Median KYC completion time across 14 regulated markets now sits at 6 minutes 12 seconds, according to operator-side data pooled from 2024 onboarding sessions. The same dataset shows 43% of players who reach the liveness selfie step abandon before submitting a usable image. The bottleneck isn't document review, it's the camera.

Where the six minutes actually go

Strip out the queue time and the arithmetic gets uncomfortable. Document upload and OCR verification consume roughly 90 seconds. Database checks against sanctions and PEP lists run in the background and rarely block anyone. The selfie step, by contrast, eats a median 4 minutes 40 seconds — and that figure only counts users who finish it.

The 43% who don't finish aren't a random sample. Mobile users on older Android handsets abandon at nearly double the rate of iOS users. Players in low-bandwidth regions bail more often than those on fibre. And the drop-off clusters hard in the 18–24 bracket, which is precisely the cohort operators spend the most on acquisition.

Why liveness detection is the hard part

Passive liveness — the kind that runs silently while a user holds still for two seconds — has improved. Active challenges, where the screen asks you to blink, turn, or read a phrase, have not. Every additional instruction is another chance to fail.

Three failure modes dominate:

• Lighting mismatch. A document photographed under kitchen light, then a face captured in the same kitchen, can still trip a model trained on studio conditions.
• Glare and reflection. Passports with laminate covers are the worst offenders. Roughly one in nine rejections traces to a single specular highlight.
• Accessibility gaps. Users with motor tics, certain visual impairments, or facial differences fail liveness checks repeatedly, and most flows offer no fallback beyond "try again."

That last point is a compliance problem, not just a UX one. Regulators in several jurisdictions now expect an alternative path, and "retry the selfie" doesn't qualify.

What the fast operators do differently

The gap between the quickest and slowest operators in the sample is 11 minutes. The quick ones share a few habits.

They capture the selfie before the document, while the user is still fresh and the app has full camera permissions. They run document and face matching in parallel rather than sequence. They accept a short video clip as a fallback when a still image fails twice. And they surface a real human within 90 seconds when automation gives up — not a ticket number.

One mid-sized operator cut selfie abandonment from 41% to 19% over two quarters by doing nothing more exotic than adding a live preview with framing guidance and killing the blink challenge entirely.

The number nobody publishes

Here's the uncomfortable part. Completion rates are measured on users who reach the selfie step. Nobody publishes the denominator — how many people opened the app, saw a verification wall, and closed it before starting. That figure is almost certainly worse than 43%, and it never appears in a compliance report.

So the question isn't whether 6 minutes is good. It's whether the industry is optimising a funnel it's measuring from the wrong end. If a quarter of your intended customers never begin, shaving 40 seconds off document review is rearranging deck chairs.

For players, the practical takeaway is dull but useful: have your ID out, find decent light, and use a phone made in the last four years. For operators, the takeaway is less comfortable. The selfie step is where growth goes to die, and most of them are still treating it as a compliance formality rather than a product decision.