Qualitative Interviews
Prior to attending Cornell Tech, I was farmer in Upstate New York for many years mostly raising sheep, pigs and sometimes cows. I had my own business Four Legs Farm that was later bought out by Maitri Farm when they wanted to add livestock to their vegetable farm. I have worked for the National Young Farmers Coalition and am involved in the local chapter — the Hudson Valley Young Farmers Coalition.
All of this is to say, changing how the food supply chain work and creating new opportunities for small farmers is personal. I feel incredibly lucky to get to interview my friends as experts. Thank you for all of your support already.
Quick recap: Our research project is building a model for many farmers with diverse products to collaborate to serve institutional buyers like hospitals or schools.
Before we started on collecting quantitative data, I wanted to make sure we were asking for the correct data and trying to solve the correct problem. I needed to talk it through with other people to make sure the problem was clear enough that we could start collecting data to build a mathematical model. So far, I’ve done five real interviews and had a few conversations where I scrambled to take notes after the fact.
The interviews were all very loose. It was mostly me asking people what their planning processes look like and then explaining what I was hoping to do. Everyone was super generous with their time and data.
Some common themes emerged for both farmers and buyers. Both rely heavily on their own historical data. As a farmer, this was definitely my experience. My first year or two managing my farm were extra nerve-wracking because I just didn’t have a good way to ballpark what to expect is terms of my grazing plan (how much grass will my animals eat?) or hanging weights (how much meat can I expect to get from each animal?).
Farmers will gladly self-evaluate on what they are good at growing and raising. Optimizing the product mix is important for diversified farmers and something that is harder to do in a CSA or farmers market where customers expect a wide spread of vegetables.
It will be harder to get comprehensive data from dining services than I had originally anticipated. There is not the incentive for them to be keep records of quantities of food purchased every week in a tidy, orderly way. We’re never going to be able to get every order, every week. This will definitely introduce some more uncertainty into how we build our model, but now I’m curious how we could extrapolate enough to get a rough idea. Once we have users, how can we make sure the model is getting better over time?
However, an exciting development is that one of the dining services general managers was able to save 8% on food costs by buying directly from farmers and cutting out middlemen. I had hoped we could keep costs the same but had assumed that the differences in scale between the local farms and the distributors would be too great for there to be any money saved by buying direct.
Prasenjit wanted some dummy data to play with last week to get started on the model while we were waiting for folks to get back to him with their spreadsheets. I said to him that I could give him my records, but the formatting is not consistent from year to year and it is probably not labelled well enough. My data looks crazy and that’s kind of embarrassing. It feels really vulnerable to share your yields which often end up being tightly tied to your budget.
Prasenjit kindly reminded me that this is exactly the point with collecting raw data from real people. It’s going to be messy and noise-y. So now, my raggedy spreadsheets are sitting in a shared Google Drive folder and it’s okay.
Thank you to everyone who has already taken the leap and sent us your data. We appreciate it. If you want to join the spreadsheet party click here and let us know if you have any questions. We will never share your data.
