Paper-writing guide, part 2: decide what you'll say before you decide how to start
Before you write a word, pin down the one sentence your paper argues. See how that core claim shapes your title, your figures and the order of your results.
"I've finished all my experiments. So how do I actually start writing?" It's the most common worry among students writing their first paper. The data is all there, but the order and structure for laying it out are anything but clear. Plenty of students begin by writing up their experiments in the order they ran them, only to watch the logic fall apart halfway through and end up starting over.
Part 1 of CRI's seven-part paper-writing guide asked why researchers write papers at all. This second part is about structure, and its advice is simple: before you decide how to start, decide what you are going to say. Settle your conclusion first, and build everything else around it.
A paper is a story, not a lab notebook
Scientific writing still needs a narrative. Think about explaining an idea to a friend. Some people lay it out clearly and make it interesting; others talk at length and leave you unsure what they meant. Papers work the same way. However important your research is, if it is hard to follow and dull to read, it won't hold anyone's attention.
A paper's job is to present your new finding so that others can grasp it easily and find it interesting. Structure is what makes that possible.
So you don't have to retrace the path you actually took: we ran an experiment, it failed, and then something unexpected turned up. Readers care less about the sequence of your lab work than about why you believe this result matters. Writing a paper starts with rebuilding its structure around the message you most want to deliver.
Start with the conclusion
One secret of good paper writing is to pin down your conclusion before anything else. That doesn't mean the Conclusion section you write at the end. It means the core message of the whole paper: a one-sentence claim that answers the question, "So what did you discover that's new?"
Settle that sentence first and the paper hangs together far better. Research rarely goes as planned. Results can point somewhere you didn't expect, or a new insight can arrive midway. When that happens, don't cling to your original aims or the order you worked in. Rebuild the paper around the core claim.
Say experiment A failed but experiment B produced an interesting result. The paper doesn't need a long account of A's failure. Organize it around why B matters and which question it answered.
Have a question you want to investigate?
Explore our faculty-guided programs, or tell admissions what you are curious about.
Boiling a conclusion down to one sentence is harder than it sounds, especially when your experiments produced a range of results and you're not sure which should take center stage. A useful first step is to pick out the three most important keywords in the conclusion you want to argue.
One published optics paper summed up its conclusion like this:
"We controlled optical near-fields by shaping the wavefront of incident light through random nanoparticle media."
Its three keywords are:
Controlled optical near-fields
Wavefront shaping
Random nanoparticles
These three are the experiment's core concepts and the message that runs through the entire paper. Build the text around what each one means and how they connect, and the paper takes on a logical flow almost by itself. The same sentence also becomes the blueprint for the figures, the placement of the data and the interpretation. Everything in the paper is, in the end, material for making that one conclusion convincing.
Let the title grow out of the conclusion
With the conclusion settled, turn to the title. Many people leave it for last, but in good papers the conclusion and the title fit together naturally.
The same three keywords can become your title. The trick is not to string them together as they are, but to turn them into a phrase that is interesting, catches the eye and gets straight to the point. The paper above was titled:
"Subwavelength light focusing using random nanoparticles"
Here is how the keywords were reworked:
In the core sentence
In the title
"optical near-fields"
Became "subwavelength light focusing"
"controlled"
Recast as the verb "using"
"random nanoparticles"
Kept as the key term, at the end of the title
The meaning stays the same, but the phrasing is more intuitive and more likely to draw a reader in. A little reordering and rewording of your conclusion's keywords can give you a title that reads smoothly and makes an impact.
Design your paper in figures
Most readers look at a paper's figures before its text. However well you write, if the figures are cluttered or their order feels awkward, the heart of the paper won't come across.
That is why many researchers draw their figures before they write a paragraph. They print out their results, spread them across the desk and ask which order would be most convincing.
