I was a consultant prior to being a university professor. I've been enjoying the years traveling around the world, facing customers and solving their real-world problems. As a professor, I would like to stay close with the industry as much as I can. Continuing my consulting practices around forecasting is probably the best way to achieve this goal.
Regarding the consulting projects, here is my order of preferences (from more preferred to less preferred):
1. A brand new problem that is not well-established yet.
    Example: forecasting at the household level would have been a good example in 2011.
2. A challenging problem that nobody else has solved before.
    Example: forecasting with smart meter data.
3. A challenging problem that needs a better solution.
    Example: solar generation forecasting.
4. A well-established problem that has been solved elsewhere but needs customization effort for a different customer.
    Example: short term load forecasting at the corporate level.
I have a desire to publish scholarly papers about the interesting findings, if there is any, from these consulting projects. In principle, the more likely a project is going to turn into a high quality scholarly paper, the more interest I have working on it.