4 ms·
Before Instacart was on the scene, my brother and I thought of about a similar service. My brother in his college days used to be a pizza delivery driver, so he
by l8again 10y ago
Before Instacart was on the scene, my brother and I thought of about a similar service. My brother in his college days used to be a pizza delivery driver, so he had the "real world" experience. Over the weekend, I quickly created a simple dropwizard java jar that would take in two variables - number of drivers, and number of orders. Everything else was randomized to be close to reality - such as delivery times, routes, etc. To be clear this was a bare bone, CLI application. I would run the service for 5 minutes, and collect the metrics. One thing was clear as mud, the more we scale up, the higher the loss. The thing was that my brother kept telling me that I am paying too less to my drivers, and that no one would work for me at that rate. Also, I will admit that my algorithm was a "simple Fed-ex" style hub-spoke model.
- tomcam 10y agoVery cool story. Love that you tried it, love that you're honest about it now.
- kovek 10y agoInteresting! You estimated the costs using a small program. Have you ever done this with other ideas? Is there some resources which teaches the best ways to do such estimates? I've done this before with an excel sheet for a specific business my friend was enthusiastic about. It turned out he would have been alright had he gotten a loan.
- marcosdumay 10y agoThere is a concept called "Monte Carlo Simulation" that can help on that. Pushing random numbers into a program can be as rigorous as you want. https://en.wikipedia.org/wiki/Monte_Carlo_method https://en.wikipedia.org/wiki/Monte_Carlo_method