The concept
This tool helps you decide whether spending time to optimise a recurring task is worth it. It is the return-on-time-investment question, made concrete: if I spend an hour automating something, how long until I have earned that hour back, and what's the most I should spend before the optimisation stops being worth doing?
The mathematics
The calculation has three inputs and one output:
- T, current time per task
- F, how often the task happens (converted to a daily frequency)
- P, the period over which you care about the savings, in days
Efficiency improvement
Efficiency is a percentage, E. A 60% improvement means the optimised task takes 40% of its original time.
Break-even formula
The maximum time you can defensibly spend on the optimisation is:
Where T × F × P is the total time the task consumes over the period, and E / 100 is the share of that time the improvement returns to you.
The matrix
The grid shows the maximum optimisation time across two axes:
- Rows, how much time the task takes today
- Columns, how often you perform the task
Cells fade out when the optimisation would take longer than the savings it returns. A faded cell means the answer is not worth it.
Tech stack
Backend
- Python with Flask, served by gunicorn behind nginx
- Just three routes, the home page, the about page, and a small
/healthzprobe used by the deploy script
Frontend
- HTML, CSS, and vanilla JavaScript modules (
engine.jsandscripts.js) - All calculation happens client-side, the matrix updates as you move the sliders
- The visual design borrows from kipjordan.com, the same palette and typography family
About me
I'm Kip, a data and analytics consultant in financial services. I built this tool because the xkcd 1205 comic is one of my favourite pieces of practical advice, and I wanted an interactive version I could point colleagues at when they asked whether some bit of automation was actually worth the time.
I'm studying a Master of Data Science at RMIT alongside the day job. More work lives at kipjordan.com, and you can reach me on LinkedIn.
Inspired by xkcd 1205.