Scope first
Most capstone trouble is scoping trouble, and it shows up in month two rather than week one. Before committing, ask three questions. Can I get the sources or data I need, with the access I actually have? Can I answer this within the word count without either padding or compressing? Is there enough existing work to build on, and not so much that my contribution disappears?
A narrower question answered well beats a broad one surveyed thinly, every time.
Work backwards from the deadline
Set the submission date, then place the milestones behind it: proposal, literature complete, method settled, first full draft, revision, formatting and citation check. Put the first full draft at least three weeks before submission. Not two. The gap between a draft and a finished project is always larger than it looks from inside the draft.
Keep a research log
A dated file recording what you read, what you decided, and why. It takes ten minutes a week and it does three things: it stops you re-reading sources you already rejected, it supplies the methodology section you will otherwise reconstruct from memory, and it lets you answer the supervisor question about why you chose this approach without improvising.
Holding the argument together
Long projects drift. The chapter you write in week ten quietly argues something the introduction does not promise. Two habits help. Keep your research question written at the top of every working file. And every few weeks, reverse outline the whole thing in one sentence per section, then read that list on its own.
Working with a supervisor
Supervisors respond to specific questions and struggle with open ones. Instead of asking whether this is any good, ask whether the second section is doing enough with the theory, or whether it is defensible to exclude a category of source. Send material in advance, with a note about what kind of feedback you want.
When it stalls
Every long project has a stretch where it seems incoherent. Usually the problem is a question that has quietly changed as you learned more, and the fix is to rewrite the question to match what you are actually finding. That is not failure. It is what the research was for.