One of the most popular tasks a data analyst or data staff at a company handles is creating dashboards. Actually, this is not only limited to working at a company. One of the major and initial skills you acquire when getting into the data field is building a dashboard. However, creating a dashboard to practice the skill you’re learning could be really different from being asked by your manager to set up a dashboard for the team. While learning, the data might have been given to you in CSV with specific questions to answer; but in a typical workplace, all these might not be readily abstracted for you and you have to source for what you need.
In this article, I will share twelve steps that you can follow to successfully set up a dashboard for your team without much worry. This piece was inspired by someone I know who reached out to me for tips to set up her first dashboard for her team at her new job. I understand that this seemingly “simple” task can feel pretty overwhelming and nerve-racking, especially if it is at a bigger company.
12 steps to set up the best dashboard for your team
- Identify required metrics: The first step in building a dashboard is identifying the required metrics needed for this dashboard. You should draw up a comprehensive list of these metrics and run it by the relevant stakeholders before anything. You can develop this comprehensive list from the initial list of metrics given by the stakeholders if you got that, or by setting up a meeting with stakeholders and asking questions to identify pain points and needs, or by joining business and product strategy sessions to identify the primary goals of the organization at the time.
It could be difficult, sometimes, for stakeholders or teams to articulate clearly what metrics they need, so you can help them identify that by asking guiding questions. Some of these questions could be, “What data points do you refer to most frequently?”, “What metrics would you like to check every morning when you wake?”, “What metrics do your most important decisions as a team rely on?”, and if the team is customer-facing, you could also ask, “What features or processes do users complain about most or seem most confused about?”. These questions can be tweaked depending on the team and can also be followed up by a couple of “whys” to really understand why certain metrics are important versus others. - Find the sources of the data: After identifying the key metrics and additional metrics you intend to visualize on your dashboard, you need to obtain the data for these. This could mean that you speak to the data engineering team or follow the organization’s processes to get access to the database you need. It is important that you take this step quite early because the process of accessing and collecting all the data you need might take some time, depending on the organization’s processes and policies.
- Think of additional metrics from the data that could tell a story or show more insights: Now that you have the data you need, take some time to think of additional metrics that could give added nuance or insights and tell a more well-rounded story. You don’t want your dashboard to just be a dump of numbers and visualizations, you want it to tell a story, in such a way that whoever views it can flow with and follow along without any added explanations.
- Choose a tool: The next step is to choose the tool for creating the dashboard. This might be dependent on your organization’s approved software, the software with the features that you need, or one that you are most comfortable with and skilled at. I think that it is important to be flexible at this point, prioritizing the features you need for your dashboard.
An example is when I decided at one of my previous roles to use a certain visualization tool because it allowed for embedding on a website and automatic refresh every thirty minutes. This was important to me for that dashboard because the data was dynamic and changed frequently, accurate data as real-time as possible was very crucial, and I wanted it embedded on a web page. Some tools you can use are Microsoft Power BI, Tableau, Metabase, or Datawrapper, among others. - Choose appropriate viz types for each metric: After deciding on the metrics you would like to visualize, and picking the right tool for your dashboard, the next important decision is the visualization types to represent these metrics on the dashboard. Is a certain metric better represented as plain text? Or, better as a line graph? Or, is a table more appropriate, a bar chart more appealing, or a combination of more than one viz type? In choosing this, prioritize legibility over complicated or sophisticated visualizations. There are certain types of data that are better represented with specific visualization types. For example, line graphs are best to represent time series, forecasts, or progression over a time period. Keep these in mind when choosing your charts and try to use the most intuitive charts for each metric.
- Design dashboard: Relating to the previous point, you’ve chosen the charts and visualizations you hope to use for your different metrics. You probably have even created these charts in different worksheets and saved them. It is now time to design the dashboard. When designing your dashboard, you should consider size, color, readability, and flow.
You don’t want different charts in the dashboard to be out of proportion, too big (as this could be overly distracting or feel like you’re screaming), or too small (as this is hard to read, is inconvenient, and can cause errors in interpretation of the data). You also don’t want to use colors that don’t match each other, are too contrasting, don’t fit in with your organization’s brand colors, or are not accessible to people with visual disabilities, color blindness, or low vision. Readability is also important. There is no need to create a dashboard, putting all the data and great charts you’ve created if people cannot read what you present. This is inclusive of people that use screen readers as well. It is good practice to always title or caption your visuals, as this gives context to each chart.
Finally, flow, although it cannot be seen directly, involves the user’s perception and viewing experience. You want to tell a story that all connects together. You are not just dumping random charts and numbers all over the place. You want to carry your audience along. You want to give the relevant metrics and data needed, with additional insights, in a concise, resourceful, and relevant way. - Connect data and create visualizations: I personally prefer to design my dashboard, at least have a wireframe of the dashboard, before creating any of the visualizations. This is because I use this mockup as a guide when creating each visual. How big is this meant to be? What colors can I use for this? Should I make this horizontal or vertical? This can be greatly affected by its position on the dashboard, so I use the mockup of the designed dashboard to influence these individual design decisions.
You also want to connect your data to your visualizations at this point. How is your data going to be uploaded and connected to the charts? Are you making API calls? Are you connecting it to a local or remote database? Are you uploading the spreadsheet files manually? It is more advisable to automate this process to reduce redundancy and possible human errors. - Test reliability and consistency of data on the dashboard: After connecting the data sources and creating the visualizations, you want to cross-check and confirm the integrity and reliability of the data in the visuals on the dashboard. Common errors could come from misspelling a table or field name, a field having some missing data that was not accounted for, or assuming a wrong data type for a field. If this is dynamic data, that is updated after a time period, think of the possible formats this data can take with each format, or possible cases/formatting that might be slightly different from what you have currently, and try to write edge cases that will catch these formats without breaking your dashboard.
- Present the dashboard to relevant stakeholders: At this point, the dashboard is ready for presentation to your team. You can reach out to the team to schedule a meeting, either in person or online, to present the dashboard, explain the metrics and visuals on it, and talk through the reasons behind the different charts. Feedback is very important, so you can leave some time at the end of this meeting for questions, comments, and suggestions. You can also give an additional number of days for the stakeholders to look through the dashboard personally, test it out, and give any feedback they might have.
- Take feedback and iterate: After receiving the feedback, you can either go through it with your line manager to decide on which changes should be implemented or not. You don’t have to make every change that was requested, as some changes might be conflicting, or even incorrect. However, it is important that you explain carefully and clearly to the stakeholders why the suggested change is not the best action to take.
- Check the connection of data after a while: If it is a running dashboard, set routine reminders to check the connection of the data frequently to make sure nothing has broken or there is no missing data. You can also work with the data engineering team to set up logs and error reporting for any possible system or data failure relating to the dashboard.
- Measure the team’s usage: Finally, a vital step to take after all this is to measure the team’s usage of the dashboard. This is very important in measuring the impact of your work and can come in handy during performance reviews, when asking for a compensation raise or promotion, or when trying to prioritize different work tasks that you are responsible for in the future.
Conclusion
In this article, I have shared twelve tips to follow to set up your first dashboard when you join as a data analyst on a new team or organization. These tips work through getting the task from the stakeholders to the planning phase, implementing the dashboard, and then measuring performance.
I hope this article has been helpful to you as a new data analyst who is faced with this first task. You can also reach out to me on Twitter, LinkedIn or send an email: contactaniekan at gmail dot com if you have more questions or if you have any suggestions for more data or career-related posts.
Thank you for reading.
Aniekan.
Aniekan's blog 