How To Chart Your Data Discoveries

Data can be an extremely powerful tool to make good decisions, but if it’s used incorrectly, it can lead to inaccurate insights and incorrect actions. With the right BI tool, citizen data scientists from a range of business teams will be given direct access to datasets to look for new insights.

There’s a lot of room for confusion and misunderstanding, so teams need to put processes in place to make sure the right people are discovering the right insights at the right time. 

Our team has compiled some tips to serve as a guide for the exploration, charting and storytelling process of data discovery. Citizen data scientists can use this information as a primer for the analytical process and a set of instructions to get the most valuable discoveries from their dives into data. 

1. How to Do Data Right 

To some people, data is a new language. They’re not quite sure how to translate the information in a way that makes insights easily apparent. For others, they’re eager to dive in and curious to explore, but reluctant to communicate their findings because they don’t have experience aligning data with operations.

These new citizen data scientists have a lot of value to add to the analysis process, and access to relevant data is the first step. They can add learnings from their business experience to draw unique connections and answer new questions, which will ultimately lead to more value for their companies. 

For this value creation process to happen smoothly, the new business-focused analysts need to be confident that they’re utilizing data properly. To empower these new data dives, we have put together a collection of tips to make sure first-time analysts are doing data right: 

Analyze the Right Dataset 

Before you start digging, you want to make sure you’re digging in the right place. Select a dataset with a preview that includes a description of the dataset, the fields it contains and some sample data. It’s a good idea to have a conversation between dataset creators and citizen data scientists to agree on a way to name and describe the datasets to ensure that analysts are always able to find the dataset relevant to their analysis. If you already have an idea of the individual data points you’re looking for, be sure to check that those expected fields are there. 

Ask Data Questions 

The right way to analyze data is by asking specific questions, and you’ll want to start simple. You’ll get to the meaningful information at the end of the process, but you can’t skip steps and look too far ahead or you might miss a key insight. Here are some good examples of starter questions: 

• How many leads did we get last month? 
• Which company pages have the most traffic? 
• What were this quarter’s sales compared to last quarter? 
• Which product generates the most revenue? 

Start your analysis with a chart that answers a basic question. If you’re looking for leads in the last month, select a dataset with a column that will display total leads, filter your information for the last month and you’ll have an answer. 

Ask Data Follow-Up Questions 

Once you’ve answered your basic questions, start thinking of related ones. If your original question was about leads generated in a month, ask follow up questions like: 

• Did leads this month increase or decrease from the previous month? 
• Which campaigns are creating the most leads? 
• What percentage of these leads converted into sales?
• What is the geographical split of our leads?

For each new question, chart the results and see if anything jumps out at you or prompts a new line of questioning. Not every new question is going to result in a valuable insight, but the deeper and broader you search, the better you’ll understand how your business operates. 

Every new analyst brings their own unique set of experience and business knowledge to the process of data exploration, so intuition will lead different analysts down different paths. For any question, there is a large range of variables that could factor into the result, so start by looking at connections that make sense to you. After you’ve done that, explore different hypotheses to see if you can uncover different insights. A lot of questions require more than one chart to answer and some results might pose new questions rather than provide answers. A valuable data analyst learns something from each new dive into the data and never stops discovering new things. 

Understand What Your Dataset Can Answer and What It Can’t 

One of the best reasons to ask your dataset specific questions is to set clear expectations around the answers you will find. 

Part of developing an analytical mindset is to start thinking in terms of the tools available to you and determining questions you can ask. You might not be able to ask data about an optimal price point, but you can look at different prices that have representative samples to search for a maximum profit. There might be more factors to consider in an experiment like that, but you won’t find definite answers unless you’re asking the right type of questions. 

Use Data to Answer Questions, Not to Prove a Point 

For a lot of business users who are new to data analysis, this is a big one. There are a lot of ways to misunderstand, misrepresent, or misuse data. If you come into an inquiry trying to achieve a certain outcome, you won’t always get a valid result. When you don’t look at a question objectively, you run the risk of not only supporting an invalid conclusion but also overlooking an insight that could add value. As questions get more and more complex, they often require analysis from many different angles before real insight is found. You could miss the knowledge found through the discovery process if you’re taking a shortcut to a certain answer. 

An Insight Isn’t Complete Unless It’s Charted Correctly

The value of a modern BI tool is that you can create charts that allow users to easily move from data insights to visualizations that can be shared with other important stakeholders. Even if a citizen data scientist performs their exploration correctly and finds a meaningful discovery, the process breaks down entirely if that insight isn’t visualized in a way that others can easily read. A chart is a vehicle that will operationalize insights, so it’s extremely important to use them in a way that best showcases what has been discovered from data analysis. 

