The secret, IME, is to test ideas with users before building the MVP.

It's a mixture of long-term thinking (3-5 years at least; as far ahead as you dare) and short-term planning.

For one product, we did assumption personas, tested them with real users, then returned to them with a back-of-the-envelope concept test (scrappy mocked-up advert listing the core attributes) which sent the team back to the drawing board before they'd done any "serious" design (let alone building anything).

They'd understood the problem well enough, but fundamentally misunderstood the needs in a solution.

Throughout development, quick-and-scrappy tests (short polls, Maze tests, interactive design protypes) were used alongside deep research projects (depth interviews of an hour or more - first to determine the problems and then to get feedback on a high fidelity prototype). Nothing at all was shipped until we were getting consistently positive, enthusiastic feedback.

This was an ancient, prestigious company: the thought of chucking out a mediocre half-finished product was absolutely unthinkable.

When they shipped it, the extensive research that had gone into building it was listed as a selling point.