By the time I joined the project, the names for Meta's first generative AI ad creative tools — 'Image Outcropping' and 'Text Variation' — had already been finalized and publicly previewed through the AI Sandbox. Reviewing the experience, I found the names technical, hard to localize, and inconsistent with a high-quality customer experience.
I built an evidence-based recommendation using AI Sandbox customer research, localization assessments across more than a dozen major languages, Meta's content design principles, and my own analysis of clarity and consistency. I presented the findings to director-level leadership and kept advocating for the change despite timeline pressure and a competitor already using similar terminology.
A name that's already been publicly previewed isn't a name that's already right. Reversing a public decision this late meant the bar for evidence was higher, not lower — the recommendation had to hold up against real localization data and research, not just a preference for plainer language.