In today’s era of big data, especially when it pertains to advertising and finance, it’s critically important to be able to utilize the data your organization has for rapid insights and meaningful intelligence. At Even, data is the lifeblood of our marketplace operations and we dive deep into the data that we collect to articulate insights and findings, all of which help our teammates make rapidly well informed decisions. This is only possible via well curated analytical dashboards that are thoughtfully designed by our analytical teams, to help provide the maximum amount of insight possible to our operational teams.
To produce a great dashboard, we need to follow a discovery, design, and execution process where stakeholders are engaged regularly to identify key points of interest. The following need to be considered for a great dashboard:
- What business questions are we trying to answer? To keep the dashboard tidy and easy to understand, make sure the business questions are clearly defined and that the dashboard is designed specifically to answer those questions.
- Who is the intended audience? Everyone has a unique communication style, and some people will want more information than others on the dashboard. Consider the audience’s role and responsibilities, and what additional ancillary information they need before designing the dash.
- Is this expected for an ongoing, one time, or periodic use case? Dashboards will require different levels of information and visualizations depending on how often they are used. A longer term monitoring dashboard will likely require easy to understand data visualizations, whereas dashboards optimized to one time or periodic use cases may require less visuals and more raw data.
- Selecting the right visualizations for the data: Identify the right visualizations for the information you are trying to show. If you’re monitoring things over time, best to use a line chart or bar chart.. If you’re just looking to provide aggregate data, a pie chart could be beneficial!
- Keeping the layout of the dashboard crisp and logical: Focus people to specific visualizations on the dash by highlighting certain charts over others, and clearly delineating “key” metrics from “supporting” metrics. This could involve enlarging certain charts relative to others, and establishing a logical flow for the dash that walks the viewer through a data story. For example, you could create a summary section at the top of the dashboard, and then proceed to more granular levels of analysis below the summary.
- Detailed labeling and formatting: The devil is in the details with numbers—make sure units are correctly formatted and articulated on the dash, and that the dashboards have clearly labeled series and titles. This is especially important to ensure that viewers of the dashboard are consistently understanding the metrics throughout the dash.
- Ensure a clean backend and easy to understand underlying queries: Make sure that the data sources from which you are pulling your data easily support the analytics you are trying to generate. This may involve working with your Data Engineering team or Analytics teams to perform ETL to ensure that you have everything you need in an easily accessible format.
- Scope the dashboard out and break up the work into digestible chunks: While it may seem like a good idea to experiment with visualizations and produce things on the go, sometimes for a larger dashboard it’s more time efficient to produce a design spec and then break down the work into smaller chunks. This allows for rapid iteration of the design while still moving the dashboard forward towards completion.
- Touch base regularly with stakeholders: Make sure to stay in touch with your stakeholders and partners throughout the process. Dashboards should be built to answer business needs—and it's rare that business needs stay unchanging over a long period of time.
- Bake in time for final revisions: Before considering a dashboard is complete, make sure to add in a week or two for final revisions. This time will allow you to refine the metrics, visualizations, and layout of the dashboard to produce more meaningful insights.
- Test the pipeline and make sure the dashboard runs in a performant manner: Be sure to perform test runs of the dashboard and perform regular QA on it for the first several weeks after launch. Not only does your dashboard need to produce the right insights to support decision making, but your audience needs to be able to get it in a quick way. Load times of 10+ mins are highly discouraged, and at Even we work to keep most dashboard load times below one minute!
At Even, we do all this and more for internal data dashboards, when we work with our enterprise partners to produce custom reporting dashboards, and during the process of developing the Even Benchmark Report. We strive to consistently enable our partners to make agile and proactive decisions by providing them extensive insights and transparency at scale. The above methodology makes that possible—and we hope you’re able to gather learnings from it to utilize in your own data dashboards!
Disclaimer: The material provided on this site is not intended to provide legal, investment, or financial advice or to indicate the suitability of any Even Financial product or service to your unique circumstances. For specific advice about your unique circumstances, you may wish to consult a qualified professional. Any information or statistical data sourced by Even Financial through hyperlinks, from third-party websites, are provided for informational purposes only. While Even Financial finds these sources to be accurate, it does not endorse or guarantee any third-party content