How to Analyze Data: A Basic Guide (2023)

Data analysis is critical for all employees, no matter what department or role you work in. Whether you’re a marketer analyzing the return on investment of your latest campaign or a product manager reviewing usage data, the ability to identify and explore trends and fluctuations in your data is an essential skill for decision-making.

Unfortunately, many companies today struggle with data organization and analysis. A global survey by Splunk found that 55% of all data collected by businesses is “dark data”: information that is collected but never used. Sometimes a company won’t even know that it has collected the information. Or, the data sits there because the team doesn’t know how to analyze it.

The same survey found that 76% of executives believe training current employees in data science will help solve their company’s dark data problem. If employees understand how to analyze different types of data, the company will be able to make better use of the information it collects.

Fortunately, data analysis is a skill you can learn. You don’t need to be a “numbers person,” have an advanced degree in statistics, or sit through hours of in-depth training modules to understand how to analyze data. Instead, we’ve put together this guide to help you master some basic data analysis skills – from cleaning data, choosing the right analysis tools, and analyzing patterns and trends to be able to draw accurate conclusions and actionable insights.

Before you start analyzing your data, you need to set some clear objectives. If you don’t have a clear idea of what you’re looking for, you’ll just spend hours staring at a spreadsheet or scrolling through countless support tickets, waiting for that lightbulb moment.

Your goals will vary depending on what team you’re on, the data you’re collecting, and your role within the business:

Your data analysis is only as good as the data you start with. If the information you’ve got is patchy, inaccurate, or inconsistent, then the insights you get from your analysis will be incomplete or misleading. So once you’ve collected your data, take some time to clean it by making sure it’s consistent and doesn’t include duplicate information.

If you’re only looking at a small data set, you may find it easiest to clean your data manually in a spreadsheet. As a starting point, here are some simple things you can do to clean up your data before you start analyzing it:

If you’re dealing with an extensive data set, it’s harder (or at least much more time-consuming!) to clean that data manually. Instead, consider using data cleaning tools like OpenRefine or Talend to speed up the process. Dedicated data cleaning tools clean up messy, inconsistent information quickly so that it’s ready to use.

You could also implement a data governance strategy to set clear guidelines for how your company manages and organizes data and cut down the amount of time you have to spend cleaning data in the future. A few data governance best practices include:

Your data is clean and you’re set with a variety of tools. Now, you can start the data analysis process.

As a starting point, look for trends in your data set. If most of your data is numerical, it’s relatively easy to plot patterns on charts and other visualizations. But if you have unstructured data like emails or support tickets, you may need a different approach. Here are a few data analysis methods you can try if your information doesn’t fit neatly into a spreadsheet:

As you spot patterns, don’t assume correlation means causation. For example, if you see a big increase in social media followers around the same time you saw a huge spike in product sign-ups, you might assume that all your new users are coming in from social media. But if you look at the source tracking data in Google Analytics, you’ll see that very few people even visit your website from social media – let alone sign up for your product.

Assuming that a correlation between two things means that one caused the other is called false causality, and it is one of the most common mistakes people make when analyzing data. There’s often another factor at play that’s caused the trend you’ve spotted, so take time to gather enough evidence and make sure your insights are accurate.

If you’re finding it challenging to identify trends and patterns in your data, it may be because you’re looking at your data in isolation. You can’t spot changes over time because all you’re seeing is a single snapshot of your performance. What you’re missing is the context: how your current data compares to previous time frames.

Compare your current data against past performance to put your findings into context. But if that’s not possible – for example, if you’re looking at usage data for a completely new product feature, or you’re just starting to analyze your support performance – then you may find it helpful to look at industry benchmarks instead.

You can find performance benchmarks for different companies, departments, and industries. Often a quick Google search for “[department] performance statistics” or “[industry] [department] statistics” will uncover useful performance benchmarks. Alternatively, industry publications and research presented at conferences are good places to look for benchmark data.

For example, the Zendesk Benchmark allows companies to compare their customer support performance data against the average for their industry:

Zendesk Benchmark is an example of benchmark data that you can compare against your support team’s performance data to put your performance into context in relation to your industry.

One note of caution: If you’re using benchmark data, it may be difficult to find companies of similar size or stage to you. So remember to use these figures as a reference point rather than directly comparing your performance against those benchmarks.

When you started analyzing your data, you set clear goals and expectations for what you wanted to learn and what insights you were expecting to find. But this can lead to confirmation bias, where you’re more likely to notice trends that support your existing assumptions or hypothesis.

Keep an open mind by looking for trends or data points that go against your expectations. You should also look for outliers in the raw data. This practice will help you avoid cherry-picking findings that support your existing beliefs.

If you find anomalies in your data, you should investigate them further, as there may be a simple explanation. For example, if your marketing team sent out a newsletter, but you’re not seeing any website traffic coming through, it could be that they sent it to an internal test list, or they forgot to add UTM parameters to the links in the newsletter.

You should also look at how much outliers in your data skew your results. Significant outliers can easily skew averages in your data, so you may need to track the median rather than the mean. The median uses the middle value of your numerical data set, so it’s less skewed by outliers. Alternatively, you may need to discount these outliers from your analysis altogether.

It’s often easier to understand and interpret your data when it’s presented visually instead of in a spreadsheet. Use tools like Google Data Studio or Tableau to represent your data in charts, graphs, or other graphics so that you can clearly explain your results to other team members.

If you’re working with large data sets, don’t try to communicate too much information at once in your visualizations. Simple charts make it easier for the viewer to understand your message and the findings from your data. We’ve put together a series of data visualization tips to help you communicate your data findings more clearly.

You can also use tools like Geckoboard to display your data on a dashboard that anyone on your team can view at any time. Geckoboard’s Send to Slack feature makes it easy to share your dashboards with your team for greater visibility by integrating with Slack, so even on remote teams, you can share your latest data insights. This feature is particularly useful for keeping key metrics top of mind by sending regular, automated updates.

Like data visualization, the way you design your dashboard will affect how useful it is for your team. Our dashboard design guide will help you create dashboards that clearly communicate your key metrics and give your team at-a-glance insights into your current performance data.