Yeah, time estimation is really tough. When I started I figured out I just needed to take my time estimate and double it. Then I got really analytical about it. Now I take my gut estimate and add 20%, but I've also gotten away from time-based billing into value-based estimating (that's a whole other very deep subject).
The "really analytical" phase wasn't too crazy by HN standards:
- Break the complete project in Pivotal Tracker down to stories no larger than 8 points
- Track the time spent on each story in Harvest
- Use that data to figure out how much time a given point score equated to, along with variance.
- Work to reduce the variance.
I should also say...I've been estimating software for, good god, I haven't done this math lately, 18 fucking years. Do anything repeatedly that long, your instincts become pretty solid.
The "really analytical" phase wasn't too crazy by HN standards:
- Break the complete project in Pivotal Tracker down to stories no larger than 8 points
- Track the time spent on each story in Harvest
- Use that data to figure out how much time a given point score equated to, along with variance.
- Work to reduce the variance.
I should also say...I've been estimating software for, good god, I haven't done this math lately, 18 fucking years. Do anything repeatedly that long, your instincts become pretty solid.