The missing piece in all of this seems to be the lack of a properly qualified researcher finding you the answers to those questions at every stage.

As a full-time researcher, I follow this process:

1. Get the PM and the whole product team into a meeting. Have them explain the context, what they want to achieve, and develop a proto-persona where they map out all their assumptions and then copy those into a 2x2 matrix to show how important that assumption is to the product, and how confident they are that it's true.

2. Develop an interview guide and share it with the team, who add comments and clarifications and requests - but the researcher still has the final say on what is asked.

3. Interview 5 users of each persona type (e.g. 5 SysAdmins who are very experienced and 5 who are new to it) - the researcher asks the questions while the team observes. There is 5-10 minutes at the end for them to ask their own questions after all the non-leading questions have been asked.

4. The researcher analyses the findings and writes recommendations.

5. The researcher invites the team to review the affinity diagram (the grouped, themed clusters of quotes and insights) and add their own observations. These findings are compared with the proto-persona - the first time, almost every assumption will be wrong! The PM now understands that they don't know users half as well as they thought they did, and is more inclined to listen to advice. The researcher then shares the recommendations.

Collectively, a chart is completed: what the real problems are, what users are doing now, and the characteristics of a solution that is better than their current solution.

This then provides the foundation from which designers and engineers can work and, since they were present throughout this process, they understand what each individual post-it means.

(There are also smaller, lighter feedback loops such as quick usability tests during development.)