Ranjan Kumar and the 95% you never say out loud
He left fintech, went quiet for four years, and came back with machines that read the face, the eyes, and the voice. The story of a founder who bet on the part of us that surveys always miss.
Ask someone why they bought a thing and they will give you a reason. Clean, confident, and usually wrong. The reason came later. The decision came first, somewhere quieter, in the half-second before words could reach it. Ranjan Kumar built a company on that gap between what people do and what they can explain.
The number he keeps returning to is 95. As in, 95% of our experience runs below the level we can talk about, and only 5% is what we manage to put into sentences. For a century, the market research industry had been interviewing that 5% and calling it the truth. Surveys, focus groups, panels - all of it built to capture the part of a person that has already tidied itself up for company.
Kumar wanted the other 95%.
It started as curiosity, not a company
The seed was planted at IIT Kharagpur, where Kumar earned his B.Tech between 2004 and 2008. Somewhere in there he got interested in cognitive science and emotion - not as a business, just as a question worth chasing. How does the brain actually decide? What happens in the moments before we know we have chosen? It was a college project. Most college projects end when the semester does. This one waited.
His early career pointed somewhere else entirely. He moved into business roles and eventually headed business at Citrus Payment Solutions, a fintech company that was acquired by PayU. The exit gave him something founders rarely get on purpose - room to stop and ask what he actually wanted to build next. He has described himself as a bit of an accidental entrepreneur, which is a generous way of saying the interesting path found him rather than the other way around.
In 2016, that old curiosity finally became a company. He co-founded Entropik in Bengaluru with the idea that emotion could be measured - not guessed at, not inferred from a questionnaire, but read directly from the signals people give off without meaning to. A glance that lingers. A micro-expression that flickers and vanishes. A shift in the voice that the speaker never notices.
The technologyTeaching machines to watch, not ask
Entropik's answer had three parts, and the team spent close to four years building them before the market was ready to care. Facial coding to read expression. Eye tracking to see where attention actually goes. Voice AI to catch the tonality underneath the words. Together they form what the company calls multi-modal Emotion AI, and they earned 17 approved patents along the way.
The logic behind it is almost stubbornly simple. Brands spend roughly $120 billion a year trying to understand consumer behaviour, and most of that money still buys the polished 5%. If you could measure the 95%, you would not just get faster answers - you would get truer ones. That is the bet Entropik has been running for a decade.
Everyone thinks they get the customer
One statistic Kumar likes to hold up is a small act of demolition. Around 80% of executives believe they understand their customers' emotions. Only 15% of customers agree. That 65-point canyon is not a rounding error. It is the whole problem, sitting in plain sight, and it is exactly the space Entropik set out to fill.
From a thesis to 120 countries
What began as a research bet turned into a working business. Entropik grew to serve large enterprises - names like P&G, Nestle, ICICI and JP Morgan - and pushed its platforms across 120 countries. The products came into focus too: Decode for consumer research, Qatalyst for user research, both built to cut the time between a question and a usable insight by roughly six times. The reach is deliberately global, with support for dozens of languages in transcription and well over a hundred in translation.
The money followed the traction. Entropik has raised around $35 million in total, including a $25 million round in early 2023, from investors including Bharat Innovation Fund, Bessemer Venture Partners and SIG. Behind the funding sits a team of about 250 people and the unglamorous machinery of a SaaS company that has to keep earning its renewals.
Trust as a product feature
Reading emotion at scale is the kind of capability that makes people uneasy, and Kumar has been open that he treats trust as something to be engineered rather than assumed. He talks about transparency, ethical practice, and clear communication as the things that actually decide whether people adopt an AI product or quietly walk away. In his framing, trust is not a marketing line. It is the cornerstone that user adoption and retention are built on.
Three companies, one obsession
Kumar now describes himself as a three-time founder, and his latest chapter is DecisionX AI, based in Bengaluru. The details are still forming, but the pattern is easy to read. He keeps returning to the same layer everyone else skips over - the part of human behaviour that resists being asked and has to be observed. Fintech was a detour. Emotion was the point.
- 2004 – 2008B.Tech at IIT Kharagpur; a college project on emotion and cognitive science plants the seed.
- 2008 – 2015Business roles, including heading business at Citrus Payment Solutions until its acquisition by PayU.
- 2016Co-founds Entropik in Bengaluru to measure emotion instead of surveying it.
- 2023Entropik raises $25M and expands its Decode and Qatalyst platforms.
- 2026Builds DecisionX AI as a self-described three-time founder.
There is something quietly instructive in how long he was willing to wait. The idea sat dormant through a whole first career. The technology took four patient years before the world caught up. Kumar did not invent the demand for reading the subconscious - marketers have wanted that forever. He just kept building toward it until the tools finally existed to try.
The 95% is still mostly unmeasured. That, more than any single product, seems to be the thing that keeps pulling him back to the desk.