Template
E-Commerce Business Overview
Combine your Shopify, GA4, Google Ads, and Facebook data for insights across your business.
.svg)
Make sure you avoid these common mistakes when visualizing data as well as some best practices to follow in 2024
As someone who’s been doing data visualization for over 10 years, I’ve come across so many mistakes people make. Make sure you avoid these common mistakes when visualizing data as well as some best practices to follow in 2024. The most common errors include:
Whether you're building a dashboard, presentation, or more, it's important to be wise with your data visualization choices.
Here are some examples of each so you can learn to avoid them.
“Which graphs should I be using?”
As a general rule of thumb:
There’s many more than this, but those are the main ones. To learn about this in greater detail, read our post about data visualization techniques.
Not all data can be visualized into graphs or charts. For instance, data pertaining to employee details: including first & last name, email address, ethnicity, job title etc.
The biggest mistake would be to present the raw data like this:
Just because a dataset contains a bunch of qualitative data like "name" and "email address" doesn't mean it can't be visualized.
There are two ways to visualize it:
Card view is good for visualizing raw data:
1.webp)
Gallery view is good for visualizing data with images (for instance: employee headshot photos). An example of gallery view is FlixGem.
Both of these visualizations aren't just to make things "look nicer." But they allow you to easily filter through the data with interactive tags. This is important for both data analysis and presentation.
It might look complicated to create, but existing tools make your job dead simple:
Pie charts are best used when there are 2-3 items that make up a whole. Any more than that, and it’s difficult for the human eye to distinguish between the parts of a circle.
Notice how it’s hard to distinguish the size of these parts.
Is “China” bigger than “Other”?
It’s hard for our eyes to tell the difference. Instead, replace this with a bar chart:
Notice how “China” and “Other” are far apart, but we can easily distinguish that one is larger than the other? That’s because our eyes are more sensitive to length of bars than parts of a circle.
Bar charts will be your go-to chart for data visualization.
It might look pretty, and you might be wondering “what’s wrong with it?”
The more colors you use, the less comprehensible the visualization will be. More colors = more categories the brain must process.
On top of that, there’s a better way to handle colors:
Colors allow us to highlight whatever information we want.
If we wanted to highlight the country with the biggest CO2 emissions, we can use red vs. grey:
Notice how China immediately sticks out and we get the point across.
Other times, it’s a good idea to use multiple shades of the same color.
Pie charts are best used when there are 2-3 items that make up a whole. Any more than that, and it’s difficult for the human eye to distinguish between the parts of a circle.
Horizontal bar charts suffer from the same issue as pie charts: once there are too many categories, you run out of space to include text and it becomes hard to digest:
Instead, it’s better to use vertical bar charts (by switching the axes around):
This gives unlimited space for including text and is easier for the brain to digest.
Here’s an example of someone trying to include too much information on one chart:
Including too much information ruins the point of data visualization in the first place. The purpose of data visualization is to allow the audience to easily digest the information and this graph does the opposite of that.
Instead, take the time to rearrange your data and create multiple graphs to convey your point.
Studies have shown that 3D rendering can negatively affect graph comprehension. It might be tempting to be creative and ‘3D’ your graphs, but there are better ways to get creative.
Sometimes it’s okay to break this rule, but in general:
Since the y-axis doesn’t start at 0, it’s easy to fool someone that product 2 is failing, but in actuality:
The same applies to other graphs like time series:
Since the y-axis doesn’t start at 0, it’s easy to fool someone that the price of something is exponentially rising where in actuality, the increase is only about 10-20%.
Spreadsheets and pivot tables with no context are meaningless.
Look at this pivot table:
It's a pivot table showing which product line and gender are generating the most income. Even though it's ordered from highest to lowest income generated, what exactly do these numbers mean? How high is $1580?
These numbers are meaningless without context.
Instead of just giving a raw number, it's highly recommended to provide a mean deviation, that is, how far a number is from the average:
Now we can look and go "Oh $1580 is 23% above the mean."
Creating these might be off-putting to some people since it takes more time and effort, but a tool like Polymer Search does all of this automatically for you - and creates pivot tables faster than Excel.
The aesthetic aspect of data visualizations is undoubtedly important. But it should not come at the expense of digestibility.
Take the graphic below, for example.
There's no doubt that the artist spent a lot of time creating this piece. But there are a handful of issues that leave much to be desired.
For one, the visualization follows no organizational structure or order whatsoever.
At first look, users may think that the headers in each "slice" are the names of the apps that produced the data. In turn, some people might think there's an app out there called "#LOVE" or "Americans."
Some slices don't even have a header at all. Only upon close inspection, which takes a couple of minutes of the audience's precious time, will they realize that these headers aren't what they seem.
The visualization also uses some questionable color pairs (just take a look at the "Airbnb" and "Twitter" sections).
