Use Data Visualizations Effectively

Hugo Cardoso
Data Engineer

Graphic drawing of different kinds of graphs

Data visualizations are the key to translating data into valuable business insight. These days, due to the volume and complexity of data, choosing the right data visualization can be difficult. When choosing the correct visualizations, you want to create a cohesive data story making data easy to read. Today I discuss how to choose the correct visualization to support a specific customers requirement and provide real business insight. 

Chart Categories Concepts 

We’d like to present a quick definition of the common chart types used for displaying data and their use:

  • Comparison – Compares values between two attributes over time or comparing items. (Bar Chart/Line Chart)
  • Relationship – Using two or more attributes, we can explore their correlations. (Scatterplot)
  • Distribution – Based on a frequency, it shows how the statistical data is grouped in a single or multiple attributes.  (Histogram/Boxplot)
  • Composition – The attributes can be represented over time or statically in terms of a size or area as part of proportion as a whole. (Stacked Bar Graph/Pie Graph)

Challenges & Solutions to Build Meaningful Data Visuals 

When a Data Analyst is building a dashboard, they need to put themselves in the position of the client. Generally, the objective of a dashboard is to make it easy for the client to interpret information for their specific goals. Two different clients may need to use the exact same data set, but for completely different outcomes. Below we discuss some pitfalls to avoid when choosing the correct visualizations.  

Common Pitfalls

Here are some common pitfalls with attempting to deliver a cohesive set of data visualizations:

Poor or noisy data – Noisy or bad data can ruin a good narrative. While not necessarily a long-term data visualization fix, there some levers which can be pulled on the front end to present data in a more meaningful way.

Most data visualization tools can filter out noisy data which hinders the end-users ability to gather insight. This can be implemented through data visualization tools and directly using data value or outlier filters. Reduction of noisy data results in faster insight as well as a more purposeful descriptive, prescriptive, or predictive story which is being told through imagery.  

Wrong choice – Choosing the right visualization is crucial for a clear decision-making for your audience. You cannot assume that they are an expert in statistics, because your data could be useless due to the wrong choice. 

Mock-up and brainstorming sessions with the client should be included in the project timeline so they feel involved in the creative process. Choosing the right visualization is a critical element in communicating your data effectively on dashboards and reports.  

Too much information – Visualizations that are large and complicated can be difficult for your client to digest; especially when you are dealing with large numbers. Keep it simple: this is the key to creating user-friendly visualizations. 

Visualizations that are overly complicated and difficult for the audience to digest can lead to bad decisions or incomprehensibility. Keeping the visualizations and presentation of data simple can be the key to getting direct and valuable business insight.  
If the data is naturally complex, try splitting out the information across multiple visualizations or alternatively, use drill-downs to answer follow-up questions. 

Misrepresentation – Contextualize your information so your audience can interpret your numbers in the big picture.

Inconsistent Scales – Always use a single scale when you use multiple variables to avoid confusion. 

Choosing a Data Visualization

The following chart, developed by Dr. Andrew Abela, serves as a decision tree to selecting a data visualization depending on the data types and narrative goals. You can use this as a tool in the design phase of your next data project. 

Chart to choose a Data Visualization

So how do you figure out if the visualization you chose is not working effectively? Generally speaking, this is discovered during the mock-up process of the dashboard and UX phase of data projects.

Client feedback is typically collected around how affective the UX and visualizations are with regards to answering the business questions. If during the interactive mock-up process, the client can’t understand the narrative or answer their key questions using the information presented, then it’s back to the drawing board on visual elements or UX. Additionally, during the UAT phase of the project 

Creating a Data Story Step by Step

At Indellient, we are committed to working with our clients closely in order to understand their business. This helps us build a unique data story, which becomes a meaningful part of their decision-making process. As such, our projects routinely incorporate the following phases to meticulously build out meaningful data stories:  


There are two pieces of the design phase which have a monumental impact on the development of data visualizations. Those are: 

  1. Obtaining the top-down business goals and value propositions of the data initiative. This helps us understand what audience will be using the narrative (who’s reading the story) and the type of decisions it needs to support (how and at what level). This information is key to define the user experience (UX) for each persona.
  2. Design a set of interactive data visualization mock-ups based on the chart decision guide presented previously. We will use the collection of aims and value propositions collected previously to validate (with the audience) the effectiveness of the narrative before implementation.  

    Mock-ups also validate whether the information being presented supplies enough answers to typical follow up questions to the decision-making process. It’s important for the mock-ups to be interactive so the audience has a general sense of how they will experience the data. From there they can start supplying feedback before the implementation begins.  


Implement the requirements and what was defined previously during design sessions. As Gartner’s research, this part of the project usually takes 70%~80% of the entire project depending on how complex the data is. With that being said, the following phases should be considered:

  1. Source Mapping Matrix  Identify the dimensions, filters, attributes and metrics based on the data sources. 
  2. Database Model – Should follow the subject-oriented schema to accommodate what was mapped on the matrix and also follow the front-end tool limitation. For example, some tools don’t have support on snowflake data schema. 
  3. Data Engineering – This is the data pipeline implementation step where you can apply some data transformation based on business rules and data quality cleansing rules to ingest and support the earlier designed data models. 
  4. Unit Test – It’s just a runtime execution to see if the data is ingested as expected into the target environment. 
  5. Data Analysis – A dashboard or report development is the final expedited deliverable of this step. It is the consolidation of what was aligned with the customer in the design phase.

UAT (User Acceptance Testing)

Typically entering the UAT phase means demoing the data visualisations and gathering feedback. Afterwards we can make additional touch-ups to how the narrative is being communicated. 

It serves as a key step towards the fine-tuning of the information being delivered and the user experience. It’s important at this point for users to be hands-on with the visualization tool and start providing feedback. At this point poor or noisy data can also be eliminated using front-end capabilities.  
The starting UAT phase serves as an important first reaction to how well the product created works. The product should support the business goals and value proposition that were established at the beginning of the project.

In Closing

While the information presented only scratches the surface on this topic, we hope it may help you on your next data project when designing dashboards or reports. 

Need help designing and developing effective dashboards? Contact us directly to get started. 


Here below some sources where helped prepare this article. You can also get additional information about this subject:

Have Indellient Clean and Organize your Data

The Indellient Data Team has lots of experience with cleaning and organizing data. We also can create custom dashboards and provide ongoing reports providing important Business Insight.

Get in Touch Today

About The Author

Hugo Cardoso

Hello, I am Hugo Cardoso, Lead Data Engineer at Indellient. My 10+ years background expertise lies in designing, building, and integrating raw data from various resources and technologies to a conformed analytics environment. Making data more useful to our customers’ enterprise and matching with their business goals. And also I have my Bachelor of Science in Computer Science from Federal University of Rio de Janeiro, Brazil.