How to make sexy, visually-appealing and insightful reports.

The interest in data processing is increasing rapidly. This is for a reason. Data insights help better determine strengths, weaknesses, threats and possible opportunities of growth. In the real world, we can see big companies once-so-great get cast aside by much smaller ones because of the tiniest details, and other small businesses with life-changing ideas failing to make it out in the world, because of lack of vision. Those so small and yet significantly important details can be, to a certain extent, easily foreseen through the practice of studying and processing relevant data. A big part of that process concerns data preparation and integration (practised by DCube data professionals), the other part is what this subject is about; Data Analysis, when data scientists and engineers finish preparing data, they pass the baton onto data analysts to make insightful reports out of it. More often than not, said reports are hard to read, barely tell a story and the right numbers are hard to make out, in other words unattractive and hardly communicative. How do you avoid that ? How do you make practical, sexy and visually appealing reports then ? We’re going to see together how to make that happen in detail.

In this article, I’m going to use PowerBI for a few demonstrations. All other reporting tools are valid and the same rules apply to all. The audience is anyone making reports.

Story

One very important element of a good report is the story it’s telling (or failing to tell). Without having to read the title, we must be able to understand  as much as possible what the story behind the charts is. In order to guarantee a good data story, one must pinpoint the big lines of what they’re trying to convey. One practical way of going about that is to ask certain questions in relevance to the subject in question. For instance, we will be able to compose a data storyline from questions like these:

“What is the annual amount of sales for the current year ? What was the amount last year ? What caused that difference ? What regions were more responsible for the growth/decline of that amount ? what external factors had to do with the growth/decline in those regions? etc…”

These questions could be answered by charts strategically positioned almost in the same order as the questions asked. When the reader/decision-maker consults the report/dashboard, they’re going to be able to read swiftly through the displayed information, make out what story the charts are trying to tell. A counterexample would be to position the graph randomly through the report and let the reader investigate. An example in real life for this would be to read a book whose chapters aren’t in order. You might succeed in figuring out which chapter comes before which, but the work and time you put in figuring it out could be put elsewhere, to better purposes. Time is money.

Choice of colors

“[Color] is what gets your audience to see what you want them to see, feel what you want them to feel, and to do what you want them to do. Which hues you choose can also affect usability and whether content is readable or not. This is what makes understanding color psychology so important for the success of your content.” – Ashton Hauff (see Color Psychology Guide)

The choice of color is of vital importance to a report. Two things must be taken into account when doing this: the identity of the report and the coherence of colors.

Identity of the report

The colors of the report will reflect the identity of whom you’re making the report for or the topic you’re talking about through your charts. Whether it’s for a client of yours, your company or an academic project, the identity is always present. For instance, the chart below is for a French underwear company from their sales report. The choice of color was resembling skin shades. This comes off as more relevant than using the colors Red, Blue and Green for example.

Another example is the public annual report of Hermes Group. In their charts, they make sure they’re simple, pretty, concise and most importantly, that they reflect the identity of the company through their color orange. 

We can also notice that the big number is represented with orange (the dominant color and the identity color for the report). This is because it’s what the report maker wants us to focus on, the information that they want us to retain is that it’s more women than men. The other color is rather discrete and it doesn’t bring up anything new.

Coherence of colors

When making a chart, chances are you’ll have one with many different colors representing different elements. When that happens, you want to make sure you avoid using random colors to present your elements. Why do you want to avoid that ? A number of studies have been carried out about the human eye’s perception of colors. It would seem that it’s a lot easier for a human to spot out the differences between different shades of the same color than the differences between completely different colors.

As we can see in the example above, I have made two doughnut charts presenting the same imaginary data. We can notice that on the right-hand-side with its one color is a lot easier to read and look at compared to the left-hand-side and its too many colors. The eye finds it more comfortable to read on the right and can make out differences even faster. We can also point-out that on the right-hand side, the shade of color gets darker as the number gets bigger making it all the more convenient.

Choice of charts

The choice of charts is no less important than the story itself or the choice of colors. This is rather very vital for good data display and also, sometimes for avoiding unnecessary extra charts that might take up more space and hence make the report ugly and full of redundancies. In order to excel at the ergonomic aspect of report making, we must be aware of the four basic presentation types :

• Comparison
• Composition
• Distribution
• Relationship

We must be able to use the right chart for the right answer (or question). The report maker must ask the questions, how many elements am I trying to present? Do the elements change over time ? How many variables do I want to represent for each item? One of the best works in this field on what chart to choose is done by Dr. Andrew Abela. See chart down below (you can download the pdf version for the chart on here)

A good example of this is when we have many elements to represent with their respective shares. Using a pie chart, you’ll be faced with not only an ugly graph, but a also completely illegible one.

Beyond the ergonomic aspect of what chart to present the data right, you will also need to ask yourself the question; is the chart easy to read, can it be represented better with a chart of the same category (i.g. column chart vs row chart). Usually, the answer to this question is yes, it can be presented with something else.

Another example of that would be, they both sort of offer the same functionalities and serve the same purposes, except that it’s easier for the human eye to make out the size of the share of a color compared to others on a doughnut chart than on pie chart. This is because the human eye doesn’t have to go all the way to the center of the circle.  (read Pie Chart vs. Donut Chart: Showdown in the Ring)

Below is a good example from the article cited above, about the difference between a pie chart and a doughnut chart. Both charts represent the same data, but for most people, it’s significantly easier to read on the right hand side.

Use of icons

Sometimes, one silhouette is better than an entire sentence. In the context of reports, icons in the shape of mono-colored silhouettes can be very interesting and make the report more visually-appealing. One example at that is down below

Relevance of titles

Titles of charts must be very concise and relevant. The sentence must include as few words as could be in order to state as much as possible.

Exports of OPEC countries in 2020 by Barrel

vs

Exports of opec 

A choice might be made by the report-maker not to include certain information in the title if it’s obvious enough in the whole report page, or the chart itself (example, the report page is about the year 2020 and it’s mentioned in the title of the page).

Font code

A font code must be respected to guarantee a more coherent view. The same font type and size for the same category types. Multiple colors also in captions and titles, should be avoided, if not, a color code must be respected. 

In order to make a nice-looking report, many aspects must be taken into account. Colors, texts, charts and the story. The identity of the report must be respected. No matter the tool we use to make our reports, the approach stays unchanged.