However, out of all these issues, the biggest problem is that the creators decided to build a single graphic for multiple, inconsistent data types.
Remember, popular data visualization formats like pie charts, graphs, and tables exist for a reason — and that is to help readers comprehend data faster and more effectively. For that to work, you need to start with a consistent, clean dataset that comes together to tell a cohesive story.
If the graphic above is meant to bombard the user with a mash of large numbers, you can say that the creators succeeded.
In data visualizations, a single setting can substantially change the way users interpret the data.
Charts that don't start at zero are a great example, and they are often used to mislead the audience.
Another example is the chart below:
You might think that interest rates soared from 2008 to 2012. After all, the bar for 2012 is several times higher compared to the one for 2008.
But if you read the scale, you'll know that interest rates actually only increased by a tiny 0.012% in four years.
The bar chart above is created to highlight the importance of the Y-Axis. By using an extremely minuscule range, the differences between the numbers are greatly exaggerated.
Here's the same data with a more reasonable Y-Axis range:
Pie charts with too many categories are bad, but at least they aren't deliberately misleading.
Charts that flat-out misrepresent the numbers with unproportionate visuals, however, are a serious offense.
Take a look at this chart, for example:
The visual may look cute, but the numbers don't make sense.
For instance, the 38.5% slice is roughly twice the size of the 31.0% part. It's also misleading that the 31%, 17.1%, and 7.2% slices are very similar in size.
If we were to guess, the creator may have built the data visualization manually and failed to use proportional sizes for the data.
With a data visualization tool, you don't have to second-guess the graphical proportions of your data. You simply choose a data visualization type, plug in the numbers, and watch the software render the graphic for you.
Granted, this version doesn't have bees and fancy shading. But it does a much better job of communicating the accurate and proportionate composition of honey.
Not to mention that it took less than five minutes to create this graphic using a data visualization tool.
Here's another good example, which is an automatically generated pie chart using pre-loaded values on Polymer:
This time, the graphic only took a few seconds to create. That's because Polymer's drag-and-drop visualization tool instantly creates anything — from column charts to pivot tables — using values from a connected data source.
You've already seen Y-Axes that don't start at 0 and use uncalibrated ranges.
The chart below uses a different tactic that can skew how viewers interpret the data.
First off, notice that the "government funding" bar is smaller than the "revenue" bar in both years. That's despite the fact that the government funding in both years exceeded $1.2 billion, whereas the revenue bars are only supposed to represent $490 million and $573 million.
Misleading, right?
Technically, the visualization didn't lie. The problem is, the Y-Axis scale is truncated from $700 million all the way to $1.7 billion (that's $1 billion jump).
While the chart itself is correct within the scale, the gap between $700 million and $1.7 billion made the "government funding" bar a lot smaller than it is.
Now, we're in no position to claim whether this is intentional or not. But, if you're creating charts, never use a truncated Y-Axis to prevent viewers from misinterpreting your data.
Whether you're creating a horizontal or vertical bar chart, never mess with your scale.
In the example below, we used Polymer to visualize the total number of leads generated per keyword in a Pay Per Click (PPC) campaign:
Notice how the X-Axis scale starts at 0 and uses a consistent, reasonable range. This gives a much clearer picture of how much better "bubble inventory download" is than other keywords in terms of generating leads.
This next data visualization is pretty interesting — and not in a good way.
As you can see, the graphic is meant to visualize the frequency of consumers doing gardening work.
There's absolutely no reason for this not to be a bar or pie chart. But instead of using a popular, more readable type of visualization, the creator decided to use a unique graphic for no good reason.
Another problem is the header "less often."
Less often than what?
Bear in mind that the visualization is based on a survey. That means people either had to respond with "less often" or something else more meaningful, like "less than once a month."
If it's the latter, why didn't the visualization just say that?
If you think about it, the story behind the visualization above is rather straightforward. It's just a survey on how often consumers do gardening work — and the job is simply to visualize the results and make the data easier to digest.
Here's an efficient way to do this:
Apart from using a crystal-clear bar chart, the label "less often" is also replaced with "less than once a month." This eliminates any ambiguity from the visualization's message.
3D and data simply don't mix — as mentioned in #6 above.
But the NYTimes back in 2008 decided that 3D rendering doesn't make a chart confusing enough.
Can you guess what type of chart is actually being used here?
If you guessed "bar chart," congratulations — you're just like most people.
But most people guessed wrong.
In fact, the chart above is a list of several pie charts.
Look closely and notice how the charts bend inwards in the middle. That's the center point of the pie chart.
Here's the thing: the data can be adequately explained with either a bar chart or a collection of pie charts. Even a simple bulleted list would've effectively conveyed the message.
The problem is, they used 3D pie charts and tried to arrange them like a typical bar graph. It's one bad design choice on top of another.
