Airbnb: A Lesson in How to Implement Language at the Platform Level

Airbnb: A Lesson in How to Implement Language at the Platform Level

Are you ready to implement language as a feature at the platform level? Do you know how to gain executive approval for the business case to achieve that? Do you even know what I'm talking about?! Read on to find out how Airbnb did it and the questions to ask your team to find out if you're ready to embark on the same journey.

What Did Airbnb Announce Related to Language Accessibility?

As part of its Winter 2021 Release, Airbnb's CEO, Brian Chesky, highlighted the company's Translation Engine as the second of four featured upgrades (out of 50) in his launch video on November 9. In fact, it was so significant to him that he took the time to demo the previous steps that his customers had to go through to translate a listing (including a required click on a translate button) versus the automatic appearance of translated content with the new Translation Engine.

Source: Airbnb.com

A multi-year initiative led by Salvo Giammarresi's globalization team and its language technology and services partner, Translated, the ModernMT-based solution will go a long way to eliminating language barriers between Airbnb travelers, hosts, and employees by granting close to seamless access across 62 languages for property listings and user reviews when it's launched at the end of this year.

But the underlying infrastructure - and the framework around it - actually go much further than what the term "Translation Engine" may conjure up for most people. The hype around the announcement within the language industry has centered around the contract size for the services and development work delivered by Translated. However, contract size is far from the most important implication of what Airbnb can now deliver to its customers - and what was required to achieve it. The far more significant implications revolve around the recognition given to the business value of multilingual content by a high-tech company such as Airbnb through its financial investment in a cross-functional collaboration initiative to greatly expand language accessibility.

CSA Research recommends that buy-side localization teams analyze five essential learnings from Airbnb's accomplishment and reflect on what they mean for their own next round of strategic planning.

Language: As a Platform Feature

CSA Research has written about the shift to language as a feature at the platform level wherein language is designed, delivered, and optimized as a feature of a product and/or service from the beginning. Language accessibility is no longer relegated as a translation task outside the bounds of product or service development and delivery. Rather, it is integrated into content and code workflows that affect almost everyone within the organization at some point. As Chesky stated in his launch video, "Technology made it possible to work from home, but Airbnb now allows you to work from any home." Obviously, language is an integral part of his platform that now makes that possible.

And so is machine translation (MT), as it allows an organization to scale its multilingual content and code enterprise-wide and for the foreseeable future, rather than simply to support a short-term effort to translate backlogged content. MT integration enables the mandating of local language accessibility - and even local functionality accessibility - to all content and code across the company. This is obviously critical to the medium- to long-term success of companies such as Airbnb, Expedia, TripAdvisor, and Bookings.com - many of which have long been machine-translating huge amounts of user-generated content (UGC) and white-label content.

Here are three questions for localization teams to explore around this topic:

• How do we successfully promote language as a platform feature?
• Once accepted, what does our team need to do to be prepared to contribute in terms of leadership, technology, and process optimization?
• Are we set up to successfully manage and optimize the continuous product/content release cycles required for scaling up language as a platform?
  

Business Value: Seamlessly Translated UGC

There's a reason that Expedia's current CEO, for example, has a strong background with media properties such as Tribune Media and Univision. Both Airbnb and Expedia recognize that they are not lodging companies, but rather high-tech (multilingual) content companies. And Chesky certainly understands this very well as he touts Translation Engine availability by the end of this year, which will deliver five million listings in 62 languages and 500 million reviews without customers having to tap on a translate button.

Chesky also recognizes that without the ability to access user-generated or white-label content, his brand doesn't exist. So the multi-year investment that he has afforded Airbnb's globalization team is recognition that this company's product (information about properties) must be accessible anywhere in any language in order to connect the people who generate Airbnb's revenue. This investment also suggests a mindset shift that other companies may yet need to make. They must recognize that, in many cases, their products and services do not function independently of information about them and that the most valuable content and code are often generated by third parties.

