Cargill Feed Data Dashboard
UX Research, UX Design, UI Design, Graphic and Brand Design
Methods& Tools
Interviews, Persona
Information Architecture
Prototypes (AdobeXD)
People
I was the sole designer on the team with two developers and one manager.
Challenges in Feeding Hogs
It is hard to believe that most of the hog farmers in the US still climb up the ladders to manually check how much feed is left in the bins every other day nowadays, which is dangerous, boring, and inefficient.
As the world leader in the agricultural area, Cargill has been trying lots of different solutions to change this situation. In this case, sensors with promising accuracy were implemented in the feed bins to record feed data. Our team's task was to design and develop a data dashboard to visualize the data intuitively to help farmers handle the feeding flow issues.
After I defined the project goal, user types, and constraints, I drafted two interview protocols for two different user types and interviewed two farmers and four people related to feed mills with another marketing intern.
There are two persona types: farmers and feed mill managers.
Focus on the Core Needs of Farmers
Considering the time limit and priority of personas, we decide to first approach the design from the farmer's end.
Know how much feed is left and when to reorder
Farmers need to make sure hogs always have food to eat to keep hogs healthy.
Make sure hogs are eating feed normally
Sometimes feed flow issues happen and feed might get stuck in the tubes when humidity is high.
Need a safer and efficient way to check bins
Farmers check the bins manually by climbing up the ladders and tapping the bins at different heights. Since many farmers are middle-aged or elderly, it is especially dangerous for them to climb up the icy ladders in winter.
Communicate with feed mills about order status efficiently and timely
Feed mills leave receipts near the feed bins, which is not so convenient and timely.
Design Challenge
How might we design a feed bin data dashboard that is informative, reliable, and effort-saving for farmers to manage the feeding process of their livestock?
Key Indicators of the Feeding Process
Based on users' needs, we identified these data which would be helpful for farmers to understand what happened inside their feed bins.
Amount of Feed Left
Percent
Visualizing the percent of feed left gives users an intuition that matches users' mental model, which is similar to what users get from their manual inspection— use tools to tap at different heights of bins.
Weight
Considering bins have different sizes, another expression is feed weight-- how much feed is left by amount.
Consumption Rate
Consumption rate is how much feed pigs consume in the recent few hours.
Another important indicator is the change in consumption rate. It is especially helpful for farmers to track if any unusual eating patterns happen.
Predicted Feed Run Out
Farmers need to know when feed runs out and reorder in advance.
Design for Farmers' Needs
Since our project timeline is very tight, we decided to finish a MVP to communicate our ideas to the clients, thus I just included three bins as an example.
I iterated from low-fi wire-framing in MockingBot to high-fi UI in Adobe Xd.
Main Page
Bin Overview
Farmers improve their efficiency by seeing the status of all bins at one glance to identify the issues and take necessary actions.
Predicted run out, available feed, and feed weight shows them how much feed is left and when to reorder.
Consumption rate and its change help farmers make sure hogs are consuming feed normally.
Alert History
The web app will send alerts via messages to farmers on phones when feed runs low or get refilled, which can let farmers know when to reorder and help feed mills communicate feed delivery status to farmers in a timely way.
Why did I use tables on the home page?
Considering that users have tens of bins (one interviewer reported that he had 18 bins), it is hard to display such diverse and rich data on mediums like graphs.
What did I do to help users identify the problems more efficiently?
Auto sorting of bins
The bins are automatically sorted based on the percentage of "available feed". Bins that are running out of feed appear on the top and bins that still have lots of feed are near the bottom of the table.
Edge case: if "available feed" is 0, they are stacked at the bottom because the bins are not being used.
Color coding
I used red for the alarming messages to help users identify the current issues quickly, such as low available feed and abnormal consumption rate change.
Individual Bin Pages
Farmers can see real-time status of specific bins and see an intuitive data trend visualization here to get a sense of hog eating patterns.
They can also see future predictions of this bin's run out dates and all alerts that have been sent before to better manage the refills.
Bin Settings Page
In order to make sense of data in the contexts of bin specifics better, users can input and adjust specifics about bins, feed, and herds in order to get more accurate data visualization.
If I don't consider technical costs, I will iterate my design this way:
1. Combine current status and bin trends modules; make a trend preview on the cards and make the cards clickable
2. Combine predictive data with the trend graph
We gave up this version because of higher costs in implementation. However, I think it combines data in the past with future predictions in a clearer way.
After Design
Near the end of the internship, I pitched our prototype at our headquarters in front of 30-50 employees. Our clients spoke highly of our outcome and this prototype provided them with insights and directions to go forward.
If we have more time, I think we can focus on:
1. Design for feed mills since they are the second persona
For the feed mills, in the long term, intelligent dashboard systems can help feed mill managers avoid urgent orders by proactively processing orders so they can optimize their production runs and delivery routes.
2. Consider pig growth cycles
- Farmers dislike storing feed since pigs eat feed with different formula during different cycles, thus investigating pigs' cycles can help farmers to be economical when ordering feed and organize feed flow better.
- Feed density varies in different cycles, thus hardcoding cycle-based information will contribute to more accurate data visualization and relieve farmers from the burden of inputting information manually.
- From research, we heard from the users that there was also a special type of bin system called double-stacking feed bins (ration 1 in bin 1, ration 2 in bin 2). Due to time limits, we didn't conduct further research. Researching how this kind of system works can help us understand users' needs further.
3. Iterate design considering the number of bins in real cases
Since in this MVP, the number of bins is limited to three. For the real product, the overview page and settings page can be modified in several ways:
- Since a typical farmer has tens of bins, some bin filters can be added. The empty bins can be folded since they are not being used currently.
- On the settings page, to make the bin setup process more efficient, they will have options including “add a new bin”, “duplicate this bin”, “delete this bin", or even import specs about bins from some spreadsheets.
.png)
.png)
.png)
