The everyday problem
Demonstrations are hard to compare when each provider chooses a different scenario. One may emphasize speed while another emphasizes navigation or integrations. A common scenario gives the evaluation team a more useful basis for recording what was actually shown.
A practical approach
Use the same fictional task and questions across demonstrations. Record demonstrated behavior separately from planned or described capabilities. Include an incomplete-information case and ask how the user understands the source and status of the result.
Try this with your team
Build a comparison sheet with columns for demonstrated result, required setup, unresolved question, and follow-up evidence. Avoid reducing every uncertainty to a score.