Kevin O’Donnell, founder of Global10x and former VP of International Growth at Dropbox, joined Jason Hemingway on the In Other Words podcast to talk about the build-versus-platform question and why he now tells companies to stop doing what he did at Microsoft.
The instinct to build your own localization infrastructure is strongest in technology-forward companies. The engineering talent is there. The conviction is that an internal team can build something more tailored than any available platform. And at the proof-of-concept stage, that conviction often looks justified.
Kevin O’Donnell has an unusual vantage point on this question. At Microsoft, he led the team that created an in-house localization platform. They revisited whether to keep building it every few months. He now advises companies through Global10x, and his advice has changed.
The build-versus-buy question companies get wrong
Kevin’s test is one question.
“What business are you in? Are you in the business of creating a localization solution, or are you in the business of delivering international experiences for your customers?”
That distinction matters because building a proof of concept is relatively easy for engineering-led organizations. Maintaining production systems over time is something else entirely.
Many companies begin with the assumption that an internal solution will be more tailored to their needs than a commercial platform. Initially, the approach can look highly effective, giving teams speed and a system designed around the specific needs of the business.
But over time, engineering resources increasingly shift toward maintaining infrastructure rather than solving problems unique to the company.
In Kevin’s experience, organizations often end up dedicating significant development resources to rebuilding capabilities localization platforms solved years earlier.
Two companies, two different outcomes
Kevin offers two cases from his advisory work.
The first involved a SaaS company coming out of a funding round with aggressive international expansion targets. The business had no internal localization team and limited engineering bandwidth.
Rather than building infrastructure internally, they connected a platform to their existing content systems and launched across multiple languages within weeks.
The business stayed focused on product and growth while the platform handled multilingual operations.
The second case was more revealing.
A large enterprise had spent years building its own localization stack. On the surface, the system worked. But once Kevin reviewed how the team operated, it became clear that much of their time was spent maintaining workflows, integrations, and tooling that were already standard capabilities elsewhere in the market.
The company eventually adopted a platform while retaining the internal team.
That changed the role of the organization entirely. Instead of maintaining commodity infrastructure, the team focused on work that genuinely differentiated the business, such as custom extensions, market-specific workflows, and challenges unique to the company’s operating model.
“While every company has its own unique use cases, they’re the exception rather than the rule.”
The underestimated value of standardization
There’s a secondary benefit to platforms that Kevin believes is routinely underestimated. They impose standardization.
Without a shared platform, organizations tend to treat every operational challenge as unique. Over time, that creates layers of bespoke tooling that become progressively more difficult and expensive to maintain.
Engineering teams slowly evolve into internal software vendors, fielding requests from across the organization while spending less time on work tied directly to competitive advantage.
In multilingual content operations, Kevin believes most companies are far less unique than they think.
Standardized infrastructure is often more valuable than highly customized systems because it reduces operational complexity and frees teams to focus on the areas where differentiation actually matters.
Where platforms fit
Platforms like Phrase exist to solve exactly this problem. They give companies the infrastructure to manage multilingual content operations at scale, with the AI capabilities, quality management, and integration architecture already built in.
That frees engineering and localization teams to focus on the work that actually requires their rather than maintaining systems for challenges the industry has already solved
The build-versus-buy question deserves rigorous examination. But Kevin’s experience, from building at Microsoft to advising dozens of companies since, consistently points to the same conclusion.
Most companies gain more by investing their expertise in the business they are actually trying to grow rather than rebuilding systems the industry has already standardized.
Watch the full conversation with Kevin O’Donnell on In Other Words.
