Berkeley Farmers' Market
In Spring 2021, my team sought to utilize the skills we’ve learned over the last four years in our Industrial Engineering studies to support an organization’s operations. We kept it local - and lended our support to the Berkeley Farmers' Market, part of the ecology center.
Background
The Downtown Berkeley Farmers’ Market is a popular outdoor market for Berkeley residents and neighboring residents to purchase fresh produce and local foods. However, due to the pandemic and limited staffing, the farmers’ market faces several new and longstanding challenges. The staff rearranged the market layout to adhere to government regulations, but the new stall placement created market congestion and forced some vendors to a separate street. Behind the scenes, staff members manually updated each vendor’s certifications and qualifications, taking up time the staff members could better use for more pressing market responsibilities. Our team analyzed the lines and the flow of people through the market to help better design the market and streamline the market’s paperwork system under a new database tool.
Problem Statements
Changing the layout of the farmers’ market
The Berkeley Farmers’ Market is an outdoor space where vendors set up stalls, and customers can walk around and purchase a wide range of items. The market must adhere to social distancing guidelines, which has changed the traffic flow of customers through the market and the overall stall placement. Amanda and the other members of the market staff expressed interest in having a fresh set of eyes on the market to ensure the setup is run in the most effective way for both the vendors and the customers. Outside of the context of COVID, the farmers’ market would like to optimize space in order to welcome more vendors into the market.
Annual Paperwork System
Each year, vendors and insurance companies send certifications (organic, egg seller, bottling, etc.) to the farmers’ market throughout the year. The staff manually adds the expiration dates for several certifications for over 90 vendors on a Google Sheet. This task becomes very tedious very fast for the market staff, and we want to make the data collection process more streamlined and easier for them. We are looking at alternatives to Google Sheets, such as Airtable, or changing the system they use with Google Sheets to include a form to auto-populate cells. We will also try to create a system to easily check which vendors have submitted all their certifications and insurance policies so that staff does not need to scroll cell-by-cell to find this information. Airtable and other software will cost additional money, so we will need to double-check the market's financial constraints.
Opportunities
Changing the layout of the farmers’ market
Pandemic Layout
-
Help facilitate safe distancing between stalls and between shoppers in line
-
Potentially add more stalls to the pandemic layout
-
Better manage the market’s capacity: adjust people’s waiting areas and traffic flow through the market
Regular Layout
-
the farmers’ market directors are always looking to expand the number of vendors - increased diversity would bring more people to the market, bringing in additional revenue for the market and for the vendors
​
Annual Paperwork System
-
significantly ease the tedious workload of market staff - there are hundreds of certifications that must be emailed, tracked, & documented by hand
-
frees up time for market staff to focus on more meaningful tasks
-
automate the data collection process and make it easier for the staff to check that all the vendors have their certifications for the year and are insured
​
Assumptions and Constraints
We received most of our information from Amanda, the farmers' market manager, but also interviewed employees/vendors/customers, researched government COVID restrictions, and consulted with Berkeley IEOR faculty.
Changing the layout of the farmers’ market
Pandemic Layout
-
COVID capacity limitations (# of shoppers in the market)
-
two main points of entrance (rather than entering from anywhere) that control the inflow and outflow of people, with two employees available to manage this flow
-
people in lines required to stand a certain number of feet apart
-
may not be worthwhile with increasing vaccination rates and decreasing restrictions
Regular Layout
-
focus on increasing the number of vendors at the farmers’ market
-
agricultural vendors and non-agricultural vendors must be clustered together
-
stalls dislike being moved around, as they notice a drop in demand immediately following a change in market location
Both Layouts
-
cannot change the physical location of the entire farmers' market
-
employee capacity
-
minimal data to model market capacity: must be collected manually in-person at the market
​
Annual Paperwork System
-
no capacity to use a databases system (onboarding and budget constraints)
-
currently using Google Sheets and Google Drive, all information and documents stored there
Data Collection
Changing the layout of the farmers’ market
Metrics
-
rate customers enter/leave the market
-
# of customers that pass through every hour
-
queuing within the market
-
customer servicing times
Collection Methods
-
time lapse videos of people moving around the market to find congestion points
-
recording line lengths for every stall
-
timing service times for each customers, finding the max and min service times
Annual Paperwork System
Collection Methods
-
interviewing the operations team to understand the current workflow in Google Sheets
-
diagram workflow and download existing paperwork data
Market Layout: Findings and Recommendations
Isolate Busy Vendors
One vendor's line ended up growing too long and overflowed outside of the market. This overflow resulted in customers confusing the vendor line for the line to get into the market, instantiating a moment where the market did not allocate enough space for a vendor’s line. The farmers’ market’s operations team decided to move that vendor to the street side adjacent to the park rather than buildings, so the stall’s line was structured to overflow into the park instead of the street or market. Due to our team’s observations both before and after this transition, we were able to notice that the overall crowding of the market decreased and flow improved.
