of applicants who complete verification with the share later flagged as fraudulent.
The team also reviews results across different groups of applicants to see whether any are failing more often than others. The case does not prescribe one identity method or jurisdictional rule. Cedar's lesson is that a smooth sign-up and a safe sign-up are measured differently, and the process needs both measures to stay honest.
Watch out
Common mistakes.
- Confusing a successful account login with proof of real-world identity.
- Treating a selfie or vendor score as infallible evidence.
- Keeping sensitive identity copies without a defined need and retention rule.
Questions
People also ask.
How is verification different from authentication?
Verification links an applicant to an identity; authentication checks access to an account later.
Does everyone need the same method?
No. Methods should fit risk and support legitimate applicants with different capabilities.
Does identity verification decide eligibility?
No. Product eligibility and consent require separate checks.
From the founder's library
Take it further with the book.
Build your financial confidence beyond this definition. Shihan's full-length guide, Accounting Fundamentals, takes the same plain-English approach and turns it into a complete, practical playbook for non-finance managers, business owners and students - with chapter-end quiz answers and presentation slides included.
25% off with code MMHQ25, applied at checkout. Priced in USD - checkout may show the equivalent in your local currency.
View the book and save 25%Related