Begin with the operating problem
A credible partner should be able to restate the business problem, the people affected and the outcome that will define success. A long feature list is not a substitute for this understanding.
Ask how assumptions will be tested before the most expensive engineering work begins.
Evaluate evidence, not vocabulary
Review relevant systems, working demonstrations, references and the team responsible for delivery. Terms such as innovative, scalable and world-class only become useful when they are connected to specific architectural and operational decisions.
Make ownership explicit
Confirm who owns product decisions, data preparation, security review, acceptance testing, deployment and post-launch support. Projects are easier to govern when these responsibilities are written down early.
Compare the whole delivery model
The lowest initial quotation may exclude discovery, migration, training, monitoring or support. Compare proposals against the same scope, assumptions, exclusions and acceptance criteria.
Apply the thinking