Telling a Story with Data | Dashboard Build Demo
1 Video View·Nov 9, 2024
In this video, Chris Dutton explains the framework for telling clear and effective stories with data, an important topic for anyone looking to improve their data literacy skills.
We live in a world that runs on data. It's how Amazon and Netflix know which movies and products
to recommend, how Starbucks manages a global supply chain, and how Uber connects drivers
with passengers in real time. But the thing is, data skills aren't just for tech companies or
professional analysts anymore. Everyone works with data to some degree, and everyone can benefit from
data literacy skills. In this video, we're covering an important topic that will help you take your
data literacy to the next level. Up to this point, we've been thinking about data visualization
strictly in terms of individual charts and visuals. Now let's broaden the scope a bit and
talk about data storytelling and dashboard design. This is another topic that we cover in depth in
our Thinking Like an Analyst course. In fact, there are entire books written on the subject,
but it's worth touching on at a high level here because it's such an important aspect of
communicating with data. You'll often see data storytelling in action in the context of ad hoc
reports and presentations or explanatory dashboards like these. Now, dashboards in general are
analytics tools typically designed to consolidate data from multiple sources, track key metrics and
information, and most importantly, facilitate data-driven decision making. When we teach
storytelling and dashboard design skills, we like to walk through a six-step process. It all starts
with defining the purpose. Similar to our data analytics framework, this involves asking questions
like who is the audience, what are their goals, and what specific questions do they need answers to.
Step two is choosing the right metrics, which is about making sure that the data you present
directly aligns with the business goals and that the level of detail is appropriate for your
audience. Step three is presenting the data effectively and selecting the right types of
visuals for the job. Step four is eliminating clutter and noise. Things like 3D formats,
excessive colors, background images, grid lines, anything that takes up space but doesn't add value
or contribute to the story. This makes me think of one of my favorite quotes, which applies quite
well here, which is that perfection is achieved not when there's nothing more to add, but when
there's nothing left to take away. Great words of wisdom in the context of dashboard design.
Step five is using layout to focus attention. This can involve using gestalt design principles
like proximity and enclosure to group related elements on the page, using attributes like size
and color to draw attention to specific patterns or data points, and leveraging common reading
patterns like z or f patterns to create a logical flow and make sure your audience sees the most
important information first. Finally, step six is about telling a clear story. This is where a
little text can go a long way and where things like descriptive chart titles and data labels
can really help you craft a narrative. So that's our design framework. Next up, I'll run you through
an end-to-end example so that you can get a sense of exactly what this looks like in action. All
right, in this demo, I'm going to walk through each step of the data storytelling framework
using an end-to-end case study. So here's the situation. You've just been hired as a BI analyst
for Maven Toys, a nationwide chain of toy stores. Your assignment is to design an explanatory
dashboard for regional sales managers to help them understand how their regions are performing.
They're going to review that dashboard each month and they need information on things like revenue
trending, product level performance, and lost revenue due to inventory shortages. Our objectives
here basically just work through that six-step framework. We're going to start by defining the
purpose, then we're going to choose the right metrics, present the data using effective chart
types, we'll practice eliminating clutter and noise using layout techniques to focus attention,
and finally we'll wrap it all up by trying to tell the clearest possible story that we can.
Now feel free to either pause the video and try to work through these objectives on your own,
or you can sit back and watch how this can come together from start to finish.
So remember it all starts by defining the purpose. Who is the audience, what are their goals, what
questions they need answers to, and how often will this dashboard be reviewed. Luckily in this case
we have all of this information in our project brief, but in reality you'd likely gather this
information through conversations with stakeholders themselves. And it might take multiple rounds of
conversations before you really nail down the scope and purpose. So in this case, who is our audience?
Regional sales managers. What are their business goals and objectives? They want to increase revenue
and minimize revenue loss due to lack of inventory. And then what types of questions might they need
answers to? Things like how is revenue trending? Is it going up or down? Which products are driving
those trends? What's performing best or worst? Are there products that are out of stock? These are
great examples of the types of questions we're going to try to answer with this dashboard. And
finally, how often will it be reviewed? Once a month. Now that we have a clear picture of the
business goals, we can start thinking about which specific metrics to include and what level of
detail or granularity makes the most sense given our audience. And a little pro tip here, creating
a table like this to map metrics against each business goal is a fantastic way to document your
thought process. And you might even add metric definitions here as well to start building out
a proper data dictionary. This is a really smart thing to do, especially if you're working with
complex or industry specific metrics, or if you're building reports or dashboards that will be shared
broadly throughout the organization. So for the business goal of increasing revenue, some metrics
we might want to track include things like total revenue, by month, by region, category, store, and
product, as well as month over month and year over year comparisons to add some meaningful context.