Huge volumes of multilingual UGC also open up the floodgates for GPT-3, a language model that uses deep learning to generate human-like text. Just think - instead of having to come up with original content to promote your beautiful, beachfront thatched-roof cottage for rental, you could choose from multiple examples from a repository fed by a machine learning system. Such a system could create an alluring description based on location, number of bedrooms and baths, accessibility, wifi speed, and other checklist criteria. There might be some entries parroting others, but at least you would be done a lot sooner with your writing task by avoiding today's copy/paste/edit cycles.

Here are two questions for localization teams to explore around this topic:

• How much does our brand depend on third-party content and code? Can our products or services exist independent of them?
• User-generated content (UGC) is often created by non-native writers who are not bound by standard grammatical rules, corporate style guides or technical writing requirements. If adaptive neural MT systems can now deliver acceptable quality for the various areas covered under this category, should we now consider implementing it for other content and code types?
  

Collaboration: Required by Content's Horizontal Nature

The Airbnb announcement is quite significant from the perspective of the degree of collaboration required from cross-functional teams. Salvo's team worked with around 100 people company-wide to achieve integration with IT, content marketing, product management, and all of the other Airbnb producers and managers of content affected by the new workflows and technology.

One notable facet of language as a platform is that content functions horizontally with no ongoing C-level representation. Therefore, individual teams usually have little, if any, control over content enterprise-wide. Instead of trying to control who creates content and where it may be created, Airbnb's MT solution integrates with all of the technology, processes, and people inside and outside of the company that produce it. With the emergence of responsive MT, this becomes achievable for other companies as well.

Here are two questions for localization teams to explore around this topic:

• Is there a strategic initiative that we should be leading or playing a more visible role in?
• Do we need a product manager - rather than a project or program manager - on our team to lead collaboration for language as a platform with colleagues in product/service design, usability, engineering, compliance, and customer support?
  

Localization: Essential Enabler for Revenue and Growth

Localization teams can only be successful leading multi-year investment initiatives if they're recognized by executives as essential drivers for their organizations' global growth and revenue. If it's not there yet, how can you evolve your team to this point? You start by recruiting an executive sponsor to promote globalization in venues that you don't have access to. Supplement that by working persistently to move your team into reporting into the optimum part of the organization that will allow you to achieve your goals. Back up your strategy and execution plans for investment with hard data at every opportunity.

In the meantime, investigate these four questions with your team:

• Is localization viewed as an essential enabler enterprise-wide for revenue and growth? If not, what's our plan to get there?
• How do we integrate our international strategy with our overall corporate strategy?
• We may not be targeting the vaunted "next billion customers," but are we preparing enough to capture and keep the next big group of new audiences outside of our domestic borders?
• What will this take in terms of people, process, and technology?
  

LSP Partnerships: An Unrecognized Gold Mine of Innovation

Many enterprises still need to learn how to mine the genuine - and often very rich - source of innovation in the companies they engage for language services and technology. They block themselves from forging new paths with these partners because they neglect sharing their challenges and goals.

What if more companies invested in this type of collaboration to optimize continuous localization, apply robotic process automation (RPA) to manual processes, or implement intelligent machine learning to meet the quality challenge, just to name a few? Expanding this collaborative approach to include paid interns from university departments such as business, economics, data analytics, AI, computer science, life sciences, engineering, product design, and language/translation can go a long way to identifying many more innovation paths than a company may have thought possible.

Here are three questions for localization teams to explore around this topic:

• What is crucial in our DNA that our language partners must display as well? Innovation focus? Technical expertise? Process optimization? Something else?
• Which partners can match our velocity for adaptation - or push as fast?
• What innovation paths are we missing?
  

Several companies have reported anecdotally that their localization teams were able to take advantage of more opportunities than usual for leading innovation or showing other teams how to deal with constant change over the past 20 months. The pandemic forced most organizations to review, tinker with, and even overhaul their business models. The result? Many localization teams still have the opportunity to launch big initiatives with their colleagues, including the implementation of language as a component of their platform. The time is now - don't miss out.

Note: CSA Research has begun work on an Internationalization Maturity Model for software architects and developers. If you have a special interest in this area, please email Rebecca Ray today.

Rebecca Ray