AddStructure Acquired by BazaarVoice to Boost Machine Learning & NLP Capabilities

Startup’s machine learning and natural language processing capabilities analyze and distill the most relevant consumer reviews
AUSTIN, Texas, February 23, 2018 – Bazaarvoice, Inc., a leading provider of consumer-generated content (CGC), advertising, and personalization solutions, today announced it has acquired AddStructure, a provider of state-of-the-art search and discovery applications for e-commerce companies. The acquisition will strengthen Bazaarvoice’s core consumer-generated content offerings of online ratings and reviews and bolster shopper profiles across the Bazaarvoice Network.

Consumers rely on ratings and reviews when searching for products and deciding what to buy, but when considering products with high volumes of review content, finding the most helpful information can be difficult and time-consuming. AddStructure’s platform uses machine learning and natural language processing technology to analyze large amounts of consumer-generated content to extract and summarize product reviews based on key themes and sentiment. These solutions make product search and discovery faster, easier and more enjoyable for consumers and boost their confidence to make informed purchases.

The combination exemplifies Bazaarvoice and AddStructure’s shared vision and commitment to leverage the promise of cutting-edge machine learning and natural language processing technology, applied to one of the world’s largest sets of consumer-generated content and Bazaarvoice’s unique shopper data, to deliver next-generation product and feature enhancements across the Bazaarvoice platform.

“As consumer behavior continues to evolve, brands and retailers must keep pace with new shopping trends and technologies to deliver engaging and consumer-friendly shopping experiences,” said Gene Austin, CEO of Bazaarvoice. “AddStructure’s advanced capabilities in natural language processing and machine learning are an incredible addition to our product portfolio and we are excited to partner with the AddStructure team to bring innovative e-commerce solutions to market and deliver tremendous value to our clients.”

“We created AddStructure to make it easy and convenient for consumers to discover products across any shopping channel and on any device,” said Jarrod Wolf, CEO of AddStructure. “We’re excited to join Bazaarvoice because of their extensive experience in the ratings and reviews space and their impressive network of brand and retail websites. Integrating our technology with their network data will help shoppers across a larger number of retail verticals and categories find the most relevant products and make the best shopping choices.”

AddStructure is one of 10 companies that participated in the first-ever Techstars Retail Accelerator in partnership with Target, a program that invests in startups disrupting the retail space. Leading retailers like Target and Best Buy have used AddStructure’s cutting-edge technology to distill the most relevant online reviews and customer questions that help consumers make their purchase decisions.

“AddStructure was in our very first Techstars Retail Accelerator program and watching the team’s ideas come to life has been incredibly rewarding,” said Ryan Broshar, Managing Director of the accelerator program. “We are thrilled that Bazaarvoice also recognized AddStructure’s talent and potential; combining these companies will lead to more exciting innovations that will disrupt the retail industry and consumer shopping experiences.”

AddStructure’s employees will join Bazaarvoice’s product, engineering and client success teams and will operate out of New York City and Chicago, where the company was founded.

About AddStructure
AddStructure provides state-of-the-art search and discovery applications for e-commerce companies. Each day, millions of consumers interact with AddStructure’s products and data. Applications of the technology range from review summarization (distilling thousands of reviews of a product into digestible pros and cons) to a complete conversational commerce system (chatting with an intelligent digital assistant to find and purchase products).

About Bazaarvoice
Bazaarvoice helps brands and retailers find and reach consumers, and win them with the content they trust. Each month in the Bazaarvoice Network, more than one-half billion consumers view and share authentic consumer-generated content (CGC), including ratings and reviews as well as curated visual content, across 5,000 brand and retail websites. This visibility into shopper behavior allows Bazaarvoice to capture unique first-party data and insights that fuel our targeted advertising and personalization solutions.

Founded in 2005, Bazaarvoice is headquartered in Austin, Texas with offices across North America and Europe. For more information, visit www.bazaarvoice.com.

Tenor reaches incredible 2 billion GIF searches every month

David McIntosh’s startup Tenor builds a GIF keyboard — but he actually hopes you’ll spend as little time searching on it as possible.

Instead, Tenor’s aim has been to collapse the amount of time it takes you to find a GIF you like and send it to a friend. Instead of trying to get people to come to the service and browse around on the keyboard or a different website, Tenor’s goal has been to figure out what you are trying to say in some kind of a GIF and get it out the door as quickly as possible. And with that approach, Tenor says its users now search for GIFs on its keyboard more than 400 million times a day and 12 billion times a month.

“It comes down to search, fundamentally we’re a search product, unlike Facebook and Instagram and Twitter and Snap,” Tenor CEO David McIntosh said. “They succeed by grabbing more minutes, our success is getting you the right thing faster. Can we take that 25-second session time and make it 20 seconds, or even 15 or 10. There’s a viral loop in place where every time you make search a little better it’s faster.”

This more or less dovetails with an approach for some companies that are focusing on pitching engagement instead of a raw active user metric. Snap, for example, has stressed to investors that it is getting people to come back to the service more and more and spend more time on it. It’s roughly the same principle in terms of using Tenor, which McIntosh says is more of a search engine than an actual hub or portal. Basically, you want to communicate what you want to tell a friend in as few words as possible — except with something silly from Friends. Tenor works across a number of platforms, but now its sights have shifted abroad.

That might even be more true as Tenor begins to expand internationally, planting people on the ground to figure out what localized versions of the service look like. One of the appeals of GIFs is that it can compress a ton of information (McIntosh refers to it as “emotion”) into a short semi-video object in a messenger screen rather than having to type out a bunch of text. As it expands to more and more countries, Tenor is able to start picking off that low-hanging fruit, as making small tweaks in certain regions can lead to dramatic improvements in engagement and usage, McIntosh said.

“Western content is so heavily exported all over the world that these things have almost become globally recognized objects,” McIntosh said. “Often western content with a local caption will perform better. Sometimes the local content performs better. You gotta have the right set of search data, share data, community uploads, it’s the combination of all of them. It’s kind of like the chicken and egg problem; it’s a slow grind until a spark happens — you’re guessing what’s gonna work. Once the flywheel is spinning really quickly you have so much data.”

It’s also begun running its first partner campaigns internationally as it’s started to expand, with the idea that it can go to potential advertisers and tell them that because people use the keyboard so much they’ll actually share that content. That includes campaigns with companies in even India and Germany. The whole goal is to, again, figure out how to get the right GIF in front of the right person in those couple of slots when they open the app and actually want to share it.

There is, of course, a data component to that problem, as well. But with 12 billion searches every month, Tenor can start slightly tweaking each search to figure out what a person is looking for based on a wider array of parameters — and maybe figure out how to get that Tom Brady strip-sack in the expiring minutes of the Super Bowl this year in front of people more quickly. Two months ago, Tenor says it had 10 billion monthly searches monthly (around 330 million daily).

It might sound a little ridiculous now, but in retrospect there’s been a blossoming ecosystem around both creator tools for GIFs as well as ones for sharing them in messenger products or the web. Gfycat, which targets creators with more robust tools, says it has 130 million monthly active users, while Giphy says it has 300 million daily active users. Either way, it means there is both a lot of competition and a lot of interest in this space — including venture financing.