Adding a personal touch to ads

Student startup uses artificial intelligence to personalize in-store ads

Computer science concentrator Shriank Kanaparti, A.B. ’21, founded the startup NXTClick, which uses computer vision and machine learning to offer a personalization service for offline ads in retail stores. (Photo courtesy of Shriank Kanaparti)

While waiting in the reception area of an eyeglass store, Shriank Kanaparti’s gaze drifted to a 55-inch screen on the wall a few feet away. The giant screen kept showing the same ad for a pair of women’s eyeglasses.

Kanaparti quickly become tired of seeing the same image over and over, and he began to think of the engagement that potential customers would feel if they saw ads that were relevant to them.

“We are living in a highly digital world, and everything online is becoming personalized. We’ve already seen highly personalized ads when we browse the web,” said Kanaparti, A.B. ’21, a computer science concentrator at the Harvard John A. Paulson School of Engineering and Applied Sciences. “It seemed to me that it is only a matter of time before this idea of personalization proliferates offline.”

He recently launched NXTClick to make his vision a reality. The startup uses computer vision and machine learning to offer a personalization service for offline ads in retail stores.

To get started, retailers plug a camera and NXTClick compute stick into a digital screen in their store. Using images from the camera, the cloud-based software identifies people in front of the screen, picks out certain attributes of those individuals, and then displays the most relevant ad.

Retailers tag each ad with the age range, gender, and emotion of individuals it is designed to target before uploading ads to the NXTClick system; a matching algorithm finds the ad that’s the best fit.

“There were a lot of challenges that we didn’t realize we would face,” Kanaparti said. “For instance, identifying someone’s emotions accurately within a two- to three-second exposure with a camera might not be completely accurate, so we are still working on that.”

Another challenge arises when there are multiple people in the frame. Kanaparti designed the algorithms to pick out as many attributes as they can, and then find the ad that satisfies the highest number of constraints for the most individuals.

Kanaparti and his team are still working out those issues as they conduct the careful debugging that must be done before they can roll out an initial product. He enrolled in Startup R&D (ES 95r), taught by Paul Bottino, Executive Director of Innovation Education at SEAS, and has relied on the feedback and support of the teaching staff and a cadre of like-minded and ambitious peers.

“I’ve learned that a lot of entrepreneurship is just about starting,” he said. “An idea is literally nothing if you don’t start; you won’t realize how easy or difficult it could have been, or where you could have gone if you tried. Entrepreneurship really comes down to having the courage to persist between the good times and the bad times.”

The NXTClick team is currently testing the market to see whether the timing is right for their product. They recently visited retailers in Boston’s busy Newbury Street shopping district, and are looking to incorporate more feedback into their design and development plans.

Ultimately, Kanaparti sees NXTClick as just the beginning—he envisions a marketing ecosystem that would allow retailers to have the same kind of robust options they have when purchasing personalized ads online.

“Currently, a lot of the offline ad space is very fragmented. People usually have to go to multiple brokers to get their ads in the places they want,” he said. “We want to reduce that friction and make the offline ad experience as smooth as the online ad experience.”

And as technology advances, he expects it will become easier to implement NXTClick at an even larger scale. Improvements in processing power would enable faster ad delivery, while hardware enhancements would make trait identification more accurate.

For Kanaparti, who has loved building with new technology since his youth, working at the cutting edge of such a high-tech field is exciting. But the best part of NXTClick is the opportunity to improve people’s lives, he said.

“A lot of the knowledge we have today about the products and services that are available to us comes from ads,” he said. “If you could see more ads that are relevant to you, that feature products and services that could really add value, that could help people live happier and more efficient lives.”

Topics: AI / Machine Learning, Entrepreneurship

Scientist Profiles

Press Contact

Adam Zewe | 617-496-5878 |