Research, Prototype, Implement
Early on in my own agency I decided we needed a new service line. Spent a month building it, hired someone to run it, told all our clients. Three months later we quietly buried it and I never mentioned it again.
The problem wasn't the idea. The problem was going from idea to full rollout with almost nothing in between.
The methodology I ended up with after years of doing this wrong is deliberately slow at the front end. Three phases, roughly a month each. The first month you research the thing properly — not whether you like the idea, but whether there's evidence it will work. The second month you build the smallest possible version and test it on one client or one team. The third month, if it worked, you roll it out.
The thing that surprised me was that people stopped resisting change once you did it this way. Asking someone to run a small experiment is different from asking them to restructure how they work. The psychology is completely different.
Twenty tasks a year, five per quarter, one per person. Every single one through the same three phases.