Office Hours
How I create an Analyst Style Guide
For an effective data story presentation
In my first blog post How I do Data Storytelling, I have mentioned that one key part of a good story telling is to create a visually affordable presentation, which should be be a great complement to your verbal communication rather than distracting your audiences from it.
With this post I would like to expand on the tips that I mentioned in the previous post, and share some good practices and examples that I find really helpful in the real life of a data analyst.
This post will cover analysis reports which are usually produced in the form of essays or slides and use charts and texts to share insights on explorative, complex topics such as “Why do we see a drop in sales revenue this month?” or “What can we do to improve user funnel conversion?”. Often they are shared and/or presented with a mixed group of audiences which can include product managers, UX/UI designers, operations managers, marketers, senior management, etc.
We will not cover how to create interactive dashboards in this post. :)
Here are a few principles this post builds upon, i.e., all tips that will be discussed later on serve one or more of these principles.
- Our time is limited. The content in our presentation should be prioritised in a way that the most important information always comes first for our audiences to extract first.
- We assume that consuming complex information that involves data is not an easy job for everyone. Therefore, we should design our presentation in a manner to minimise the effort that it takes our audiences to extract information from it.
- Human brains love patterns and consistency. A clear and consistent structure can train our audiences to get used to our storyline and make it easier for them to ingest and digest the presented content.
That’s all! Below you can find four tips that I find really useful in producing an easily digestible presentation to share data insights with a diversified group of audiences. At the end of this post, I also share a link to my own template.
I hope you find them useful as well — any feedback, thoughts or questions will be highly appreciated!
Always put conclusions first and let details follow.
Conclusions first — this is my biggest takeaway from the book The Pyramid Principle. The idea behind this approach is to help your audiences to get the key idea, usually a “what”, at the first sight and let them decide on their own whether or when they want to know more details behind the “what”.
For example, if I am asked to look into why there is a drop in our sales revenue this month, I would put a summary slide as the very first page of my report. On this summary slide there are usually three sections with each section containing 1–2 sentences.
Why three? This is also a trick I learn from the Pyramid Principle. Three is the magic number of ideas in a group that can most effectively communicate your message. In case you are interested, I find this post very helpful in explaining this rule in details: https://medium.com/lessons-from-mckinsey/the-rule-of-3-c1cd82dbc96
This slide helps an audience who checks his or her inbox in the morning, opens your attachment and has only 5 minutes per email. The first slide gives the audience a chance to capture the key message, and let one decide whether he or she wants to dive into the details now or save it for later. He or she does not have to view through the whole presentation before making that call.
Another application of this tip is to use meaningful sentences as slide titles. This means that, instead of titling a slide with “Sales outlook”, write down “Sales outlook is promising in the next 12 months”.
Why do we bother to write a longer headline? Again, imagine you are a the Director of Sales and you have only 5 minutes to browse the 10-page report. When you go through the slides, the headline is the first thing that catches your eyes. If you see the headline “Sales outlook is promising in the next 12 months” — Phew! That’s a good news and you don’t need to worry about the sales outlook now. Maybe you are curious about some specific details about the sales revenue forecast, but that is not your priority now and you can definitely read the report in detail later on when you have more time. However, if you read only “Sales outlook” as the headline at your first sight, you are not sure whether that is bad or good. You would have to go down to the charts and texts to find out that, and this unfortunately adds to the effort it takes you to get the exact, simple information you need right now.
Meanwhile, compared to “Sales outlook”, headline “Sales outlook is promising in the next 12 months” contains a thought judgement that can better engage your audiences with the content of the slides.
Use layouts to create patterns for your audiences to learn.
Whether we admit it or not, our brains love patterns. When we see a series of [1, 3, 5, 7, 9, ?], our brains automatically replace the question mark with the number 11. The same applies to your presentation! When your audiences open or see your slides, their brains start to learn your patterns and apply that knowledge when they absorb the information from your slides.
