1 month ago
December 14, 2020
Farfetch’s Alex Katsambas Credits Research With Localization Success
When an in-person visit to Fifth Avenue or Orchard Road is off the table, luxury fashion platform Farfetch brings clothing, shoes, and accessories directly to customers in more than 190 countries.
At SlatorCon Remote December 2020, Farfetch’s Head of Linguistic Services, Alex Katsambas, discussed the company’s localization strategy — which goes a long way in establishing Farfetch as an authority in high-end fashion, and as a credible marketplace for designer goods.
Founded in 2008, Farfetch sells products from about 1,300 brands and boutiques based in more than 50 countries. To cater to its international clientele, Farfetch’s localization strategy combines a large internal team with on-the-ground assistance, including support from freelancers.
Although most writers are located in London, Katsambas said, “wherever we have Farfetch offices, we also have writers.”
The Farfetch platform is currently localized into 14 languages with plans to add more in 2021. The company expects to push out 2.5bn words in 2020; about double the output just 18 months ago.
When it comes to deciding which locales to approach next, data and controlled studies lead the way. The preparation often entails comparing pre- and post-launch performances to more stable “control” markets.
Katsambas said this research paid off when Farfetch entered several relatively small markets — Denmark, the Netherlands, and Sweden — about a year and a half ago.
Reach over 13,000 newsletter subscribers and 120,000 high quality monthly pageviews by leveraging Slator's unique audience of senior decision makers.
“That was a really good exercise for us to actually show that it’s worth investing into those markets,” he explained, refuting the common assumption that countries with high rates of English comprehension neither need nor want localized content.
Farfetch’s integration of machine translation (MT) is still in its early stages. “For us, at the moment, we go with an open mind to figure out what works for what,” Katsambas said.
In his team’s experience, MT can be helpful with technical details in product descriptions, “as long as it has enough data and terminology to fit in so it doesn’t translate tanks as actual military tanks rather than tank tops — we’ve seen that before as well.”