An A/B test of the new designs initially increased registrations by 10-15%, but completed resume purchases declined. We realized that while the UX Researcher provided valuable user insights, these didn’t directly shape business strategy. It was hard to separate research from execution and to tell if the problem was flawed data or how the team used it during development. Positive feedback in controlled studies showed user interest under certain conditions but didn’t ensure commercial success after launch.

Instead of scrapping the project, I recommended slowly introducing features as optimizations to test each one against the baseline. Many features succeeded and improved performance. Using these results, we revised the site for international users, boosting performance by 35% there.