Prior to founding Localization Technology Partners, I built out and led the localization teams at leading startups, including Notion, Lyft, Medium and others. I typically would be brought in to set up translation tech and service providers and build out the localization ops team, with a focus on tooling and process automation so a lean team can support a variety of languages. Following is a brief overview of how these projects unfolded. You can also find my Linked In profile here.
I joined Notion in 2021 to build out the localization systems and team. At that time, the product had been partially localized in Japanese and Korean, where the company was already seeing a lot of organic demand. There was no in house localization team, and most of the translation work was being done manually.
The first thing we did was to source a TMS (translation management system) that integrated with our CMS (content management system), and that had good workflow automation. This enabled us to localize all customer touchpoints, and keep translations in sync with updates to English content. We also expanded the number of non-English languages from 2 to 12. The company is now in a state where translations happen by default across most touchpoints. About 70% of Notion users are now from outside the US, and also represent a majority of revenue. There are now three full time people working on localization, plus a small army of vendors and freelancers who do the bulk of the translation work. The team also works closely with international sales offices in Europe and Asia to support their efforts, and to get feedback.
Takeaways: Notion did almost everything right when it came to preparing for and supporting international expansion. With a lean team (3 people) and a modest translation budget, we were able to unlock 100x as much in revenue and establish Notion as a global company. The product itself allows users to create content in any language they want, and had a design motif that worked across languages. It also helped that the founder, Ivan Zhao, is an immigrant and understood the need to support other languages.
I joined Lyft in 2017, initially to help them expand their rideshare offering to Toronto and other English speaking Canadian cities. This involved a lot of work on currency support and (my favorite) value added tax (VAT). Following the Canada launch, I worked on localization. Lyft had a large amount of tech debt due to operating in US English only for many years. This involved a cross functional effort to refactor the many systems that powered the mobile app. This took 18 months, and cost well over a million dollars. Be sure to read this tutorial on future proofing your code base. If Lyft had followed guidelines like this earlier in their development, they would have been able to eliminate much of this delay, and would have saved over a million dollars in engineering time.
We launched Lyft in six languages. Our initial focus was on the US market, where a large percentage of drivers spoke English as a second language. We were also targeting the US Latino market to catch up with Uber, which had been operating in Spanish for some time. We were doing this with an eye toward international markets, although Lyft had difficulty exporting its rideshare model to other regions due to regulatory barriers. In retrospect, they would have done well to position the app as a booking agent for licensed cabs (Uber has done this with success), something I pushed for.
Takeaway: Lyft had accrued a lot of tech debt over the years, which significantly slowed down the project, and also incurred significant engineering costs to refactor the code base to support multilingual operations. The company’s core product was also hindered by regulatory barriers in most non-US markets that limited its ability to expand internationally. Although we got high marks for the quality of the localized products, localization by itself wasn’t enough to insure success outside the US and a few Canadian markets (an important lesson for companies that are starting their expansion into new markets).
I joined Medium in 2016 when the company was operating only in English, but despite that was seeing strong organic demand in foreign markets, Brazil in particular (a market most companies would do well to invest in). Although the product UI was in English, people could author content in any language they wanted. Medium had also done some interesting product experiments around crowd translating user authored content. Medium was a publishing platform that attracted talented authors, and this was a way to expand their reach.
The codebase had not been written with localization in mind, so there was a substantial effort to refactor it to support multilingual operations (I did much of this work since they had a lean engineering team). We added four interface languages (Brazilian Portuguese, Spanish, French and Italian if I remember correctly). This was particularly well received in Brazil where we had a large user community.
Unfortunately the company hit a wall financially about a year into the project and decided to concentrate on the English speaking market, which was probably the right business decision at the time (that’s life in tech). If timing had been different, the outcome would have been as well. The product itself worked well in other languages, and there were no barriers to people using it anywhere.
Takeaway: Medium did a great job designing a platform that allowed users to write in any language. So even though the product itself wasn’t localized, that didn’t stop writers from authoring content in other language, which drove organic growth in markets like Brazil. On the other hand, their code base had a lot of tech debt that had to be retired, which slowed things down.
Some Common Themes
It’s not only important to have product-market fit in your home market, but your product also needs to be adaptable to new regions. SaaS products like Notion and Medium have few barriers to usage, so localization is an easy way to unblock users and grow your audience. Products that have an on the ground component or that are affected by regulations have lots of challenges when it comes to international operation (localization by itself may not be enough to insure success, although it is a pre-requisite for many markets).
Most small to mid-sized companies can run localization with a lean team (1-3 people) that manages translation vendors and freelancers. The bulk of the department spend will be on translation services (AI is bringing unit costs down, but it still requires people to be in the loop to maintain quality, at least for high visibility content). With good automation, this can largely be outsourced, with a more junior project or program manager to oversee the work. Localization can unlock 10-100x in revenue compared to its spend, making it a powerful growth lever for mobile and web services.
It is important to follow global ready design and coding practices early on. Otherwise tech debt accumulates and it is time consuming and expensive to retire later on (spoiler alert: almost nobody does this).
There is usually a lot of upfront work, with a focus on engineering work (refactoring, tooling, etc). Then once that is out of the way, and with good automation, a lean team can support the program at most small and mid-sized companies.