MQM was the most robust answer we had for human evaluation. Automated evaluation is a different problem, and the seams are starting to show.

When AV runs outside your TMS, terminology drifts and consistency breaks down. Phrase Studio closes that gap by running subtitling, dubbing, and voice-over inside the same workflow your text content already uses.

Most announcements work like this. A company describes what it intends to build, frames it as a vision, and asks you to believe the future they're describing. This post is different (and longer than most!).

The instinct to build is strongest in engineering-led companies. But maintaining a localization platform over time is a different problem than building a proof of concept -- and most organizations underestimate the difference.

The companies capturing real value from AI aren't the ones with the biggest budgets or the most sophisticated tooling. They're the ones that understood early that the technology was never going to be the hardest part.

When you break global revenue down by individual market, patterns emerge that should be driving strategic decisions. Most companies don't have the reporting infrastructure to see them.

Shifting left means bringing localization expertise into product and go-to-market decisions at the planning stage, not at launch. Here's what that looks like in practice.

International growth fails when it is treated as a bolt-on. Kevin O'Donnell explains why SaaS companies need global-first operating models, dedicated accountability, market-level dashboards, and localization infrastructure built for scale.

For many localization teams, video still sits outside the system. Text moves through mature workflows, while audio and video rely on separate vendors, tools, and review processes. The next step is not simply localizing more video, but integrating it into the operating model for global content.