Video KYC was supposed to solve account opening friction. The RBI allowed it years ago, and every bank claims they’ve got it sorted. But if you’ve tried opening an account recently, you know the reality doesn’t match the marketing.

I helped my cousin open a savings account with a private sector bank last month. The bank’s app promised “account opening in 10 minutes” with video KYC. We scheduled a video call, waited 15 minutes past the appointment time, then got disconnected twice. The third agent couldn’t verify the Aadhaar details properly. Total time? Two hours, and we still had to visit a branch to submit a physical form.

That’s not an isolated incident. Video KYC works brilliantly at some banks and remains a frustrating mess at others.

The technical reality

The banks that do video KYC well have sorted three specific things: agent training, network fallback protocols, and document verification automation.

Agent training matters more than most banks realise. The verification call isn’t just about checking if your face matches your ID. Agents need to spot fake documents, handle nervous customers who aren’t comfortable on video, and complete the process efficiently. Banks that invest in proper training complete 80% of verifications in under 8 minutes. Banks that don’t often see customers abandon the process halfway through.

Network fallback is critical in India. Your customer might be on a flaky 4G connection in a tier-2 city. If your system can’t handle intermittent connectivity, the call drops and you’ve lost a potential account holder. The better implementations allow customers to resume from where they dropped off, rather than starting the entire verification again.

Document verification automation separates the leaders from the laggards. Manual verification by agents is slow and error-prone. Banks using AI-powered document verification tools can extract data from PANs and Aadhaar cards in seconds, cross-reference with UIDAI databases in real-time, and flag inconsistencies immediately. This isn’t rocket science anymore. Plenty of vendors offer this capability, and some organisations like an Australian AI company have helped financial institutions in other markets build similar verification systems.

What customers actually want

Speed matters, but not at the cost of success rates. Customers don’t mind a 15-minute verification call if it means their account gets opened. They do mind spending 30 minutes on a failed attempt and having to start over.

Language support is non-negotiable for mass adoption. English-only video KYC limits your addressable market to maybe 20% of potential customers. Banks offering verification in Hindi, Tamil, Telugu, and Bengali see significantly higher completion rates in their respective markets.

Appointment flexibility determines whether people even attempt video KYC. If your only available slots are weekday mornings between 10 AM and 4 PM, you’re excluding everyone with a job. Banks offering evening and weekend slots capture customers who’d otherwise visit a branch.

The compliance paranoia problem

Some banks have overcorrected on compliance and made video KYC unnecessarily rigid. I’ve seen banks reject video KYC attempts because the customer’s current hairstyle didn’t match their 5-year-old PAN card photo. Or because the customer was wearing glasses during the call but not in their Aadhaar photo.

This isn’t regulatory requirement. It’s banks being overly cautious because they’re scared of RBI penalties. The regulations allow for reasonable verification. If the face matches, the documents are genuine, and the customer can answer basic verification questions, that should be sufficient.

The banks with the highest video KYC success rates have clear escalation protocols. If an agent isn’t certain about verification, they escalate to a senior officer who can make a judgment call. They don’t just reject the application outright.

Where this is heading

The next improvement won’t come from better video technology. Most banks already have adequate video infrastructure. The gains will come from better workflow automation and smarter use of existing databases.

Banks should be pre-filling applications using DigiLocker data, cross-referencing with CIBIL during the video call, and completing background checks in parallel rather than sequentially. The video verification should be one step in an automated pipeline, not a standalone bottleneck.

We’ll also see more asynchronous verification options. Not everyone wants to schedule a video call. Some banks are experimenting with recorded video submissions where customers film themselves with their documents, and AI verifies everything offline. This works for straightforward cases and frees up agents for complex verifications.

Video KYC is improving, but unevenly. If you’re a bank and your completion rate is below 70%, you’ve got work to do. If you’re a customer and your first attempt fails, try a different bank. The variation in quality is that significant.