Far more people read the abstract and then skim the figures from Figure 1 onward than read a paper from start to finish. Each figure therefore has to deliver its own message, and together the figures have to lead naturally to the conclusion. A typical plan:
Figure 1: the background concept, the problem and the core setup of the experiment
Figure 2: the first experimental result, supporting part of the core conclusion
Figure 3: further results that settle the remaining questions and strengthen the claim
Every figure converges on the same sentence: your core conclusion. Fix the role and position of each figure in advance, and the introduction, results and discussion fall into place when you write. The text comes afterward.
Experiment order is not paper order
Experiments never go entirely to plan. The first attempt fails; unexpected results pop up along the way. But a paper isn't a diary of everything you tried. It reconstructs your results to deliver one conclusion as persuasively as possible, so the question to ask is:
"In what order should I show these results so readers accept my conclusion most easily?"
Suppose you ran experiments A, B and C in that order. If that sequence would leave readers asking, "Wait, why did they do this?", the paper doesn't have to follow it. B, then A, then C might work better:
Order in the paper
Experiment
Its job
First
B
Shows the core message most directly
Second
A
Supplies the background needed to understand B
Last
C
Answers doubts about the core claim
Real research involves chance discoveries, failures and repetition, but a paper is an argument, and its results belong in the most efficient order. Reordering this way lets readers answer "So what's new here?" as quickly and clearly as possible.
As you plan the structure, ask:
What role does this experiment play in the paper as a whole?
Which experiment contributes most directly to the core claim?
Which experiment gives readers the information they need first?
Which experiment answers the questions or objections that might come up along the way?
Then arrange every figure and section to follow the flow of your argument. A paper is persuasive writing: it has to show why your finding matters and how you proved it, in the way readers will find most convincing.
A worked example: how one published paper was built
If this still feels abstract, the quickest way in is to look at how a real paper was put together. The example below is reconstructed from the structure of the same published optics paper.
Read in order, a published paper's figures reveal how its argument was designed.
Put in the words of its title, the sentence at the heart of the paper reads:
"We achieved subwavelength light focusing using wavefront shaping through random nanoparticle media."
The figures that support it appear in a somewhat different order from the one in which the experiments were done.
Figure 1 presents the key concept. A schematic of the whole experiment shows which physical principle brings the light to a focus and previews the paper's storyline. From the outset, readers can tell what the paper sets out to show and through which experiments.
Figure 2 presents the results. Its core data show how light converges after passing through the random nanoparticle structure, pairing actual experimental images with an intensity graph. It proves the "focusing achieved" part of the core sentence, and showing that result early gives the argument a strong start.
Figure 3 fills in the supporting evidence. Verification experiments take on the questions the main result might raise, such as "Is this effect statistically reproducible, or just a fluke?" Control conditions, repeated measurements and noise comparisons address likely objections before readers can raise them, which makes the argument more convincing.
The figures and the flow were redesigned around the message the paper set out to argue, not the order of the experiments. Someone seeing it for the first time should be able to look at Figures 1 to 3 and think, "I see what this paper is saying." That is what good structure looks like.
What to remember
Writing a paper means persuading readers of one message, so its structure should grow from your conclusion, not from the order of your experiments.
A paper is a story. It isn't finished until readers can understand it.
Decide your conclusion first. Design the structure around a single central sentence.
Pull out your keywords. They steer the flow of the text, the title and even the figures.
Build the structure around figures. Design the results visually first; the writing follows.
Logical flow beats experimental order. Rearrange your results in whatever order helps readers understand.
Papers get easier to write with experience, but practicing conclusion-first thinking from the start is a foundation every researcher needs. Right now, just making sense of your results may feel like plenty. This way of structuring a paper will still pay off whatever topic you take on next, because it teaches you to design a paper rather than merely write one.
The paper-writing guide series
Why write a research paper at all?
Decide what you'll say before you decide how to start (this article)
Figures are the face of your paper
Writing results and discussion that work
When and how to write the introduction and abstract
Finding, citing and managing your sources
Proofreading, the final design step
Every part is on the CRI blog. Part 3 picks up from here with how to plan and build your figures.
If you'd like to try conclusion-first design on a paper of your own, the Seoul Research Program brings small groups of students together with visiting professors from US and UK universities to carry out real research and write it up in English.