This Polymer-powered table offers the fastest and most efficient way to convey the information above:
Not only is the visualization clean and readable, but it's also interactive.
Users can filter out professions they're not interested in reviewing. Additionally, the data can be sorted alphabetically or based on their prestige rating.
Here's the same data represented as an interactive bar graph:
Unlike the original chart from the NYTimes, this graph features plain, flat bars with a value at the center. Everything is the same shade, but the graph still does an immensely better job of conveying information.
Before you look at the next visualization, take a deep breath and prepare for a little headache.
So, what do you think is going on here?
The funny thing is, data visualizations are tools to ease the interpretation of data. They're not supposed to be brain teasers that leave viewers with more questions than when they started.
It might take you a while to piece this puzzle together. To save you the trouble, here's what's happening in the graphic above.
The full green bar visualizes the "business should take responsibility" value. That's the percentage of respondents who believe the companies should be responsible for the items listed on the left (create jobs, drive innovation, support local communities, etc.).
The blue line in the middle marks the actual performance of the business in each expected responsibility. For example, when it comes to creating jobs, 50% of businesses are actually doing well.
Lastly, the shaded part of the bar to the right is the performance gap. In simple terms, that's the difference between the performance of businesses and the expectations of the respondents.
Ultimately, the goal of the visualization above is to compare the public's expectations with the actual performance of businesses when it comes to specific issues.
Rather than stacking both metrics in one bar, just create two bars for each.
That's why this version is far superior to the mess above:
A cumulative chart has its uses, but it can be misleading when measuring certain values.
For example, check out this cumulative annual revenue chart:
The line is going up so the business must be doing well, right?
Keep in mind that cumulative charts can only go up. If you look really closely, you'll notice an almost imperceptible slope that indicates a slowdown.
Instead of cumulative data, this chart tracks the YOY revenue of the business:
If you're the business owner, this visualization is scary — the complete opposite of what the cumulative chart makes you feel.
Regardless, the YOY is the visualization you need to see. It shows the reality that your business could be well on its way to shutting down, giving you the opportunity to diagnose the problem and possibly turn things around.
You might think that using raw, unadjusted data leads to accurate analyses.
While it's true in some cases, especially with smaller datasets, "unadjusted" means you're completely ignoring the importance of data quality.
In other words, you might end up with a chart like this:
The chart above is meant to represent yearly recorded temperatures. But since the creator carpet-bombed the plot area with blue dots, it's impossible to use the data for insights.
There's a reason why data cleansing is important.
Biases, errors, outliers, and incomplete data can all lead to faulty readings. For one, the chart didn't organize the data by season — a huge factor that affects temperature records.
How can you tell if the dots above 82 degrees aren't due to heat waves (which are, by definition, abnormal)? Should viewers just assume those values were logged during summertime?
That's why this chart, which segments the data by season, is a lot better in helping the audience understand what the data is trying to say.
A lot of data visualization mishaps could've been avoided if the creator only used the appropriate chart type.
Flexible data visualization and Business Intelligence (BI) tools like Polymer are available. There's no need to forcefully use a sankey chart for a dataset that's best visualized as a table, bar graph, or even a set of pie charts.
To give you an idea, here's a quick look at the data visualizations you can add to your Polymer dashboard.
Every visualization tool you need for just about any dataset is here.
You have zero reason to experiment with an incompatible chart and end up with something awkward or borderline unreadable.
Data visualization can be incredibly powerful when done correctly. Not only does it make complex data more understandable, but it can also reveal patterns, correlations, and insights that might not be visible otherwise. Here are a few examples of good data visualization practices:
While there are many ways to effectively visualize data, there are also common pitfalls that can render a visualization confusing or even misleading. Here are some examples of what to avoid:
In the context of the current content, it's also essential to emphasize the importance of choosing the right type of chart or graph for the data. For example, while bar charts and pie charts have their place, they aren't always the best choice. It's crucial to understand the nature of the data and the message you want to convey before selecting a visualization method.
Once you learn the many data visualization techniques, know when to use each graph and become aware of all the good and bad practices, you’ll be a pro data analyst in no time!
There are many ways to enhance your visualization skills - with Polymer Search you’ll be able to instantly generate interactive graphs/charts/pivot tables in a matter of seconds. You simply upload your data and the AI will automatically turn it into an interactive spreadsheet and provide quick and easy data visualization tools that are available in no other tool.
You’ll also be able to create your own web app in a couple of minutes with no coding experience required. Simply upload your dataset and Polymer will automatically transform it into a web application where you can share all your visualizations. Unleash Polymer's AI to unlock more insights in your data.
It’s braindead easy to use!
See for yourself how fast and easy it is to uncover profitable insights hidden in your data. Get started today, free for 7 days.
Try Polymer For Free
.svg)
.webp)
.webp)
.webp)