For example, I have a layout that is called the Driver Layout. With this layout I usually plot the trend of the Goal KPI on the left side with a text description in the same box. I use the larger space on the right side to plot the trends of the Driver KPIs that can explain the development of the Goal KPI. In the following slide, the Goal KPI is Sales Revenue and the Driver KPIs are Leads (#), Conversion Rate (%) and Order Value (EUR). In addition, I also use some pop-up comment boxes to comment on irregularities.
I use this layout across the whole presentation when I want to explain the development of a Goal KPI based on Driver KPIs. When my audiences see this layout the next time, they would already know how my storyline would look like and it makes easier for them to follow my story.
Another layout that I frequently use is the Contrast Layout. I found this pretty helpful when I want to highlight the difference in two or more KPIs given the same segmentation. With this layout, I usually divide the space equally depending on the number of the metrics I want to compare with. I also apply contrasting colour schemes to let my audiences know that they should pay attention to the gap or difference between different segments.
Meanwhile, I also find that the choice of colour schemes plays a key role in creating patterns. Use contracting colours when you want to highlight the difference, gap or deviation and choose sequential palette when you want to address sequences or temporal trends. For example, with the following chart I want to show the improved monetisation over cohorts. I create a sequential palettes where the younger the cohorts the darker the curves, so that my audiences can easily follow the change of colours and notice that the darkest curves (and thus more recent customer cohorts) are always located higher on the plot, in other words, the recent customer cohorts generate more revenue on average than the older cohorts did.
A random palette, on the other hand, fails to help the audiences establish a connection between the darkness of the curves and the recency of the cohorts and thus adds the challenges for your audiences to read the chart.
Let consistency rule every corner of your slides.
Once you have laid out the structure and patterns of your slides for your audiences to follow, the next step is to minimise any tweaks that may distract your audiences them from following the patterns you create.
Use the same colour scheme for the same dimension across your slides whenever possible. For example, if you have used gender as a key dimension in multiple charts for comparison where blue stands for men and yellow for women, stick to this pairing across your presentation wherever possible. By doing so, the next time when your audiences see the “Blue/Yellow” pair on your visualisations, they know intuitively this is a comparison across genders without having to read the little legend.
Use the consistent vocabulary in your written and verbal communication — in particular, stick to your domain-specific glossary. If the word “Lead” in your organisation refers to anyone who visits your website but does not register for an account, then stay with this definition in the rest of your presentation. If you out of sudden start to use the word “Lead” when you describe someone who registers an account but does not place an order, your audiences may get confused and ask for your clarification.
Use the same date/time labels for the same temporal granularity. For example, if you have used “July 2020” somewhere as a month label on a chart or in your texts, then stick to this date format in your whole presentation when you discuss monthly trends. Switching it to “2020/07” or “2020–07–01” out of sudden would require the brains of your audiences to transform that into a previously learnt format and thus create unnecessary burden for processing information.
Keep everything simple.
Though I am always impressed by advanced and comprehensive data visualisations that are able to communicate very complex trends and dimensions (like this well-known data presentation by Hans Rosling: https://www.ted.com/talks/hans_rosling_the_best_stats_you_ve_ever_seen), I always try my best to keep the data visualisation in my reports as simple as possible.
Theoretically one can visualise as many as six dimensions with a static chart using common BI tools or visualisation packages of R and Python, the rule of thumb I apply to my work is that no more than three dimensions should be presented on one chart. Remember — three is a magic number :)
Use simple and short date and number formats whenever possible. For example, if I need to present a number series ranging from 2,000 to 200,000, I prefer to use data labels “3K” instead of “3,543” in both written and verbal communication. It is true that by doing so we are unable to show easily the difference between 3,200 and 3,240 (because both are now depicted as 3K), this should not be a big issue unless we are in a very accuracy-sensitive setting, e.g., monthly accounting reporting, investor board meeting or PR release.
Summary
That’s it!
Here I attach the Analyst Style Guide I have created for this post with all examples: https://docs.google.com/presentation/d/1OuPyxbJN9mFI_awe74-fK6OtL4y3i5iUMKfQdIbsVLU/edit#slide=id.p
I would encourage you to create your own Analyst Style Guide, use it in your slides and keep improving it!
