Picture a customer in small-town India opening a leading app, ready to borrow a personal loan. She clears the early screening, reaches the video KYC step, and starts turning her head on cue for the liveness check. Her signal wavers, the session stalls, and the app asks her to begin all over again. She is exactly the customer the lender wants, and she’s just one screen away from dropping off.
That customer represents the majority. According to our data, small town customers make up the bulk of video KYC users in India. In states like Odisha, Chhattisgarh, Andhra Pradesh, Tamil Nadu and West Bengal, this share climbs to 87%.
And yet, most video KYC workflows are tuned for a metro professional on office wifi with a minute to spare. When a flow built for that customer meets the one who actually shows up, the gap results in a customer who wanted in and walked away because the process was too frustrating and not built for them.
The opportunity hiding in the drop-off
Onboarding completion is one of the most valuable numbers a lender can move, and small gains compound quickly. In fact, 70% of financial institutions worldwide saw customers slip away last year due to slow onboarding, the highest figure recorded and up from 48% two years earlier. Research shows that 38% of customers step away mid-onboarding when the process asks too much, too soon.
For small-town India, the prize is large. A customer who reaches the video step has already chosen the product and cleared earlier hurdles of customer acquisition. Helping them finish turns an intent that already exists into a funded account, and the financial institution captures value.
Where most video KYC workflows have room to grow
The RBI’s November 28, 2025 KYC Master Direction, which brought payment aggregators into V-CIP-equivalent compliance, and the 2025 amendments added deepfake-related requirements set a clear and sensible bar. It calls for a live official, a liveness check, geo-tagging, an encrypted recording, and a second video officer who reviews the first. Those controls are the reason regulators treat video KYC as equal to a branch visit.
Meeting that bar at scale is genuinely hard. The early versions solved for the metro customer with a strong connection, but the traffic has since shifted to small towns with modest networks, making it important to build workflows in regional languages that encourage financial inclusion.
When a session fails, the easy assumption is that the agent’s connection or staffing was the problem. Our data tells a different story: the customer’s network is the single biggest point of failure. A flow tuned for office wifi reads a wavering signal as an error and punishes it with a restart, when for most of the country that signal is simply how the internet arrives.
As digital finance becomes more accessible, it’s important to protect against fraudsters running a real-time face swap off a stolen ID, who test the surface that live video and face match. A modern flow must answer both these challenges at once so it smooths the path for the genuine customer and raises the price of attack for the fake one.
The balance worth getting right
Security, compliance and customer experience reward each other when the design is sound. The stakes here are high because late in 2025, several large banks tightened their fully-digital account opening and started asking customers to visit branches after the RBI acted on weak KYC in online onboarding.
Audits in the past have found institutions unable to promptly retrieve what the rules require them to hold: the geo-tag, the timestamp, the full trail of who reviewed what. Retrieval is a design requirement. You should be able to pull any session’s evidence on demand. Manual post-call checks across full volume don’t survive scale, because the error rate defeats the purpose.
A flow that holds the compliance bar and stays kind to the customer keeps the digital channel open and makes financial inclusion genuinely possible for the small-town customer. When the journey meets the customer where they’re at and complies with regulation, control and the conversion arrive together.
Six changes that lift completion
Hold the same regulatory bar, ask the genuine customer for less, and raise the effort for the fraudster. Quality checks at the right moments in the onboarding workflow do much of the quiet work, confirming a good frame instantly and asking for a retry when it’s unclear.
The technology enabling video KYC must read the customer’s ID and pre-fill the fields so the customer can confirm the details and move on.
It must catch frame quality early on. An on-device check should confirm sharp focus and good lighting in the first two to three seconds, so any redo must occur on the spot while the customer still has the camera raised to perform the liveness check.
It must survive the signal dip without dropping the session. This is the largest lever for weak coverage. Standard video implementations panic when packet loss climbs; they freeze the frame or cut the call, and the customer starts over. A well-built stack does the opposite. It prioritizes the audio channel so the conversation with the officer continues even when video quality falls, buffers the encrypted recording locally, and completes the upload when the signal recovers. The session stays live, the compliance record stays intact, and the customer never sees a restart screen.
The platform must run the session in regional languages. Prompts, error messages, and the officer script must be customized in Hindi, Telugu, Kannada, Bengali, etc., as it goes a long way in keeping small-town customers confident of using digital platforms for onboarding. The industry is already moving this way, with video KYC now built to run at bandwidths as low as 75 kbps, holding the audio steady even as video degrades, and across regional languages.
Onboarding workflows must confirm each step in plain words. A short message that tells the customer their face is not framed well, or asks them to move to a place with better lighting, keeps the customer oriented and carries them to the finish.
It must layer deepfake defense behind the scenes. Passive liveness that flags AI artifacts, randomized officer questions, and geo-tagging work together to turn real-time face swap into a costly, low-odds attempt. iBeta-tested ISO-aligned liveness gives a financial institution a benchmark it can point to. The genuine customer glides through while synthetic ones meet a wall built to hold.
Also Read: 10 Insurance Myths to Burst This Dussehra
Verification has long asked every customer to prove they belong, and the weight of that caution settled hardest on the honest person with the weakest network. Designing onboarding workflows for such customers while increasing the cost for fraudsters ensures a stronger and more inclusive ecosystem.
Views expressed by: Kedar Parikh, Chief Product & Delivery Officer – India, HyperVerge