For the goal of minimizing loss from lack of inventory, we'd want to be able to track stock
on hand at the product level for each store and estimate the monthly revenue loss based on what
we would have expected had those products been in stock. Now keep in mind that this isn't a black and
white exercise. I'm moving through this very quickly, but it's not something that you should
expect to get right on the very first pass. In reality, you'll likely iterate and evolve this list
as you start designing the dashboard, and more importantly, start getting actual feedback from
your end users. So this takes us to step three, presenting the data effectively. Here's where we
can take all that information from step two, the metrics and the level of detail, and start thinking
about the best way to visually represent them. So for example, revenue by month, we'd likely use a
line chart to show trending over time. For revenue by region, which is probably the single most
important number that these regional managers care about, we might use something like a KPI card to
really give it some emphasis. Revenue by product category, maybe a simple bar chart. And for our
product level details, we might choose something like a table so that we can include some additional
details like the product name, category, stock on hand, revenue, and so on. And if we continue this
process, we'll end up with something like this. And note that these are all very simple visuals,
right? We're talking lines, bars, KPI cards, and tables, nothing custom, nothing complex,
but this should get the job done at least as a first pass. Again, this is another case where
you'll likely iterate multiple times before landing on the final version. So don't stress
too much about getting this perfect on your first pass. So let's see what it might look like if we
took all those visuals on the last slide and just dumped them on the report canvas. Well, we'd end
up with something that looks a little bit like this. Technically, we've checked all the boxes so
far, right? We've defined a clear purpose, we've chosen the right metrics, the most effective chart
types, but obviously our work isn't done. It's cluttered, it's noisy, there's no intuitive layout
or flow. So it's time to leverage some of those visualization and storytelling best practices,
starting with eliminating clutter and noise. There's a ton of simple things we can do here
that will go a long way in terms of simplicity and polish. For example, we could update our revenue
line chart to simplify axis and data labels, remove grid lines and borders, change the colors to draw
more attention to the current year, which might look something like this. We could apply similar
changes to the product category bar charts as well, and also use color a bit more strategically
to draw attention to positive or negative trends, like this. And helpful tip, if we want to make it
a bit more colorblind friendly, we might use colors like blues and oranges rather than red and green
like we have here. Then finally for our product tables, we might kill the borders, adjust text
alignment, or add some conditional formatting to really focus on which products are driving the
gains and losses, like so. Now that's feeling much better already, even with just a handful of
relatively simple design tweaks. With that, it's time to move on to step five, using layout to focus
attention. This is where we can start to think about how to assemble these visuals into an
intuitive layout that helps us craft a narrative and focus viewers attention on the right places.
Now in most western cultures, viewers typically start by looking at the top left corner of the
page, they scan to the right, then make their way down the page from left to right, following either
a z reading pattern, which looks like this, or an f pattern, which is similar and looks like this.
Either way, the point is to feature the most important high-level information towards the top
left of the canvas. So with that in mind, one thing we might tweak is rearranging our header to put
the dashboard title in that top left corner, and we can bump the logo, which is less important,
to the right of the page, like so. And for the rest of our visuals, let's start by clearing the deck
and see if we can improve that layout and flow, and also practice applying some design principles
to make sure we're drawing the eye to the numbers that matter most.
So given that the top business objective is focused on revenue growth, it makes sense to
put our revenue KPI cards and trending chart in that prime real estate on the left side of the
canvas. We can also play with font size and color to emphasize that current month revenue total,
and de-emphasize the month over month and year over year figures, since those are really just
providing additional context. Note that we're also using a subtle design principle here known
as enclosure to visually group these elements together, since they're all part of the same
revenue story, and we want users to focus here before shifting their attention away.
Another best practice is to start high-level and progressively get more granular as the viewer moves
through the dashboard. So with that in mind, let's pull in our category-level bar charts next,
and if you look closely, you'll see that we've actually removed the axis labels
from the revenue change chart, and align these two visuals to create continuity between them.
This is another example of a gestalt design principle that we're putting to use here.
Now we're getting down to product level, so we can pull in our top five and bottom five product
tables like so, and because everything we've shown so far has been revenue-focused, I'd like to create
a bit of separation and make it very clear that the stock and inventory-related data
is telling its own distinct story. So we could do something like this, add some more enclosure
around our table of products with no stock, and pull in that KPI card here to really emphasize
the potential revenue loss, since that directly ties into our second key business objective.
So now we're looking really good, but remember that the final step in our framework is to tell
a story, and one of the simplest and most effective ways to do this is by replacing default
chart titles with meaningful descriptive text, and using things like font colors and styles
to help users quickly interpret what they're looking at. Here's an example of what that might
look like. Now instead of just feeling like a report, this actually reads like a story.
How did the region perform in September 2021? This is how much revenue we generated,
this is the revenue split by product category, where these five products drove the increases
and decreases. It's so clear, so much easier to follow, and such a great way to help your
end-users and stakeholders cut to the chase and get the exact information they need.
Now of course the final step would be to tie it all up with a few specific insights and
recommendations, for example like launching new product offerings similar to PlayFoam,
phasing out Rubik's Cubes, restocking headphones at the JFK location,
but hopefully this was a helpful demonstration of data storytelling in action, and a great
example of what the dashboard design process can look like from start to finish.
If you enjoyed this content and want to see more, we've got a brand new Data Literacy Foundations
course, and it's entirely free. You can check it out at mavenanalytics.io. So whether you're
an individual looking to build confidence, a leader seeking to empower and upskill your team,
or a data professional just trying to stay ahead of the curve, this is the course for you.
We've got a lot to cover, so let's dive in.

