How AI and ML makes language translation more efficient for non-English speakers

Imagine visiting a website, clicking on one link after the other, and getting nowhere – the links are either dead or they keep sending you back to a language that you don’t read or speak. The frustration would ratchet up immediately.

Now consider that this is a daily experience for millions of non-English speakers in the world.

“The native market should provide the same great customer experience as in all other markets. And that’s just not true for most companies right now,” said Spence Green, CEO of enterprise translation software and services company Lilt.

While most companies pride themselves on being global and meeting customers “where they are,” their websites, customer support and social media channels often indicate otherwise.

Of the thousands of spoken languages in the world, only about 100 are on the internet. The biggest companies support roughly 70 languages, but the average company caters to fewer than 10, according to Lilt’s research.

Add to this the fact that by 2030, there will be more than 4.7 billion middle class and affluent consumers in non-English-speaking regions, as estimated by Brookings Institute.

The power of diversity

Language is central to the customer experience, and it’s a critical component of global commerce. By not making themselves linguistically diverse, businesses lose existing and potential customers, revenues, and brand loyalty, Green said.

“The question to ask is, ‘Why isn’t every business doing this right now?’ and the reason is the same as it is for individuals: We kind of don’t see the whole other world out there,” Green said.

But, he added, “It should be intuitive that language is part of a personalized customer experience.”

Lilt is staking its claim in the $23.8 billion global language services and technology industry. The company, which today announced a $55 million Series C funding round, offers translation technology and services through a human-in-the-loop AI approach. This branch of AI creates machine learning (ML) models from both human and machine intelligence.

The company is providing scalable enterprise translations with a system that Green described as adaptive, self-training, interactive and predictive. This is critical, he said, as the industry transitions from high labor intensity to high automation.

He pointed out that the average consumer uses 10 channels to communicate with businesses, and that 80% of customers now consider their experience with a company to be as important as their products – not the other way around. Still, a study from online payment portal Stripe found that 74% of European e-commerce websites had not translated their checkout pages into local languages. So visitors couldn’t even understand the information in front of them. Such failures on the checkout page account for 9 out of 10 lost sales in Europe.

Test it yourself

Just test it yourself by going to any company website, changing the language, and clicking around on links. As Green noted, you will invariably be kicked back to English.

“That’s the equivalent of having broken links and broken images on your website, which no company would ever tolerate,” he said.

And there’s no reason that it should be happening, he said. “It’s just fixable, it can be fixed.”

Companies simply have to make it a strategic priority. If they do, they will see measurable returns. “Incorporating this into your customer experience and your digital presence is just a totally obvious thing to do,” Green said. “It leads to growth. Every business can be doing this right now.”

Green’s vision is automatic translation capabilities – just a click of a button, wherever or however a person is writing. That button would be integrated directly into business systems. That shouldn’t be far off, as he underscored the fact that across five years of R&D, Lilt has accelerated the number of words it machine generates from 50% to 80%.

“It’s getting measurably more efficient,” Green said.

The company has worked with Intel, ASICS, Emerson, UIPath, and Canva, with local governments such as Lewiston, Maine, and with the U.S. government. For example, Lilt is engaged in Ukraine in defense and intelligence applications. As Green noted, the feds haven’t invested in Slavic language capabilities since the end of the Cold War, and the situation is too pressing to train human talent, which can take months.

The new funding round was led by Four Rivers, joined by new investors Sorenson Capital, CLEAR Ventures and Wipro Ventures, and with participation from existing investors Sequoia Capital, Intel Capital, Redpoint Ventures and XSeed Capital. It brings the company’s total raised to $92.5 million.

The company will use the new funding to expand its team and international footprint, while also further bolstering its R&D efforts.

“We are taking another significant step towards achieving our mission of making the world’s information accessible to everyone,” Green said, “regardless of where they were born or which language they speak.”