Data Storytelling 101: Do’s and Don’ts for Effective Visualizations
Hola Data Freaks😎,
With the rapid increase in demand for data analysts, proficiency in data visualization has became indispensable. As someone who has been in this domain for quite some time now I believe sharing the mistakes I made during this journey and the lessons I learned from those mistakes will definitely help you avoid those pitfalls in the path of your learning and take you one step closer to becoming a better storyteller with data…
There are many posts for what and how you can present your “observations” from a set of data (yes observations as many people confuse them with insights), but this post will tell you what all things you should definitely avoid if you want to become a successful storyteller in the world of data analytics.
CHICAGO CAR CRASH ANALYSIS this is the Power BI dashboard I recently worked on as a personal project for my Intra-Soc competition at NucleusCBS. When I completed the dashboard it looked very vibrant and the insights (along with observations I presented) were very decent, but looking back it feels this dashboard has major flaws I overlooked and hence a lesson for today. So let’s get started…
Mistake 1: Not identifying who my stakeholders are?
The Biggest mistake i made was by far presenting each and every observation/data point from the dataset onto the dashboard. Instead of allowing myself time to think about who will be using the dashboard or what purpose it would be used i.e. who the stakeholders are and what they would likely want from this dashboard, I just included every single bit of info I can possibly find from the dataset and presented them in form of various vibrant and fancy graphs.
Let’s take an example, Here in this case after getting a grips of data I must have put efforts on whom am I making this dashboard for, is it the insurance company, the government, the car manufacturers or the car repair businesses and it is only after identifying the stakeholders, I would have figured out what matrices were require to deliver the value they needed.
Key Takeaway: Always identify your audience before starting on your dashboard. This will guide your focus and ensure your work is relevant and impactful.
Mistake 2: Showing Too Much Info and Too Many Graphs
Remember the 80/20 rule i.e. 80% of the results are from 20% efforts similar is the case in data analytics, major deviations are usually a result of few business decisions. One significant mistake I made was overwhelming the dashboard with excessive information and numerous graphs. Instead of adhering to the 80/20 rule and focusing on the most critical insights, I included a plethora of charts that often showcased trivial observations.
For instance, I created multiple sheets filled with various graphs that could have been consolidated for clarity and impact. A clear example is Page 4, which although contain detailed data, could have been clubbed together for efficiency. Moreover, the animated bar charts on Page 3 represents similar information as those on Page 2 regarding the number of car crashes.
Key Takeaway: Applying the 80/20 rule ensures that your dashboard remains focused on delivering the most valuable insights to your stakeholders, avoiding unnecessary complexity and enhancing usability.
Mistake 3: Forgetting the Context
A critical mistake I made was neglecting the importance of context in data analysis. Focusing solely on the data without considering the broader context can lead to misinterpretations and flawed conclusions. For instance, I presented a pie chart showing significantly fewer accidents in the east of Chicago. However, I overlooked the crucial context that the east side has a lower population density compared to other areas. This oversight skewed the interpretation of the data.
To avoid such pitfalls, it’s essential to consider factors like weather patterns, climate, humidity, geographical location, types of vehicles, and population density. These contextual details provide a deeper understanding of the data and help in making more accurate and insightful analyses.
Key Takeaway: Always integrate relevant contextual factors into your data analysis to ensure a comprehensive and accurate interpretation of the findings. This approach enhances the relevance and reliability of your insights for stakeholders.
Mistake 4: Using Fancy Charts
One big mistake I made was using fancy charts instead of simpler ones. Fancy charts might look impressive, but they can distract from the main goal of showing important insights clearly. They often need more time to explain to managers and stakeholders.
For example, the Ribbon Chart on Page 1, despite its visual appeal, might confuse those unfamiliar with its complexity and require significant time to explain. Instead, using simpler charts like a line chart could have effectively shown trends in reasons for accidents over the years without unnecessary complexity. In short it’s better to stick with simple charts that highlight the main points clearly. The goal of a chart is to focus on what’s important and not get bogged down by unnecessary details.
Key Takeaway: Choose simple charts that are easy to understand. Clear visuals make your data analysis more effective and easier to use for decision-making.
So what’s the conclusion…
The world of data visualization can be a challenging yet rewarding journey. Through my experience with the Chicago Car Crash Analysis dashboard, I learned valuable lessons that I hope will aid you in avoiding common pitfalls. Remember to always identify your audience, adhere to the 80/20 rule, provide context for your data, and favor simplicity over complexity in your visualizations. As the saying goes, “Torture the data long enough, and it will confess to anything,” but true insight comes from understanding the story behind the numbers and presenting it clearly and accurately.
“Torture the data long enough, and it will confess to anything” — Ronald Coase
If you found this post helpful and are an aspiring data analyst or just passionate about data analytics, follow me for more such nerdy stuff. Link for the dataset I used: bit.ly/4bHKjpc
See you in the next post till then Sayonara,
Happy analyzing! :)🙌
