Designers often want to hop into "blank slate", and run interviews with non-customers. The expectation is to get insight on unattended needs of people, and only then the design work would start bringing solutions.
This is a theoretical view of things. In real life, companies will rarely design from/for a clean slate. They either have a customer base (and want to improve the product) or they have a customer base and want to launch a NEW product.
TL;DR: Get answers to needs, not promises about your solution
When you interview a non-customer, ask only about the needs they have concerning a certain universe. It won't make any sense to gather people according to demographics. People create more relevant clusters when it comes to behaviours (purchase behaviours, habits) rather than education levels or age groups. This research is expensive. Why? Because you don't know if despite what they say, they will ever actually be interested in your product.
So if you are designing for IMPROVEMENT, check these boxes:
Research your BEST CUSTOMERS, to understand the core value of your product/service, to make sure you will IMPROVE it, and not kill it.
Look for needs, and focus on answers they know. They know: what they have done in the past. They don't know: if they will actually buy your proposed solution to their need. Also: sometimes they know their need, sometimes they don't.
A/B test your improvements. That can be a big website improvement, a new retail layout, or a new offer to which you collect "I'd like to know more" clicks.
Implement what works.
If you are designing NEW SERVICES, you are close from a clean slate. A startup, for example, is often in this situation. Here's the tee:
Research your CUSTOMERS-TO-BE, starting from need. Don't go after "people in their 30s", go for the group of people generated by the need you want to solve: "people who need to manage a lot of parking tickets", for example.
Spend a little time talking to people with that need, but just to get your first ideas. Remember, that's BEFORE you have any idea of what to offer them, AND if you are starting from a clean slate. Importantly: Ask questions they know. Come up with the conclusions they don't know.
Your solution is a business hypothesis, and you want quantitative validation for it. Ten interviews will not validate an idea. You end up with something like: "I would buy that". It's not nearly enough a predictor of success. When you don't know the customer, win them before you interview them. How? Take the this is not a test approach, and run value proposition ads or live beta services.
THEN you can spend time understanding WHY people clicked your new offer, and more about them. What's the difference? Simple: you are investigating people who are actually moved to give you their money in exchange of what you offer. You will be able to improve your offer, and hit your target with better rates. Stick with indications, not promises.