Retail & Market Research Dashboards 101
A guide to designing effective, efficient, and engaging dashboards
These days, we have in our thumbs access to tons of data points about customers’ buying behaviors and habits. The real challenge posed to us as analysts and retail professionals is presenting this information in an effective, efficient, and engaging manner. Effective, as the analysis should ultimately address specific business objectives. Efficient, as data needs to be updated in a timely manner, otherwise they run the risk of being outdated and stale. Engaging, to encourage regular use of the dashboard.
Business intelligence dashboarding is pretty common among organizations, and has helped analysts visualize data in a more engaging manner. Its interactive capabilities have helped professionals in the retail, consumer products, and marketing industries explore the data more easily. Static reports, often presented in powerpoints, are not only grueling to put together but can also be very difficult to scale. Dashboards, when planned and executed well, makes for a more efficient way to present huge datasets.
As a data analytics freelancer, I deal with these types of dashboards on a daily basis. I find that the key to creating a successful dashboard is planning it well. This article can help you with that, by breaking down the process into simple steps.
8 steps in designing your data dashboard:
- Determine the dashboard user/audience
- Set the key business objectives
- Determine the level of data granularity
- Decide on the performance benchmarks
- Design the data implementation
- Select the BI tool
- Create the mock-up design
- Test and iterate
Step 1: Determine the dashboard user/audience
Who will primarily use the dashboard? Answering this question will determine the level of detail required, and the language to be used in titles, labels, and captions in the dashboard. If it’s the CEO or any top executive, your dashboard should mostly consist of high-level components that tell something about the bottom line (e.g. sales this year versus last year, and the most profitable brands this quarter).
On the other hand, when users are involved with actual business operations, more details are to be expected (e.g. Which specific SKU or product variant contributed to the decline this year? Marketing executives would probably be more concerned about which consumer groups are buying their brand or the competitor’s brand).
Step 2: Set the key business objectives
It is critical to set the business objectives that the dashboard intends to answer, otherwise, you run the risk of putting together something that would not be useful. There is also a big risk of putting too many KPIs than what’s necessary. As essentialists would put it: “when everything is important, then nothing is important.”
With this, I summarize some types of dashboards that I have built in the retail and CPG context, and the corresponding objectives that they address:
- KPI / brand health metric trackers
They contain performance data on key business indicators such as total sales, profits, costs, customer lifetime value, etc. Apart from the absolute figures, these dashboards communicate performances of these metrics versus a certain benchmark. It can often be a time-driven benchmark: “How is our brand doing this month compared to last month?”, or a competitive benchmark: “How is our product selling versus our key competitors?”
This type of dashboard is very common in the industry — it’s probably around 90% of all the dashboards I have created for my clients so far.
- Data exploration dashboards
Sometimes the users would need to cut and slice the data to get patterns about their customer base. This is very common in marketing — “To which consumer demographic does my brand appeal to the most?” or “Which products are usually bought together?” which helps a lot in product bundling decisions.
- Optimization and Other Advanced Analytics
Pricing and promo optimization dashboards are common examples of these. “How much sales lift did the promotional campaign bring?” or “What types of discounts were effective to bring in more customers?”
- Self-service dashboards
This is the most detailed dashboard among the four. It primarily consists of tables that the users can interact with, and most of the time be able to export the data for further use.
Step 3: Determine the level of data granularity
There are some trade-offs to consider when setting the level of data granularity (How detailed can the data be drilled-down or filtered by the user?).
The first consideration would be the analysis requirement anchored on the business objectives. Some more sophisticated analytics would require up to user-level or transaction-level data granularity.
On the other hand, the two things to consider when having too granular data would be the speed of dashboard loading, and data privacy issues. All of these should be taken into account when planning.
All of these can be weighed and assessed, and compromises can be reached depending on the technologies to be used (which I will discuss more in step 6).
Step 4: Decide on the performance benchmarks
As discussed in step 2, KPI dashboards include benchmarks to assess the product’s performance. I cannot overstate the importance of this step as it may ultimately decide if your dashboard is going to be useful or not. Benchmarks provide context to the numbers and charts that you see. Without these benchmarks, numbers and charts would become meaningless and not actionable.
Step 5: Design the data implementation
Careful designing of the data pipeline is critical, especially if you intend to design a dashboard that automatically updates with new information at a desired frequency. It is best to brainstorm with a data engineer or an ETL (Extract, Transform, Load) developer about the technologies to be used, the frequency of updates, and other specifics for these steps.
Step 6: Select the BI tool
There are tons of options for dashboarding tools, depending on your needs and budget level.
As a freelancer, I mostly use Google Data Studio. It is free, has a lot of great visualization features, and integrates well with Big Query, Google’s Cloud data warehouse. It also integrates with most of Google’s digital marketing suite like Google Analytics, Search Console, and Ads for free.
Tableau and Looker are two other great options and are worth investing in for more sophisticated visualization and data integration capabilities.
For advanced data practitioners who are not afraid of coding, Python and R along with the libraries Plotly and Shiny are also good alternatives for more flexibility and customization.
A lot of websites offer a comprehensive comparison of features and pricing schemes of these dashboard services. Choose a couple of great product alternatives, reach out to their support team for a demo, and avail of a free trial if offered.
Step 7: Create the mock-up design
It’s time to design the dashboard!
A simple sketch pad or whiteboard can help with layout ideas, but based on experience doing a first draft of the dashboard using actual data is a much better mock-up.
Pro-tip: A lot of dashboards easily offers customization of themes. Take advantage of this feature to make the color palette and other aesthetics consistent with the company’s branding.
Step 8: Test and iterate
The ultimate test for the dashboard is to have the actual user interact with it and actually use it. Get specific feedback from the user, from data presentation to the layout and aesthetics. You are likely to get more valuable feedback as users adopt the tool more regularly.
Data dashboards can definitely aid swift decision-making when planned and executed well. Again, do not forget to: identify the key dashboard user, set the specific business objectives, determine the data granularity, set the performance benchmarks, plan out the data implementation, choose a good BI tool, create mock-up designs, and finally, test and iterate.
In my next posts, I will further break down the steps in making your own dashboards, using some of the tools that I have mentioned in this post. Stay tuned for more!