To assist with the chart selection process, the next section of this guide will cover the right way to choose a data visualization and input your data to accurately uncover and share valuable insights. 

2. Selecting the Right Chart 

Once you’ve done the analysis and found insights in the data, determining the right visualization is crucial to telling the right story. Even if you’ve done the work of selecting the right dataset, asking the right initial question, following up with the right additional queries, and using your personal expertise to pinpoint valuable information, all of that analytical work loses value if the charting process is done incorrectly. 

Choosing the right chart doesn’t need to be intimidating. Like the exploratory phase, you’ll get to the right answer by starting with the right questions. The difference is that, in the charting phase, the questions are closed-ended. 

The First Question 

It’s a simple process and it always starts the same: what are you showing? There are four basic ways to answer. Here’s a short summary of the possible responses:


Comparison charts are used to display values of two or more data points side-by-side. They are an easy way to pinpoint the highest and lowest values in a dataset. These charts can show a snapshot of several items at a single point in time, changes to a single item over a period of time or even track the progress of several items over a period of time.

Use comparison charts to:

  • Rank products by quarterly sales revenue
  • Show monthly sales trends for a single product over the course of a year
  • Show monthly sales trends for a series of different products over the course of a year


Distribution charts visualize the range and density of data points. They can be useful to get a general sense of a dataset through averages, illustrate trends or patterns and outliers. Use distribution charts when the goal is to combine a large number of data points to show a more generalized picture.

Use distribution charts to:

  • Illustrate total sales revenue by age cohorts
  • Establish a correlation between cost of a purchase and number of products purchased


Composition charts illustrate how a whole is created from its parts. These charts can be broken out over time or some other segment to illustrate how individual components change relative to each other.

Use composition charts to:

  • Show how much of the total budget is spent by each department at a company at any specific point in time
  • Illustrate how certain revenue sources contribute to total revenue over time
  • Illustrate how marketing resource allocation has changed over time


Relationship charts show how two or three factors change relative to each other. They can be used to show a positive or negative correlation, which can lead to a better understanding of what may be influential and thereby actionable.

Use relationship charts to:

  • Show a correlation between customer size and revenue
  • Map a customer base by revenue and satisfaction rating

The Follow-Up Questions 

After you’ve determined what you’re trying to illustrate, use that answer to ask appropriate follow-up questions to determine the correct chart type. Below is an explanation of the right line of questioning for each of the four basic answers to the original question. 

Comparison Charts 

Start by asking whether you want to compare a measurement among items or a measurement over time. If you choose to compare items, you’ll need a bar chart; you just need to determine whether your chart will show information vertically (one or few categories) or horizontally (multiple categories). A bar chart is a great way to compare the magnitude of values, with the option to rotate it horizontally to make many items more easily readable. For cases where the number of items may be very large, you can easily limit the number of items to the top 10 or 20. 

If you’re showing comparison over time, there are three possible charting options, all with time moving forward along the x-axis. You can show how a single data point changes over many periods of time (line chart with a single line), how a single category changes over a few periods of time (vertical bar chart) or how multiple categories change over a few periods of time (line chart with multiple lines).

comparison chart for Sisense Analytics

Distribution Charts 

Start by asking whether you want to illustrate the distribution of one or two factors. If you choose a single factor, you can display the information in a vertical bar chart with that factor shown on the y-axis. If you’re showing two factors, you’ll want to use a scatter chart, with the two factors displayed on each axis and potentially a trend line if it adds to the narrative.

Distribution chart

Composition Charts 

To show composition properly, start by asking whether you are showing information from one period in time or information that changes over time. If you’re showing a static snapshot, you need a pie chart. 

If you’re showing information that changes over time, there are two questions you need to ask. Start by examining how many periods you’re comparing, is it many or few? If it’s only a few, you want to use a stacked bar chart. For many periods, use a stacked area chart. 

Regardless of whether you use a stacked bar or stacked area chart, the second question is whether or not your information displays relative differences or absolute differences. A shortcut to this answer is that percentages mean relative differences and raw totals mean absolute. If you’re showing relative differences, a single item can only increase if another decreases. A chart that displays relative differences is most easily made by converting each item to a percent of the total, so you want to show this information with a 100% stacked bar chart (for few periods) or a 100% stacked area chart (for many periods). If you’re showing absolute differences, you can use standard stacked bar charts (for few periods) or stacked area charts (for many periods).

