Research Foundations › Module 6: Planning a project
Scope and timeline
Lesson 1 of 3 · 2 min
Most first projects that fail do not fail because the idea was bad. They fail because the project was too big for the time available.
Make it small enough to finish
A good first project can be finished in 8 to 12 weeks of part-time work. That usually means one question, one main experiment or study, and one or two datasets. You can always extend a finished project. You cannot publish an unfinished one.
Plan in stages
| Weeks | Stage | Done when |
|---|---|---|
| 1–2 | Read and refine the question | You can name 10 key papers and state your question in one sentence |
| 3 | Set up | Data downloaded, licence checked, baseline code runs |
| 4–5 | Reproduce a baseline | You get close to a published number, or know why not |
| 6–8 | Main experiments | Every planned run is done, logged and backed up |
| 9–10 | Analyse | Figures and tables made, results checked twice |
| 11–12 | Write and share | A full draft, read by someone else |
- 1–2
- Stage: Read and refine the question
- Done when: You can name 10 key papers and state your question in one sentence
- 3
- Stage: Set up
- Done when: Data downloaded, licence checked, baseline code runs
- 4–5
- Stage: Reproduce a baseline
- Done when: You get close to a published number, or know why not
- 6–8
- Stage: Main experiments
- Done when: Every planned run is done, logged and backed up
- 9–10
- Stage: Analyse
- Done when: Figures and tables made, results checked twice
- 11–12
- Stage: Write and share
- Done when: A full draft, read by someone else
Plan for things to go wrong
Something will go wrong: data that is messier than expected, code that will not run, a result that makes no sense. Add buffer time, roughly a quarter of the total. Decide in advance what you will cut if you fall behind, for example one extra dataset.
Milestones, not just deadlines
A milestone is something you can show: "baseline reproduced", "first results table". Share each milestone with a mentor or study partner. Regular check-ins catch problems weeks earlier than a single deadline at the end.
Keep a research log
Write a short dated note every working day: what you tried, what happened, what you will try next. It feels slow, but it saves you when you write the methods section months later, and it is your record if anyone asks how a result was produced.
RememberReproducing an existing result first is not wasted time. It proves your setup works, and many good projects grow out of the gaps you find while doing it.