Biometrics are unavailable for some people
Aliases: biometric exclusion · enrollment failure · cannot enroll fingerprint
What it is
Some people cannot present a usable biometric template: worn or missing fingerprints, certain dermatologic or ophthalmic conditions, faces covered for medical or religious reasons, limb differences, dry older skin, children’s fingerprints not yet stable. Unavailable means enrollment fails or policy excludes them—not that one match missed. A product that treats biometrics as the only way to open an account or the only unlock locks those people out. This entry is about population and enrollment. It is not about how an already-enrolled person exits a failed match, and not about fallback strength.
Why it happens
Biometrics assume a body part is stable, readable by the sensor, and willingly used as a key. None of the three is universal. Industrial work, chemotherapy, and skin disease change ridge detail; some eye conditions and facial change break face templates; public tests have repeatedly shown sensor differences across skin tone and lighting, with magnitudes that vary by hardware and must not be collapsed into a single percentage. Mandatory enrollment reads failure as “the device is broken” or “the person cannot use it,” and support scripts miss the real cause. Unavailability is prior exclusion; retrying the sensor does not repair it.
Studying it
In the enrollment funnel, separate “never started,” “started and failed,” and “succeeded,” and stratify by reportable traits (age band, declared disability, occupational exposure)—not overall pass rate alone.
Independent variables: whether enrollment is skippable, whether failure allows never enabling, sensor modality (fingerprint, face, voice). Dependent variables: enrollment failure rate, share who can still use the product after failure, uninstalls or human support caused by enrollment failure.
Convenient lab samples are young, with kind skin and lighting, and will understate unavailability. Public fairness tests report differences on particular datasets and sensors; they are not field numbers for a given phone. Interviews must allow people to refuse to disclose body reasons; “chose never enable” is the behavioral measure.
Where it stops holding
Government identity and border checks may legally capture biometrics, so in-product skip does not apply; a human lane is still required for people who cannot be captured. When work devices mandate enrollment at hiring, exclusion happens off the app, and another in-app skip is meaningless. Voiceprints fail for some languages and vocal conditions; swapping biometric modality is not a population fix. If biometrics are only a local device shortcut and the account still accepts a password, unavailability does not lock the account—it only removes a fast path.
Applying it
- Treat enrollment as a skippable enhancement that does not block signup or sign-in; offer “don’t use biometrics” and remember it.
- Word enrollment failure as “this sensor could not finish enrollment,” with a control to continue without biometrics, not implying fault in the person’s body.
- Do not treat a second biometric modality as the accessibility substitute; the excluded group may merely change.
- Verify that people who fail or explicitly skip enrollment can still complete core tasks over the next seven days. Toggle mandatory enrollment on and off and compare how many are stuck on the enrollment screen. Walk enrollment with people across disability declarations and age, and record whether inability was environment, hardware, or body condition.