This article is going to describe … would be a terrible opening for this article. It’s six words that convey nothing. You already know this is an article, and you already know that it’s going to describe something. We don’t see this, fortunately, because the importance of a strong and compelling opening sentence is well recognised. At the paragraph level, however, it’s easy to forget the importance of the first sentence. In scientific cases, a symptom of poor or lazy writing is opening a paragraph with Figure n shows.
When it comes to visualising your data, the most important question to ask yourself is what’s your point. Wording a paragraph by starting with Figure n shows will not convey the point. It tells me what you did, but not why I should care. Using this phrase would be like putting the Methods section of a scientific paper before the Introduction.
Other comparable opening sentences to watch out for go along the lines of An analysis of A and B shows … or A plot of x vs y shows … These are just disguised variants of Figure n shows. They all focus on the data themselves, not the context of said data.
Structuring paragraphs is something you probably learned in high school. I remember the general idea being to start with a topic sentence, provide some evidence and lower-level description, and then link this description back to the start of this paragraph or the beginning of the next one.
A similar, but more general approach to structuring paragraphs it the Uneven U, introduced in The Elements of Academic Style and neatly summarised in a post by the Thesis Whisperer. This says that the first sentence should be somewhat general, yet focussed (the second-highest level on a five-point conceptual scale). The following sentences should proceed downward to level 2 (descriptive or plain summaries) and level 1 (concrete examples or raw data). Level 4 and 5 sentences should conclude the paragraph to return to the overarching issues.
The problem with starting paragraphs with Figure n shows is that it is a level 1 or 2 sentence. It promotes focus on the raw data or the descriptive summary that a figure embodies. If you subsequently explain the high-level relevance of the figure by ending the paragraph with level 4 or 5 sentences, then you’ve just created a continually uphill path for the reader. Instead, start with a level 4 sentence, and then substantiate it with a reference to Figure n. You’re now creating a downhill path, which will give the reader some momentum for their return to the higher level sentences that conclude or generalise the previous sentences.
Here’s an example from a paper I’m currently writing. I’ve paraphrased to make it three sentences, the minimum necessary to create a U-shaped trajectory. First, the uphill ordering you should avoid:
Figure 3 shows that along-wind shear is stronger earlier in the day and the cross-wind component increases later. Further, the respective directions indicate a rightward turning. However, this observed evolution throughout the day is only qualitative evidence of inertial turning.
In an actual scientific paper, I might have to wade through 10 sentences or more in order to reach the last sentence, the one containing the useful, generalisable result. Here’s how I would rewrite the paragraph above:
In the northern hemisphere, we expect a rightward turning of shear. This is well demonstrated in Figure 3, which shows that along-wind shear is stronger earlier in the day and the cross-wind component increases later. However, this observed evolution throughout the day is only qualitative evidence of inertial turning.
Compare the underlying organisation of the two paragraphs graphically:

Forcing yourself to place a figure reference within the paragraph, rather than at the beginning, ensures that the first few sentences express your results in a larger context that is independent of the exact figure.
Another benefit of avoiding figure references at the beginning of the paragraph is keeping the reader’s attention on the text. When I come across Figure n shows at the beginning of a paragraph, I have to decide when to divert from reading the paragraph to consult the figure. If I flip to the figure immediately, then I’ll be looking at a figure with no context. Or, if I read the whole paragraph first, I don’t have the complementary information that the figure provides.
Putting the figure reference within the core of the paragraph encourages the correct flow for the reader: get the context and overarching information first, then see the data depicted in the figure, and finally return to the paragraph for the interpretation. (Put the context first and foremost in figure captions as well.)
Unfortunately, I’ve now made you aware of a problem that, if you’re like me, you can’t unsee. Each time you now come across a paragraph starting with Figure n shows, it will bug you. For that I apologise. I first became aware of the issue thanks to an anonymous reviewer of the paper I wrote back in 2014. They noted it is a minor point, but it is preferable not to start a paragraph with a callout to a figure. It did indeed seem a minor point at the time, but it’s one that’s stuck with me and influenced my writing ever since.
I am guilty of doing this and am suitably shame-faced now. Great post. I will change my ways.
I’ll also stop hyphenating shamefaced…