Mirroring is the beginning
As a native Arabic speaker, I notice quickly when an interface has been translated but not reconsidered. The text may be Arabic and the layout may flow right to left, yet the experience still carries assumptions from another language and market.
RTL support reaches into navigation, icons, animation direction, charts, mixed-language strings, numbers, forms, truncation, and the tools teams use to build and test. A mirrored screenshot can look correct while the product remains difficult to use.
Language reveals the system
Localization exposes whether a design system describes intent or merely stores components. Does an arrow mean next, forward, submit, or move east? Should a timeline reverse? Which identifiers remain left to right inside Arabic text? Those questions cannot be answered reliably by flipping coordinates at the end.
The same is true organizationally. If research, content, engineering, and quality practices only work in the source market, internationalization becomes a recurring exception instead of a product capability.
Design with the market, not for it
The most useful shift is to include local language and cultural context early enough to influence the product. Research should test the underlying proposition, not only the translated interface. Local teams and customers should be participants in the decision, not validators at the end.
RTL work is a distinctive example, but the lesson is broader: products become more resilient when teams treat their default assumptions as hypotheses rather than universal truths.