Age estimation

Security, compliance, and improved customer experience for age-sensitive services.

Sumsub’s Age Estimation Solution uses in-house liveness technology to define an individual's age based on their facial features:

  • The technology is 99.9% accurate for filtering out underaged users (if a 7-year buffer is applied).
  • The average error across all ages is within the 2.9-year limit.
  • Age estimation quality remains consistent regardless of gender, skin tone, and nationality.
  • The average inference time for the age estimation algorithm is less than one second (95th percentile is 900 milliseconds; median is 600 milliseconds).

Apart from these, Sumsub’s Age Estimation Solution can detect sophisticated fraud attempts with video footage, masks, and deepfakes.


Did you know?

Sumsub is approved by the German Commission for the Protection of Minors in the Media (KJM) as an age verification solution in line with German regulatory standards.

How it works

Sumsub’s Age Estimation solution is based on AI and uses probabilistic models with continually improving accuracy.

Liveness technology is an advanced biometric technology that captures the user’s face from all angles, creating a 3D map.

The Sumsub AI uses the map to generate multiple age estimations for each map. Based on the accuracy of the estimation, the applicant is attributed to an age cohort.

Cohorts in Sumsub’s age estimation model

Infants and toddlersChildrenTeenagers

Starting with age 16, each age is considered a separate cohort, while ages 70+ are classified as seniors.
Cohorts can be used in your verification flows to reject certain age groups, route to additional ID verification, or automatically approve users who are definitely determined to be of age.

Get started

Use age estimation to route applicants to different verification flows depending on their age and the probability level established by the system:

  1. Contact our support at [email protected] to have the age estimation feature turned on.
  2. Set up the WebSDK and/or MobileSDK integration.
  3. Create a verification level and add Age estimation as a verification option. Contact your Customer Success Manager or Sumsub Support team if you need any assistance or if the Age verification option is unavailable in your subscription.
  4. Create multifaceted flows no-code using the Workflow builder. to guide the applicant through verification.
  5. Review verification results.

Review verification results

To review the verification results:

  1. Open the Applicants page and select an applicant who passed age estimation.
  2. Scroll down to the Age Estimation results tab in the profile.


Leverage age estimation to speed up verification and increase conversions. The algorithms usually allow for a non-significant 1-2 year difference. You can use these estimations to tailor the verification flow.

To identify underage users accurately, we suggest using an age buffer of 3, 7, or 10 years, depending on the sector. For example, implementing a 10-year buffer would mean that any user who appears to be 28 years old or younger will be asked to present an ID. In this case, our algorithm has an error rate of only 0.009% (1 underage individual per 1121 people).

If your case allows for fewer restrictions, we propose using a 7-year buffer, which means having an age threshold of 25 years.

Lastly, if the age threshold is 21 years, a 3-year buffer results in an error rate of 0.02% (1 in 540 people).

This strategy can be easily implemented through our condition-based Workflow builder.