Designing a Dashboard
A design task
I’m currently job hunting, and sometimes companies ask me to do a design task. Normally I don’t share, but this one was different as it was relatable to my previous experience as a merchandiser. The design task was for a company that use data and AI to optimize staff, inventory and predict future sales for restaurants and retailers. In my previous role as a merchandiser I used to have to do that in Excel, and it was a nightmare (a bit like the government counting coronavirus cases in a spreadsheet!), so I felt very passionate about it.
Design task: improve the current forecasting tool
I’m going to refer to the company as xxx. Xxx is trying to solve the below problems:
- Managers often spend hours building reports on Excel. At the end of the day, they spend more time collecting data than they spend on analysing it.
- Managers often struggle to build relevant forecasts.
Step-by-Step approach:
- Usability Test with 5 users
- Conduct User Research (interviews, survey, affinity map to discover trends and patterns)
- Persona
- Defining the problem
- Analysing the forecasting tool
- High fidelity wireframes
Given the design task’s constraints, I could not conduct a usability test and talk to users. To get around this, I search the web for reviews, feedback and observations regarding the product. I found that user satisfaction was 95%, and they had 222 positive feedback on social media and only 6 negatives.
I also found a guide on “how to use” the forecasting tool, which gave me insight into the user’s pain point and the goal. The forecasting tool’s goal is to give managers insight into the predicted sales, so He/She can motivate the team to reach their weekly sales target and approve the AI-generated forecast. From that, I also created a persona.
Persona
Charlie is a Manager at a café chain with at 7 locations.
Behaviour: He uses xxx daily to adjust sales, guest and approves forecasts
Needs & Goal: Need an overview of forecasted sales to motivate and empower the team to take action and to achieve their weekly sales targets.
Pain Point: Not enough time and find big tables frustrating.
Scenario: It’s Tuesday and Charlie are using xxx Forecasting Module to review the AI-generated forecast for the next week. He updates this based on insights, e.g. unusual weather and numbers of guest booking. He also checks xxx to see if his team will archive their weekly target or if he needs to take action.
The problem
Charlie needs a quick and easy way to see the forecast for his locations, so he can motivate his team to achieve their weekly sales targets.
Analysing the Forecasting Tool
Accessibility: When improving an online experience, I always check if the colours and texts are accessible. In this case, they weren't, so I changed the colours slightly to increase the contrast.
Visualising data: When you first look at the forecasting tool, you’ll realize there is no overview. “Print”, “Export to CSV” and the Search bar is at a very prominent place. Based on my research and assumptions, the managers don’t print/export these tables very often. If the managers' goal is to see the forecast, why is the forecast not the first thing you see? Would it not make more sense to have forecast first then how it compares to last week and last year? This is something I would have liked to test with the user.
Approval buttons: The stages of the approval button seem confusing. Which one is approved and which one requires you to take action.
The Forecasting Dashboard
Below is my suggestion of the forecasting dashboard. As you can see, forecasted sales are first, and you can easily see if the team is on track compared to previous data. Further, you also have an overview of how many approvals you have to do.
Next Step
- Usability Test to confirm/disconfirm the assumptions
- Test prototype with 5–10 users
- Explore option for implementation of new features
- Ongoing Feedback — Track increase in user satisfaction
