The biggest mistake with a first version is trying to build the finished product. A first version exists to put something working in front of real users quickly. Scoping it well takes one focused conversation.
Start with one person and one problem
Describe the single person who will use it and the one problem it solves for them. If you have to say and more than once, you are describing two products.
Define the outcome, not the features
Write one sentence about what is different for that person after they use it. Features come second. This sentence is what you test every decision against.
Sort everything into three lists
Must have for launch, nice to have later, and not now. Be strict with the first list. If removing an item does not break the outcome, it belongs on the second list.
Agree on the edges
Write down what the first version will not do. This protects you from scope growth, because every new request is checked against a list you both agreed to.
Set the date, then fit the scope
Choose when it must be working, then shape the scope to fit. This keeps the project honest. A fixed date with a smaller scope beats a bigger scope with no end.
Leave with a plan you can approve
A good workshop ends with a scope you can read in a few minutes, a price and a date. If you cannot, the scope is still too vague to start.
Have a rough idea and no spec?