Place more vendors in "dead zones"
The area around the information booth has significant dead space. The vendors in this area did not get busy enough to justify the size of their line space allocation. Therefore, it appears that another vendor could likely be added into this area.
General vs Specialized vendors
The longest lines at the farmers' market are for farms that sell a wide variety of produce. Other stalls that specialize in tinctures or granola rarely have more than 3-4 people in line. We recommend that the busier general stalls be placed on the park side of the streets; this will allow them to be kept close together as their lines can be overflowed into the park itself. The less busy stalls should be placed on the non-park sides of the street; they can then be kept close together as they do not need large line allocation space.
Create a 2 way street
The middle central path often becomes congested with lines, people running into each other, or people walking in opposite directions and bumping into each other. The farmers' market could consider creating a 2 way street with cones or chalk so people move through the market consistent. People will experience less congestion and a heightened market experience as a product of easier navigation through the market.
Annual Paperwork System: Workflow Recommendations
Airtable ended up being the best balance of a more organized and automated database with ease of use and understanding. In order to make the transition from Google Sheets to Airtable easier, we tried to make the workflow and tables similar to their current workflow.
Vendor Table: This includes all vendor contact information and core information like whether they are an agriculture or non-agriculture vendor. On the sidebar, we set up example views with filters that employees can easily switch to (ie. Tuesday Vendors, Ag Vendors, etc.).
01
Problem: When employees requested vendors to fill out the required annual documentation, they had to send multiple back-and-forth emails and then manually update the spreadsheet and Google Drive folders with the correct data and attachments.
Solution: In order to automate the workflow, we set up the "Forms" view on Airtable which would allow BFM to send out forms that vendors could fill out. The recorded answers, which can include attachments, would automatically populate in the Airtable database.
02
Problem: The marketing manager often had to send emails to vendors with missing materials or expired certifications. With Google Sheets, this involved manually checking a spreadsheet column for expired certifications (color-coded) and then emailing each vendor about their documentation.
Solution: We implemented an email automation tool on Airtable that would send an email automatically to a vendor whenever the current date surpassed the expiration date. This automated email autofilled the vendor's email, subject line, and body.
03
Problem: Whenever a vendor received an infraction, a manager would have to send an email to the vendor and update the Google Sheet manually. Because there are a lot of infractions per farmers’ market, emailing all the vendors took a lot of time.
Solution: We set up an email automation with different infraction templates that managers could use to send out to vendors. When an infraction is entered into the table, an email is automatically sent to that vendor. Because the tables are linked, the vendor email is auto-populated.
Challenges and Learnings
Working with non-technical clients
The Berkeley Farmers’ Market team is unfamiliar with simulations, modeling, and databases. We needed to balance application of classes over the years with practicality so that the market team could understand our results and use our work. Overloading the client with too much technical information would turn our proposed solutions into problems and burdens for them.
Test data as its being collected
Our first few rounds of data, while collected accurately, ended up being the wrong kind of data, which resulted in uncertain and inaccurate models. The variable service times of customers made the queuing model yield infinite lines. Luckily, we were able to change direction before we spent all the remaining weeks collecting this data, and focus on line lengths and time lapses to provide recommendations to our clients.