Ranjan Kumar, Founder and CEO, Entropik; Image: Selva Prakash for Forbes India; Imaging: Kapil KashyapTo a layman, consumer research would either mean surveys, focus group discussions or interviews. This requires responses to be either filled in manually or analysed from a group discussion or interview. However, with the onset of artificial intelligence (AI), Bengaluru-based startup Entropik is reinventing these traditional quantitative and qualitative research methods.
Founded in 2016, the startup is primarily looking at AI’s application in the consumer research domain using three proprietary technologies—facial coding, eye tracking and voice AI. With large scale clients ranging from P&G and Nestle to ICICI and JP Morgan, Entropik works across sectors such as consumer packaged goods (CPG), health care, media and entertainment, and finance. With a team of close to 250 people, they work with clients across US, Europe, Southeast Asia, India and the Middle East, with 70 percent of its revenue coming from the US and European markets.
Ranjan Kumar headed business at Citrus Payment Solutions, till it was acquired by PayU. That's when he went back to the drawing board and found a clear problem statement: Brands spend about $120 billion globally on understanding consumer behaviour. “Market research is a 100-year-old industry, where things have been done in a very traditional way. We saw this is a large-scale industry, has scope for disruption with new technologies and the ability to scale,” reckons Kumar. With this premise in mind, he launched Entropik.
With surveys, there is a large sample of about 1,000 people, who give a directional sense of the best choices; this is quantitative data. Focus group discussions and interviews help understand the 'why'; this is qualitative data. Both of which could be time consuming, and not too accurate. The team spent almost four years in building technologies to solve this problem. Currently, the company claims to have 17 approved patents.Also watch: Why you should worry about Generative AI
Facial coding analyses the expressions of a person using a webcam. This technology analyses about 58 facial expressions and muscle movements, on a second-by-second basis. The second one is, eye tracking. This analyses where the user is looking, like a heat map on the screen. For instance, when a user is looking at an advertisement, this technology can analyse, what part of the screen has the user's attention. “In this way, the brand finds out what has the user's attention in an advertisement—the brand ambassador, the product or something else. They find out if they are spending money in the right place,” explains Kumar. Putting these two together is Entropik’s first product, Affect Lab, for quantitative analysis.
Voice AI, its third technology, analyses voice transcription, translation and tonality. This means, it is not just about what the user says, but how they say it. Based on the analysis, a confidence score is given: How confident was a consumer in making the claim they made. This is the technology used for its second product, Decode, for qualitative analysis.
Typically, for a consumer survey, an advertisement is shown to a set of consumers who are then asked questions about it. “We are changing that to, no questions asked. As you watch an advertisement, our technology tracks your facial expressions and eye movements. This helps provide deeper real-time actionable insights to brands,” he says. With such detailed insights, brands can expect a much higher return-on-investments as well.
With Affect Lab, a brand manager can use and pick the kind of target audience they want, and the sample size. The platform has an online panel of 60 million respondents across 120 countries, who have consented to participate in these market research tests. “Once the test is made live, the link goes to the respondents, they need to turn on their cameras, and watch whatever is being tested, for instance advertisements or new packaged designs. Once the session is over, the data is available immediately for the brands to look at,” explains Kumar. The advantage here is 4x faster insights and high-quality insights, as compared to traditional research.
With Decode, all forms of video and audio consumer interviews (qualitative data) can be added to the platform and it automatically transcribes and translates these to text. Decode is an AI-powered Online, Scalable, DIY Qualitative Research Platform. Their system has 58 language transcriptions, and 135 language translations available. With the onset of ChatGPT, “we have added a layer of ChatGPT LLM model on this, which automatically summarises these conversations, making it easier for the brand manager.”
So far, the startup has raised $35 million, including a round of funding in February 2023 of $25 million. Some of their investors include, Bharat Innovation Fund, Bessemer Venture Partners, SIG Venture Capital, Trifecta Capital and Alteria Capital. “Entropik was using unique technology—a combination of AI, brainwave mapping, and facial coding—that promised to cause a disruptive shift in the way brands can better understand customer responses to their physical and digital touchpoints,” says Ashwin Raguraman, co-founder and partner, Bharat Innovation Fund. The company follows a SaaS-based revenue model and claims to have an ARR of about $8 million.Also read: Unravelling the ethics of large language models: Navigating concerns in ChatGPT and beyond
While the company has a large repository of emotion data, it is working on how this collective intelligence can be used to make the existing technology predictive. It is also working towards product consolidation, as boundaries between quantitative and qualitative research blur.
With the sudden interest in AI, Kumar is getting a lot of interest from investors, even though he doesn’t need funding. “I wish I had delayed the last round by a little bit,” he jokes. Now deep tech is truly getting appreciated for its value, and not just as a revenue-driver. He adds, “Pretty much everyone in [Silicon] Valley is rallying around AI as the new hope out of recession.”