Experiments Are Fun
Quick Recap: Empire State of Food is a two-sided marketplace that connects local farmers with institutional buyers like schools and hospitals. It is a way to decentralize the food supply chain, create new opportunities for farmers and reduce carbon emissions in our food system.
We’re a team of masters students at Cornell Tech in New York City. I used to be a farmer in Upstate New York.
New semester. New teammates. New adventures.
For our third and final semester working on Empire State of Food, Prasenjit and I get to add two new teammates. Daniel and Paula are both data scientists. Data scientists are better at statistics than software engineers and better at software engineering than the statisticians. It is so flattering that they approached us to work on this project. I am so thankful for all three of you.
This semester is going to be a lot about testing and experimenting. Our first test was a landing page experiment. This is really just to check to see if people are interested enough in the idea to engage with it and to get some new contacts for our pilot later in the semester. We also asked for pricing information about veggies in a survey, but we know that this data isn’t really granular enough to give us a good idea of the kind of prices that people expect. We blasted it all over social media with this video, my co-star is Daniel.
Results of our landing page experiment:
- 160 views
- 11 survey submissions
- 4 people interested in the pilot
- 5 follow up interviews
- 1 dinner invitation from farmer friends I haven’t seen in two years
Not bad for a three-day campaign for a project that most people had not heard of before! We had been pushed to get some kind of commitment with the landing page experiment (usually it’s some kind of pre-order). But, since we’re still figuring out who is participating in the pilot on a case-by-case basis, we wanted to avoid making promises we couldn’t keep. This is still a good sign that people are excited about it.
It has also been very helpful to have Daniel and Paula around to question the assumptions we’ve been making (and therefore what exactly we should be testing). Prasenjit and I have been explaining this to each other for a year and it is really helpful to have fresh eyes take a look at it.
For example, trust is going to look different for Empire State of Food than it does for other two-sided markets like Uber and AirBnB. In those apps, we are matched with strangers because we need a ride right now or have plans to travel on a specific weekend and we decide to trust them based on ratings and reviews.
Empire State of Food is different. We’re building ongoing relationships between farmers and buyers. We need to build our platform knowing that both sides of the market are likely to have each other’s phone numbers and maybe get invited to the same parties. It would be easy for them to meet in person while doing a delivery, exchange contact info and then take their business off of the platform for next week’s delivery. Trust and feedback between people who know each other might look different than just ratings or reviews. They’ll be collaborators.
This is also potentially two different matching problems, both long-term and short-term. Ordering vegetables for this week is a different problem from planning your menu for the next six-months or year. Our pilot will be testing the short-term matching because it will be a one-off delivery. We’re treating the short-term matching as our minimum viable product.
Now all my brain wants to do is figure out how to hack together experiments. Like could we just start taking orders and then run to the farmers market to fulfill them? Could we run a second pilot in New York City with urban farms? What if we added the farmers and buyers to a group text to see how they communicate with each other? I need to remind myself that I still have other schoolwork, it’s too much fun.
