Finding the right numbers is only half the job when working with data. The other half is communicating those numbers in a form that is accessible and usable to people.

Suppose you've been working for hours to gather sales data, scrub customer information, and crunch monthly performance figures. Finally, you have a dependable set of data. You must now share your conclusions with a manager who is concerned about declining profits.

You develop a vibrant dashboard featuring a number of charts. All is in order, but the manager still can't figure it out.

It may not be your data. It may be to do with how you've been presenting it.

Using a chart can help make an otherwise complex pattern more understandable, but the wrong visual can obscure key elements, lead to misinterpretations or cause people to miss the problem altogether. There are a plethora of chart types in Excel, Power BI, Tableau, and Python, and it can be very important to know which one to pick.

So, how do you determine the right chart for your data?

The answer starts with a simple question: What do I want the audience to understand?

Start with the question, not the chart

A common error in data analysis is to make a chart menu first, before determining what to discover. A survey was conducted of the most popular graph options, such as pie charts and line charts, because they are often selected for their attractiveness or suitability for a dashboard. One better way is to determine purpose of the analysis before beginning the analysis.

Consider these questions:

1. Which product has the greatest turnover?

2. What has been happening with customer demand in the last 12 months?

3. Is advertising spending associated with increased sales?

4. Why do some warehouses have longer delivery times?

5. Which of the following category of expense is the highest percentage of the budget?

All of these questions are based on data but it is the type of vizualisation that is needed that that will differ.

This is a starting point rather than a strict set of rules. The final choice depends on the structure of your data, the number of categories, and the decision your analysis needs to support.

Understand what different charts are good at

1. Bar charts: When comparison matters

Suppose an online store sells four product categories and records the following monthly revenue:

A bar chart makes the comparison straightforward. The reader can immediately see that laptops generate the most revenue, while accessories generate the least.

A horizontal bar chart is especially useful when category names are long or when you need to rank many products.

Use a bar chart when you want to compare independent categories, identify the highest or lowest values, or highlight differences between departments, customers, or locations.

However, a bar chart is not always the best choice. If you are trying to understand how revenue changes every day over six months, a line chart will generally communicate the pattern more effectively.

2. Line charts: When the sequence matters

Time-based data often contains information that is difficult to recognise in a table.

Imagine that a company records monthly revenue for an entire year. A line chart connects the observations in chronological order, helping the reader identify growth, declines, seasonal changes, and unusual movements.

For example, revenue might increase steadily from January to April, fall in May, and recover during June. The line makes that change visible without requiring the reader to compare twelve separate numbers.

Line charts are useful for monitoring sales, website traffic, customer activity, production output, and operational performance.

One important detail is that the observations should have a meaningful order. Connecting unrelated categories with a line can suggest continuity that does not exist.

3. Scatter plots: When you want to investigate a relationship

In some cases, it is not about comparing the value or following a trend. Rather, you want to determine if two numeric variables are correlated.

Imagine an ad agency that is interested in exploring the link between ad expense, and sales dollars.

For a scatter graph, advertising expenditure would go on the x-axis, and sales would go on the y-axis. An observation is represented by each point, which could be a campaign or a day.

It would seem that, overall, the more advertising that is spent, the more sales that occur in the data that is observed. If the dots are dispersed, the connection might be less strong or less steady.

This visual can be used to analyze performance, pricing and demand, delivery distance, delivery cost, and customer spending or purchase frequency.

An important caveat: Causality can't be established by a relationship between two variables. Discounts, seasonality, product availability, or other factors could also impact sales.

4. Histograms and box plots: When the distribution matters

Averages obscure information.

Suppose that there are two delivery services, both of which claim an average delivery time of 3 days. Their performance at first glance seems the same.

But one service could get almost all packages delivered in 2 to 4 days, while another service could get many packages delivered in one day, and have some customers waiting a week for delivery.

That difference can't be accounted for by the average. A histogram displays the frequency distribution of data when the data are partitioned into intervals. It will help you identify the focus of observations, if the distribution is skewed or if there are unusual values.

A box plot is a simple method of comparing the distributions of multiple groups. It can show variation in spread, variation in median values and any potential outliers.

These charts are especially useful in analysing delivery time, transactions amount, processing times, customer spending and quality control measures.

5. Pie charts and stacked bars: When the parts matter

Suppose a business spends $100,000 each month across four expense categories:

A pie chart can communicate the basic proportions of the total. Salaries account for half of the monthly expenditure, marketing represents one-quarter, and the remaining categories account for the rest.

This works reasonably well because there are only a few categories and the values add up to a meaningful whole.

However, pie charts become harder to read when there are many small slices or when the audience needs to compare similar values. A sorted bar chart may be more effective in those situations.

A stacked bar chart is useful when you want to show the total and its components across several groups. For example, a retailer might compare total revenue across regions while showing how much comes from online and in-store sales.

If the main question concerns the percentage composition of each region rather than its total revenue, a 100% stacked bar chart may be more appropriate.

Similar charts can answer different questions

A chart doesn't always appear to be a perfect match for one question. There are cases where two charts look like an appropriate choice, but with each one showing a different side of the data. Consider three common situations:

1. Which chart should you use to compare the frequency of data?

Compare sales of products using a bar chart. Follow sales across months by using a line chart. A column chart might be used if you want to compare individual months, not focus on the trend.

2. Which of the following is the correct format for this? Grouped bars or stacked bars?

Use grouped bars for comparisons within categories. Use stacked bars to show a total and its components when you want to see the total and its parts.

3. Which one's better: Histogram or bar chart?

A histogram is a graph that organizes numerical observations into intervals, like $0 to $50, $50 to $100, and $100 to $150, showing the number of observations in each interval. A bar chart is used to compare different categories of data, like different products or different sales regions.

These distinctions may seem small, but they change what the audience can easily interpret.

A practical process for selecting the right visual

When you receive a new dataset, follow a consistent process instead of choosing a chart at random:

Step 1: Formulate a question.

Make a note of what you would like to find out. Are you comparing performance, observing a trend, examining a relationship or studying a distribution?

Step 2: Understand your variables.

Recognize in which columns dates, categories, numerical measures, or geographic information are found. Before visualising the data, identify any missing values, inconsistent units and unusual observations.

Step 3: Determine the relationship you are seeking to connect with.

When comparing, bars are suggested. A line is sometimes indicated by a time sequence. When two numerical variables are indicated, they suggest a scatter plot. A distribution would be a histogram or a box plot.

Step 4: Choose the simplest effective chart.

Use a familiar visual first and then an expert visual to add useful information. If the reader is required to refer to the table for exact numbers, it may be better to use a table.

Step 5: See if the message is clear.

Think about how you might be able to tell someone the main finding without having to give them quite a long explanation. Look at the title, labels, units, scale and colours.

Step 6: Check the conclusion with the data.

A visual might indicate a pattern that is not apparent on a longer time frame or if other factors are included. Ensure that your interpretation is backed up with the evidence that supports it.

Conclusion

There is no single chart that works for every situation. A bar chart may be ideal for comparing products, while a line chart is more suitable for tracking their performance over time. A scatter plot may reveal a relationship, while a histogram can uncover variation that an average conceals.

The trick is in knowing what information you can glean from the data and selecting a visualization that makes it easy to see that information. The next time you create a report or dashboard, don't get tempted to begin with the Chart menu. Begin with the question first.

What does the audience need to understand? What evidence would help them understand it? Which visual presents that evidence most clearly?

Once you've answered those questions, it'll be much easier to pick out what chart you want to use. Great charts start with great questions.

Better the questions you pose, the better the decisions you make.