The introduction as a funnel

A good introduction is a funnel: it starts wide, where every reader is, and narrows step by step to exactly what this paper does.

The four steps

1. The big picture. Why this area matters. (One or two sentences, not a history lesson.)
2. What is known. The key work so far, with citations.
3. The gap. What isn't known, or what's wrong with current approaches. This is the turning point, often signalled by "However," or "But".
4. This paper. What you did to fill the gap, and, briefly, what you found.

Example(1) Plant diseases cause large crop losses, and early detection helps farmers act in time. (2) Deep learning detects diseases well from leaf photos when large labelled datasets are available. (3) However, most crops and regions have only a few hundred labelled images. (4) Here we test whether simple colour augmentation closes the gap for small datasets, and find it recovers most of the accuracy lost.

Common mistakes

• Starting too wide: "Since the dawn of agriculture..." Start where your readers already are.
• A review without a gap: paragraphs of related work, but no clear "however".
• Hiding the contribution: readers should know what you did by the end of the introduction, not on page 6.

Many papers end the introduction with a short list of contributions, which reviewers appreciate.

RememberWide to narrow: why it matters, what's known, the gap, what this paper does.