Composition chart for Sisense Analytics

Relationship Charts 

How many factors are you mapping? If it’s two, you can use a scatter plot with the two factors on the x and y axis. It you’re mapping three, you can use a bubble chart with two of the factors on the x and y-axis and the third factor will be shown by the size of each bubble.

Relationship chart

3. Telling the Right Story With Data 

Once the data has been analyzed for discoveries and the correct charts have been selected to accurately display that information, there’s still one final step before that data can turn into value for a company: telling a story with it. In order to make sure their insights are translated into business value, they’ll need to tell the stories of their discoveries in an accurate and compelling way. 

The tools that will help citizen data scientists tell their data-based stories are charts and dashboards. Charts are depictions of individual trends in data. Dashboards are collections of charts, tables, and number overlays that combine to show a larger picture of the data. In other words, charts are the individual sentences, dashboards are the entire data story. 

Here are some tips to more effective visual storytelling with charts and dashboards.

Tips for Telling the Right Story With Charts 

• Include descriptive titles on charts. The data in a chart will tell the story, but you have to set the stage with a title before that can happen. A chart that shows an up-and-to-the-right hockey stick is valueless without knowing whether you’re looking at the profit for a company or complaints to the support team. A good rule for chart titles is to be descriptive. Labeling the visualization with the name of the variable or variables you’re showing is usually a good idea. When you think about the chart as part of a bigger dashboard with other charts, it is especially clear that the information is labeled correctly. 

• Avoid clutter. The point of a chart is to illustrate a specific phenomenon or trend clearly. Having extraneous information in that visual is going to distract from the intended narrative or even block people from seeing it altogether. Before finishing a chart, examine it for information that isn’t needed, even taking into consideration the axes and chart infrastructure. 

• Does the chart illustrate the insight without explanation? Between the title of the chart and the trend that is illustrated by the data, the insight gained from that information should be clear. Keep in mind that once charts are made, they might be shared or consumed by other people who won’t have the benefit of reviewing the information with the chart’s creator. Charts will also be included in broader dashboards where the information will need to speak for itself to tell a bigger story, so the discovery should be clear without much investigation. 

• Advanced tool: mixing series types. Bar, line, and scatter charts can be mixed and matched to allow you to put multiple pieces of information into a single graph. Use series types to show the difference between unique categories of data, such as a line and a bar for Y1 and Y2 axes, or a bar and a scatter to show the realized revenue and projected revenue. 

• Use color as a tool, not an accent. Color can be crucial to the way a reader understands a chart, so don’t be afraid to use it as part of your story. Consider the following chart, which displays a change in daily active users. It shows a pretty clear picture of volatility.

Change in Daily Active Users

Now consider the same chart, colored differently, with green to show increases and red to show decreases. Adding this color shows a different story — not just that the daily active user count is volatile, but that the gains and losses come in multi-day chunks. The story shifts from showing instability to questioning the cause of this pattern.

Change in Daily Active Users 2

Tips for Telling the Right Story With Dashboards 

• People read left to right and top to bottom; use the order of information to tell your story. Consider the beginning (initial inquiry and anchoring points), middle (follow-up questions and hypotheses) and end (conclusion and recommendations) when putting together your dashboard. Remember that these dashboards should show not only the insights you discovered but also how you got there. 

• Putting charts side by side will change how they are read. Be very considerate when placing two charts next to each other. Putting a chart showing an increase in revenue over the last few years will have its conclusion changed when it is placed next to a chart showing an even greater increase in costs. This can be an advantage if the goal is to show profitability, but to a detriment when the costs should not be considered in the analysis. 

• You’re building a narrative, not just displaying information. The dashboard is your primary storytelling device, so consider the entire question/answer narrative that guided the exploration and insights when making a dashboard that will tell that story. Keep in mind that the first piece of information people see will influence subsequent charts and overlays. For example, look at the two dashboards presented below. They contain the same information, presented in two different ways. The chart on top shows a long-term growth pattern without a focus on the recent loss of active users. The chart on bottom puts attention on the 15% decrease in active users, making readers question what has caused this recent trend.

Change in Daily Active Users 3
Change in Daily Active Users 4

Start Exploring with Sisense for Cloud Data Teams 

With data discovery for businesses, your team’s vital business decision-makers can access data instantly. With greater access to prepared datasets, your company can improve overall data literacy and continue evolving to be a more data-driven organization. 

If you want to see how your team can use Sisense for Cloud Data Teams to increase the value of citizen data scientists, watch a demo or set up a free trial today.

Watch Demo Start a Trial