Almost every website project that runs over budget and over schedule shares one root cause: the scope was fuzzy when the contract was signed. "A new website" is not a scope. "Eight pages, three templates, migrated content, and a launch by the end of Q2" is a scope. The difference is whether you and your agency are arguing about change orders in week six.
Start with pages and templates, not features
Count the unique page templates, not the pages. A site can have forty pages and four templates — and it is the templates that drive design and build cost. Write down every template you need (home, service, article, contact) and roughly how many pages sit on each. That single list does more to control cost than any feature wishlist.
Decide who writes the content
Content is the silent killer of website timelines. If you are writing it, your project is gated on your slowest writer, not the agency. Be honest about whether your team can produce the words on schedule, and if not, scope the writing in from the start. The fastest projects are the ones where content was ready before design began.
Name what is explicitly out
List the things you are NOT doing this round — the blog redesign, the customer portal, the integrations — in writing.
Agree on what triggers a change order, and what is a free in-scope revision.
Set the number of design revision rounds up front; "until we are happy" is not a number.
A scope is as much about what you exclude as what you include. Write the exclusions down and you turn week-six surprises into week-one decisions.