The Midnight Epiphany That Changed From An Over-Hyped Failure To A $100 Million Acquisition

“I wonder what they are searching for? Why are they using our site instead of Google or Yahoo Images?” With these two questions, a new direction emerged. did sound promising. Big-name venture capitalists like Bay Partners and Leapfrog Ventures had invested $19.5 million in the startup and cofounder Burak Gokturk was gifted with the technical know-how: he was a Stanford PhD and an image recognition expert with more than a dozen patents to his name. The facial recognition startup was barely out of alpha testing in 2005 when the acquisition hype began. "The rumors about a Google acquistion (sic) were neither confimed (sic) nor denied by anyone in the know…but there sure is a lot of buzz around this company right now," wrote TechCrunch's Michael Arrington in a blog post following Riya's alcohol-fueled launch party held in his backyard.

Riya promised to automate the process of image search by dispensing with the need for metadata and actually recognizing the faces and objects in a photograph. Computers were already pretty good at tracking down a string of text in 2005, but ask one to look at a photograph and tell you what it sees, and it would look at you with a dumb expression on its face. Google attempted to solve its own image search problem by turning it into a game: Players would create text to describe the figures in a photograph and Google’s search engine would sift through these words to project the images on your screen. The founders of Riya (which would subsequently be rebranded thought there would be a great demand from people wanting to tag friends and others in pictures, which would make it possible to organize these into photo libraries. The company could then amass a database of user-generated images that weren’t covered by copyright protection and eventually sell ads against them. Interface

They were wrong.

The system was overly complicated, a five-step process with two branches. And the business plan itself was built on too many assumptions--“and this will happen, and then when that happens, and after that happens, then we’ll make money and it’s really oblique,” remarked cofounder and CEO Munjal Shah in a 2009 talk. (After several requests for an interview, Shah declined to comment for this article.) Riya needed a new strategy just a year after launch.

On April 25th, 2006, just after midnight, Shah was looking at Riya’s stats when he noticed that for every person uploading personal pictures and tagging them, 20 others were using Riya for search. He immediately emailed one of his cofounders, Gokturk, who replied within five minutes. “I wonder what they are searching for?” Gokturk wrote back. “Why are they using our site instead of Google or Yahoo Images?”

It didn’t take long to unravel the mystery. Riya wasn’t a replacement for Flickr or Y! Photos; it had become a search engine for images on the Internet. By their actions, people were indicating that they wanted a search engine for pictures that was smarter than Google’s image search.

Riya quickly rebranded itself as and attacked a new vertical by turning itself into a visual shopping engine for shoes, handbags, watches, and jewelry. This second iteration of the Riya’s technology allowed users to find an image, say of a strappy red shoe, and request to do a “Likeness search” to find similar items. Users could find variations of products in different colors, shop for clothing similar to what celebrities were wearing, and upload images of their favorite items, then scour the web for similar items.

With this new spin on Riya, Shah and his team were able to raise an additional $50 million from investors. The company’s annual revenue grew to $20 million generated through a simple, direct process: customers browsed for products, clicked, were diverted to retail sites like and and pocketed the 5- to 15-percent commission. By 2009, was selling $100 million in products through leads to merchants.

In 2010, Google showed how much it either liked was threatened by a competitor--by acqui-hiring the team that built it for $100 million and eventually shutting down the service. Needless to say, Shah and Gokturk would have never gotten to that point if they hadn’t managed to pull off a pretty elegant pivot.

[Image: Flickr user Tara Hunt]

Add New Comment


  • Chetan P.

    I heard Munjal speak at an MIT-Stannford event about face recognition and his take on that.  I guess the most interesting view he presented was that he did not believe in the live, real-time recognition of the face recognition technology.