An MVP is the smallest version of a product that genuinely solves users' problems and lets you verify the idea in the market – before you invest the full budget in features nobody may need.
The key word is "viable": an MVP is not a shell or a demo – it must deliver real value, only in a deliberately narrowed scope. The art is cutting features, not quality.
A good selection question: "without what does the product make no sense in the first week of use?". Everything else – integrations, statistics panels, the mobile app – can wait for proof that the core works.
Traps: the monster MVP (18 months of building "the minimum"), the hollow MVP (so much cut out it verifies nothing) and no measurement plan (we built it – how do we know it works?). In our product workshops we scope MVPs around verifiable hypotheses.