When Sanjay Parthasarathy shifted base to India in 2009, after having spent nearly two decades in Microsoft, it dawned upon him that “everything had changed” about the country. He hunted for places to buy things for his kids and family, and realised by and by that search engines didn’t work well outside the US and the UK. “Google’s page rank system threw up results from the US and the UK only,” he says.
For some time, he toyed with the idea of starting a local search engine tuned to what was happening in India. After all, starting up was second nature to the man who “retired” (“quit sounds very wrong, I just decided to move on,” he says) as the corporate vice president of Microsoft’s Startup Business Accelerator (a division he created at the tech behemoth for building startups). “A lot of interesting work was going on at Microsoft, but a lot of it got lost in engineering. We wanted to see if there was a way to build new businesses out of these ideas,” he says.
But, as he found out later, in India, starting a local search engine wasn’t a great idea because the market wasn’t conducive. “The business model wasn’t there, and Google had all the oxygen. Ninety-five percent of search in India was Google,” says the 49-year-old Parthasarathy, an MIT alumnus.
He spent the next couple of years talking to people and travelling to conferences around the world “with no clear idea in the head”. “Now there is a great deal of energy and enthusiasm about startups. At that time, it was more about the economy,” he says. But he sat through conferences quietly and “just listened” to the CEOs and CTOs of the world. “That’s how the idea [of Indix] started… just by hearing 200 people,” he says.
When Parthasarathy incorporated Indix in December 2010, the idea was to start a search engine for numbers (prices). “Prices were interesting in a lot of different ways. People would pay to get knowledge of prices. Just like Bloomberg offers prices of financial instruments, our original concept was to create a Bloomberg for non-financial instruments,” he says. But, very soon, he realised that prices were just one attribute of products.
“Products have an incredible number of attributes: Colour, size, texture, manufacturer, seller, shipping codes, tags, reviews, etc. There is such a rich environment around a product and so much data that if we can get it, look at it, analyse it, and help people make decisions based on it, then that’s a business. That’s Indix,” says Parthasarathy, co-founder and CEO of the company, which is often dubbed the “Google of products”. “But what Google doesn’t do is organise it in such a way that you can analyse it, visualise it and take decisions based on it,” he says.
For instance, if an apparel retailer wants to take stock of a certain product, say, a black cashmere shawl, in its inventory, it can use Indix to find out a variety of things: The inventory size or the number of such shawls it stocks, stores with excess inventory, areas/cities where the product sells most/least, the product’s price in a competitor’s store, the kind of people buying it, the things people are saying about it, etc.
To provide the retailer such detailed information, Indix uses crawlers (a search engine that browses the web systematically for the purpose of indexing) and smart algorithmic filters to mine public websites for product information, which it then classifies into ‘catalogue data’ (static details like category, description, tags, etc) and ‘offers data’ (dynamic details like prices, promotions, availability, etc). The company then structures that data, analyses it in real time, and gets visualisation engines to throw up charts and insights for product managers—its chief target group. Depending on the results, the retailer can take real-time business decisions in terms of price changes, promotional offers, inventory management and more.
That is serious technology in one system. And to ensure its smooth functioning, Indix has minimised human interactions with the use of deep learning or machine learning—a concept where machines interpret data and provide analytics. “People are here to just code. You don’t need them for anything else. If you had them anywhere else, it doesn’t scale. Because humans can process data for hundreds and thousands of products. Machines can process for millions,” says Parthasarathy.