You can't deliver any real value at all to a problem if you don't spend at least 30 minutes on it.

After that, you add value quickly, until a point of diminishing returns (minutes or days more, really, it depends on the task). What you can do in short stints is gain understanding, input, feedback, and context that represents the “potential for value”. But just gaining context isn't enough, you need blocks of time to release that value, to take all that context, reform it into hypothesis and framework, and render solutions and insights against it.