"Ebay reaches out to advanced small business' technology in what they call 'acqui-hired"

McLean’s AppTek is latest to be ‘aqui-hired’ by eBay

About 18 months ago, eBay faced a linguistics problem.

The e-commerce giant had about 650 million items listed for sale, by 25 million people worldwide.

Interested buyers typically searched for products by typing words into a query. The site would then find items whose descriptions contained those words, written by the seller.

But what if the buyer and seller didn’t read or write the same language? Buyers were limited to items with descriptions they could understand, and sellers could only reach a subset of the global market.
So the company started investing in text translation, aiming to automatically translate descriptions into the buyer’s preferred language. As part of this project, eBay recently agreed to acquire translation software developed by AppTek, a McLean-based technology company.

AppTek is the latest target of an eBay “acqui-hire” — a strategy in which a business buys a company, or its intellectual property, with the intent of hiring its technology development talent. Between 2008 and 2014, eBay has completed more than 35 acquisitions; since 2010, 80 percent of the founders of those companies have remained at eBay, according to eBay. As per the agreement between AppTek and eBay, about eight of the Virginia company’s 20 employees — including its co-founder, Mudar Yaghi — are joining eBay’s translation team.

“These language scientists are few and far between when you scan the globe,” said Wendy Jones, eBay’s vice president of geographic expansion and cross-border trade.

She noted that in this case, the experts who developed the software are just as valuable to eBay — if not more so — than the software itself. “What we’re trying to do is marry the [translation] capability with all [of eBay’s] insights around e-commerce,” Jones said.

For instance, Jones said, eBay is trying to make its translation software more sophisticated — shifting from, “Is this translated correctly?” to, “Have I done this in a way that is compelling enough to the consumer so that they’re going to behave differently?”

The AppTek deal is part of eBay’s “ambitious journey to extend our footprint into markets we traditionally haven’t focused on,” Jones said. Better translation could allow more global online trade, she said — not only among large nations and newer markets such as Brazil, Russia, India and China, but also in parts of Europe such as France, Italy and Spain.

Before deciding to acquire AppTek’s technology, eBay had been developing its own software in-house, called “eBay Machine Translation,” Jones said. The division, which AppTek’s former employees will join, is currently made up of about two dozen developers.

“We started with the [Brazil, Russia, India and China] focus, but the bigger opportunities are in the existing markets where eBay already has strong, healthy business,” Jones said, explaining that people who already use the site regularly will soon find many more options than they did before every time they search — five, six, seven times as much, she added.

In some countries, as much as 70 to 80 percent of demand on eBay can be fulfilled through cross-border trade, Jones said. “It’s a great opportunity to help sellers reach markets they wouldn’t otherwise think about participating in.”

In some countries, as much as 70 to 80 percent of demand on eBay can be fulfilled through cross-border trade, Jones said. “It’s a great opportunity to help sellers reach markets they wouldn’t otherwise think about participating in.”

For AppTek, the sale offers an opportunity to narrow its focus on speech recognition software, co-chief executive Mohammad Shihadah said in an interview. Before the sale, the company’s time, talent and resources were divided between newer technologies and machine translation.

For instance, AppTek hopes to market its technology to TV news sources for closed-captioning, Shihadah said. AppTek’s speech recognition software is currently being used at al-Arabiya, an Arabic-language news network, for subtitling and captioning.

The company is also working on speech recognition for mobile phone users, as well as voice response systems.

Compared to the machine translation software they sold to eBay, he said, “we believe this is equally important in terms of the future.”