To Future-Proof a Language Service Provider, Embrace Data, Tech, and Client Goals

Michal Antczak of PayPal at SlatorCon Amsterdam 2019

It is more than a little ironic. Oftentimes, when language service providers (LSPs) talk to potential clients, it can seem like each side is speaking a different language.

Michal Antczak, Head of Localization Technology at PayPal, told the SlatorCon Amsterdam 2019 audience that LSPs tend to focus on information that does not answer the customer’s biggest questions. Instead, they highlight their affordable services, top-quality translations provided by the best translators, cutting-edge technology, and scores of satisfied customers.

According to Antczak, the message LSPs try to send to potential clients is, “If you work with us, you can be sure you’ll get optimal quality, cost, speed, everything.”

A glowing testimonial from a past client, however, does not guarantee success for a new customer. Generic marketing rhetoric promises clients that the LSP can enable clients to enter to enter new markets, but in reality, LSPs can only help their clients enter new markets.

“Business stakeholders know localization is just a tiny fraction of what needs to be done,” Antczak pointed out.

For example, the specific industry in which a client operates can impact the potential profit earned through efforts to localize. Antczak recalled the surprising response to an effort to localize professional services materials into Chinese: The end users wanted to keep the materials in the original English to use as resources in preparation for exams given in English.

This demonstrates that different users have different needs and expectations; so again, what worked for one client may not make sense for another.

“What the business people really want to talk about is this stuff: their metrics, their KPIs, their vision,” Antczak said. “We, the localization people, need to start talking to these people in their language.”

Life for Localization Beyond NMT

To shift the conversation to clients’ goals and priorities, Antczak recommended that LSPs focus on how the LSP will build trust, reduce product-use friction, and increase product-use comprehension.

Slator 2020 Language Industry Market Report

55 pages. Total market size, biz dev and sales insights, TMS & MT review, buyer segment analysis, M&A, Covid impact & outlook.
$480 BUY NOW

Taking this idea to the next level, Antczak said LSPs should move toward enabling their clients to speak to end users in their language and in the appropriate cultural context, rather than LSPs doing so on their clients’ behalf.

This new focus on clients’ needs, combined with a creative use of existing and future technology, could transform an LSP’s work.

In a twist on going hyper-local, Antczak suggested that tailoring localization to specific end user parameters could have a big impact on user experience. Localization professionals could take into account a user’s demographic data, preferred voice and tone, recent or past experience with the company, and current sentiment toward the company, and use translation memory (TM), neural machine translation (NMT) and, in the next two or three years, natural language generation (NLG) to craft highly targeted content.

Michal Antczak

“That would be true innovation,” Antczak said. “That’s where I feel we should be heading as an industry.”

The rest of the localization workflow would be modified accordingly. The quality assurance (QA) stage, for example, would be replaced by a process to interpret user feedback. The LSP would then measure the impact of feedback on KPIs, allowing business stakeholders to confirm that their investment is worthwhile. Based on those results, language experts and translators would continue to review and improve content. 

Typically, Antczak explained, big companies look at analytics in English and, maybe, one or two other locales. In this new model, a business would work on data collection, products, instrumentation, and data management, while the LSP or in-house localization team would build expertise or technical systems based on TM, MT, natural language processing (NLP), and text-mining techniques to understand what users are saying in their own languages.

Antczak sees this model as a way for LSPs and linguists to provide added value in a tech-saturated market —  life for localization beyond NMT.

“The beautiful thing is that if LSPs try to take this route, it’s a multi-year journey,” Antczak said. “This is not something that’s going to be achieved in half a year and then we move on.”