How to draw a technical roadmap

Technical roadmaps for modeling papers and proposals: structure, method, prompt template.

A technical roadmap is the standard overview figure in math-modeling papers, research proposals and grant applications: one figure that lays out the path from question to method to outcome. What reviewers read first is its hierarchy — question layer, method layer and result layer each occupy a distinct visual zone, and arrows only express how data and decisions flow.

What a good technical roadmap looks like

• Top: the research question or goal — one sentence, not a paragraph.
• Middle: stages arranged as rows or columns — data preparation → modeling → solving/experiments → validation, each stage in a dashed grouping box.
• Inside each stage, a chain of boxes carrying method names and key parameters (algorithm, model, number of constraints).
• Arrows between stages are labeled with what they pass on (data / features / candidate solutions).
• Bottom row: evaluation metrics or final deliverables (Pareto front, accuracy, section mapping).

A prompt template you can copy

Draw a technical roadmap for a math-modeling paper, landscape 16:9, white
background, academic style, no title bar, no decorative icons.
Layout: four stages top-to-bottom, each in a dashed rounded box with the
stage name in its top-left corner.
(1) Question layer: one box with the goal "<one-sentence goal>" and 2-3
    sub-question boxes below it.
(2) Method layer: data preparation (<source and scale>) → preprocessing
    (<method>) → model construction (<algorithm/model + key parameters>) →
    solving and experiments (<solver / setup>). Method names inside boxes,
    no long sentences.
(3) Validation layer: sensitivity analysis, comparison against <baseline>,
    metrics (<metric names>).
(4) Output layer: one row of three deliverable boxes — <result 1> / <result 2>
    / paper-section mapping.
Arrows between stages are labeled (raw data / feature matrix / candidate
solutions).
Semantic multi-color palette: question layer light blue, method layer light
green, validation amber, outputs light purple; consistent within a stage;
never a single blue for the whole figure.

Common mistakes

• Drawing it as an infographic poster: big title bar, numbered modules, conclusion strip — that is marketing art, not a roadmap.
• Boxes containing only empty words like "modeling" or "solving", without method names or parameters.
• Unlabeled arrows — the reader cannot tell what flows between stages.
• Everything squeezed into one column with no visual grouping, so the hierarchy is invisible.

Every number in the figure (sample size, constraint count, metric values) must exist in the paper itself; double-check any value the model fills in before it goes into the manuscript.