Guide clusterResearch Roadmaps
A research roadmap template gives your proposal a clear structure for the question, inputs, method, validation and translation stages. Start by working backward from the research question, replace each placeholder with your study details, and label every arrow with the evidence or decision it carries.
Why a roadmap is not a timeline
A timeline answers when. A roadmap answers why, in what order, and on what evidence. If your diagram can be read as a Gantt chart with the dates removed, it is a schedule wearing a diagram's clothes. Reviewers reading a proposal rarely want to know that Phase 3 ends in month 30; they want to know what Phase 3 establishes that Phase 2 could not.
The practical test is the arrows. In a timeline, an arrow means and then. In a roadmap, an arrow carries evidence: a validated dataset feeds a model, the model produces predictions, and those predictions are tested against a held-out sample. If you cannot say what travels along an arrow, the two boxes on either side of it are probably not causally connected, and the diagram is overstating the design.
If an arrow cannot be labeled, remove it or insert the missing stage.
Start with the question, not the boxes
Open a blank document and write the research question as a single sentence with a measurable outcome. Everything downstream has to earn its place against that sentence. If a module does not change what you can conclude about the question, it is either a service activity or a distraction, and neither belongs in the diagram.
Write the research question in one sentence, naming the population or system, the comparison, and the outcome you will measure.
List the specific unknowns that must be resolved before the question can be answered. These become your stages.
For each unknown, name the data, sample, corpus or material that resolves it, and confirm it is actually accessible to you.
Name the analysis, model or experiment that converts that input into a claim.
State the validation criterion that would tell a skeptic the claim holds.
Only now draw boxes, in the order the reasoning runs rather than the order the calendar runs.
The five stages most roadmaps need
Across disciplines, the defensible roadmaps tend to share a five-stage spine. The vocabulary changes with the field, so a clinical study says cohort where a materials project says specimen set, but the logical dependency between stages does not change. Use the table below as a checklist for your own draft rather than as a template to fill in.
| Stage | What it contributes | Evidence it produces |
|---|---|---|
| Problem framing | Narrows a broad field into one answerable question and states the gap | A testable hypothesis or aim statement |
| Data and inputs | Names the sample, corpus, instrument or material the study depends on | A defined and accessible dataset or specimen set |
| Method | Describes the analysis or experiment that converts input into a claim | A result with a stated level of uncertainty |
| Validation | Sets the test that could falsify the claim, including baselines and controls | A comparison against a baseline or independent sample |
| Translation | Explains who uses the result and in what form | A decision, design, guideline or follow-on question |
Choosing the right level of detail
The most common structural error is mixing laboratory steps and program-level milestones in one figure. A roadmap for a funding panel should sit at the level of aims. A roadmap for a lab meeting should sit at the level of protocols. Decide your reader before you decide your zoom level, because two audiences need different diagrams even for the same project.
- Panel version: one box per aim, arrows labeled with the evidence that links them, nothing below the level of a deliverable.
- Committee version: add the validation stage and the failure risk of each aim, because a committee will ask what happens if an aim does not succeed.
- Lab version: expand the method box into its real sequence of protocols, instruments and checkpoints.
- Slide version: cut to three or four boxes and move the detail into speaker notes rather than into the figure.
Testing the roadmap before you submit it
Hand the diagram to a colleague outside your subfield and ask them to describe the project back to you. Every point where they hesitate marks either a missing arrow label or a box that assumes knowledge they do not have. This takes ten minutes and catches more problems than a week of polishing typography.
Then run the falsification pass. For each validation box, ask what result would make you abandon the aim. A roadmap with no branch that can fail is not a research plan; it is a description of an activity. Including one honest contingency branch signals methodological maturity rather than doubt.
Worked example
Worked example: a soil carbon measurement roadmap
A doctoral project asks whether cover cropping increases stable soil carbon on degraded farmland within five years. Framed as a roadmap, the problem box states the gap directly: routine soil surveys measure total organic carbon, which cannot distinguish fresh residue from stabilized carbon, so the effect of the practice remains unclear at the timescale farmers actually plan on.
The data box names two sources rather than one: paired fields under cover crop and conventional management, plus archived baseline samples. The method splits into two linked boxes, physical fractionation to separate particulate from mineral-associated carbon, followed by isotope analysis to attribute carbon to the cover crop rather than to pre-existing stocks. Splitting the method matters, because the second step only becomes interpretable once the first has succeeded.
Validation names a specific test: the same fractionation applied to archived baseline samples, so any difference is measured against a known starting point, with a modeled counterfactual as a second check. Translation shows where the result lands, as a decision rule for growers about which fields respond and as an input to regional carbon accounting. The final figure holds nine boxes at panel level, and a reviewer can trace the argument in one direction without backtracking.
Common mistakes
Frequently asked questions
How many boxes should a research roadmap have?
At panel level, five to twelve boxes is usually enough: roughly one to three per stage, plus a contingency branch. If a reviewer needs more than a minute to trace the main path, the roadmap is carrying detail that belongs in the accompanying text instead.
What is the difference between a research roadmap and a research framework?
A framework shows the concepts and relationships you are testing, usually arranged by role. A roadmap shows the sequence of work that produces evidence. Many projects need both, and the framework often fits naturally inside the first stage of the roadmap.
Can AI tools be trusted to draft a research roadmap?
An AI assistant can organize stages you have already described and suggest a layout, which saves formatting time. It cannot know your methods, your data access, or your field conventions, so treat every generated box as a draft to verify. You remain the author of the design.
Does the roadmap need to show every risk?
No. Show the one or two risks that could invalidate an aim, and place them where they occur rather than in a separate list. A risk paragraph at the end of a proposal is easy to write and easy to ignore; a contingency branch inside the flow is harder to skim past.
Sources and further reading
- Academic writing and structure guidancePurdue Online Writing Lab, Purdue University
- CRediT - Contributor Roles TaxonomyNISO
- LaTeX templates for theses, papers and proposalsOverleaf
- EQUATOR Network - library of reporting guidelinesUK EQUATOR Centre, University of Oxford
- ORCID - persistent identifiers for researchersORCID
These are the organisations whose published guidance the recommendations on this page follow. Listing them does not imply that they endorse Pathform AI, and this page is not affiliated with them.
Editorial update: Reviewed October 27, 2026. The roadmap guidance connects research planning intent to editable PowerPoint structure; confirm the proposal or thesis requirements before submission.