|
The friction of getting started with Mito is something we spend a lot of time focusing on. For example, when it comes to the installation process, not only do users install Mito through a CLI, but because JupyterLab 2, JupyterLab 3, and Jupyter notebooks all support extensions in different ways, there are different installation commands that users need to run to get it working for their specific environment. Initially, we just gave users instructions in our docs about which commands to install for which environment. Now we've built a completely new Python package, the mitoinstaller package, that handles the entire installation process. It downloads Jupyter if they don't have it, detects which version of Jupyter they have installed, runs the installation commands for their JupyterLab version and Jupyter notebooks, and finally starts up the Jupyter server with a tutorial notebook. In the success case, users run two commands and then 2 minutes later have already imported data into their first Mito spreadsheet. That initial friction reduction is important to our target users, who I would describe in two buckets: 1. Target open source adopters. These users are beginner to intermediate Python users that want to / need to write Python for data analysis. Most of the open source users that adopt Mito are already on their Python journey -- we're not teaching them what Python is or what a notebook is in the vast majority of cases. Many of them have gone through Kaggle courses, taken a couple data science classes at school, or are particularly enginuitive. For those beginner users, and even for people like me who have written pandas code for a few years, some things are just much easier to do in a spreadsheet interface, like creating a pivot table or graph (two of our most popular features) 2. Decision makers at large enterprises responsible for moving their company from Excel to Python. Much like us, these decision makers think a ton about the friction of getting employees started with Python. In most cases, they set up JupyterHub (https://jupyter.org/hub) so users don't need to go through any installation processes themselves, and they control things like version controlling, turning notebooks into reports, etc. They generally also offer/require Python training courses, provide template notebooks, and have data scientists available to help the business end users when they get stuck. |