Have a new new design question about user needs or motivations? Consult your personas.
If the personas don't answer the question? Do some additional research and expand your personas.
Personas not the right fit for the question? No problem, use another tool like classic functional user requirements for this question.
Personas for food ordering apps demonstrate the usefulness nicely. You want to know the preferences that segment your users like restaurant quality, cost, and healthy options as well as demographics like age and income. From this you can derive a few clusters, and ultimately personas. Useful ones like "Jeff, the young workaholic that needs dinner brought to the office each night" or "Tom and Sarah, parents trying to find decently healthy takeout options for their kids". Now you can ask questions like, "do we have enough healthy options that Tom and Sarah's kids will like?" and "do we have an easy way for Jeff to reorder weeknight meals? Do we need a subscription plan?" You can do similar things with other user research tools but personas are very natural and intuitive for many scenarios.