How YC advice helped me build a Coronavirus delivery service
I currently live in a 250 square ft cabin in my parents’ garden. My dad chopped down a couple birch trees and built me an outdoor shower. When it comes to hacker mentality, no MBA in the world can compete with a 79 year old Swedish farmer.
When I emerged from the W20 batch the world was locked down by global pandemic. I worried about my parents — they live in rural Sweden, old and out of reach from existing grocery delivery services. If they don’t go to the shop they die from starvation, if they do they might die from Coronavirus. I decided to fly home to help them, and ended up starting a mutual aid group in my community. In doing that I’ve constantly found myself applying advice indoctrinated on me by YC.
Build something people want
There are 12,000 people in my parents’ municipality, and 4,000 of them are members of a local Facebook group. Before going to Sweden I posted in that group about my plans to deliver groceries to my parents and maybe other at-risk people in the community. I wanted to find out if something like this existed already, and if not, if anyone else was interested in helping out.
The response was overwhelming, this didn’t exist yet, but lots of people wanted to help. The local grocer offered to put signs up, churches and the local Red Cross were interested in collaborating. This was clearly something people wanted.
Find the 90 / 10 solution
More people than I had anticipated signed up to volunteer, so we needed a way to coordinate volunteers and deliveries. I started building a Django app with a database to store orders. I even got a logo made on Fiverr.
But as the situation in Sweden was deteriorating rapidly, I needed a stopgap solution while I built out the app. I made a spreadsheet for orders and a mailing list for volunteers. That was the simplest, lowest-tech hack I could think of.
Since many older people don’t use the internet, I bought a Swedish Twilio number that they would call and leave a voicemail with their grocery order. I’d transcribe voicemail messages into the spreadsheet and send a message to the mailing list when a new order was ready.
I also made a Facebook group for general discussions. Within a couple of days it had grown to 150 members.
Talk to users
While new volunteers kept joining the Facebook group, only 5 people signed up to the mailing list. At the same time, the spreadsheet was causing confusion on Facebook.
I decided to ask the group what they thought of the process. One person wondered why we couldn’t just do it all on Facebook, why did we need a spreadsheet and a mailing list when everyone was already in this group?
Of course, she was right. Facebook was the database. Easier to maintain than a spreadsheet or a web app, but also more usable since everyone already uses Facebook.
The current process is:
- Someone calls the Twilio number and leaves their name and address
- I (or another coordinator with access to Twilio) post the caller’s first name and approximate location in the Facebook group
- A volunteer responds to that post
- I DM the volunteer with the caller’s phone number and exact location
- The volunteer and the caller speak on the phone to work out order, payment, delivery details, etc.
After each order has been delivered, I DM the volunteer again to thank them and ask if anything in the process can be improved.
Do things that don’t scale
By using Facebook as both frontend and database, I had spent 30 minutes building what would have taken a week to build as a Django app. I don’t have to write a single line of code if we need to tweak the process, just a group announcement in plain text.
It’s not a scalable solution, this Facebook group wouldn’t work if we had thousands of users and several orders per minute. But do we even want that?
Instacart needs an app because they need to scale and take payments in order to grow their business. We don’t need either of those, so we don’t have to copy Instacart’s patterns. We can keep working in small groups where we all know and trust each other.
Scale in this context becomes a problem of scaling communication and learnings across groups. I’m a member of a Facebook group of Swedish mutual aid group admins, and I lurk in the UK admins’ group, the Help with COVID Discord, and similar Slack channels. There I often find interesting tips that I can bring to my local group, and sometimes I share my local experiences back.
The volunteer response to COVID-19 is like a giant corporation with thousands of teams, except there’s no management structure that information can flow through. There’s plenty of duplicated work, and knowledge is siloed and scattered around Google docs, spreadsheets, Slack, Whatsapp, etc. It’s not ideal, but I don’t think centralization at a global scale is an option if we want to maintain speed. What we need are better tools to propagate information between our various subgraphs. Until those tools exist we should all do our best to be active nodes in the graph, and use the edges we have to share knowledge.
Be nice! Or at least don’t be a jerk
We all have egos, we expect recognition when we volunteer to build a website or design a logo. We get upset when we find that someone else has already built the app we’re working on, or when the local government wants to co-opt the mutual aid group we created. I’ve seen groups almost implode because of damaged egos. This crisis is too important for that. Get to know yourself and your insecurities.
Launch now
It’s easy to volunteer, join an existing initiative or start something new. Do some Facebook searches around your local community, find or start an interesting project on helpwithcovid.com, build a knowledge base or search engine that indexes all this stuff (please!). Or just write a note with an offer to help and put it up on a lamp post.
If you do a grocery run for an elderly neighbor, you’ve made something people want. You’ll probably have at least one user who loves your product, and you might have saved their life!
Feel free to reach out to andreas dot s dot t dot jansson at gmail.com.
