How I Estimate Time

A key aspect of working with other people is being able to accurately estimate how long it will take me to do something. I'll be in a meeting with my collaborators, talking, brainstorming, and making plans when someone will say: "I like this idea, how long do you think it will take you to do it?"

Usually, at this point, a hypothetical completion date with no factual basis will pop into my head. Over 95% of the time, this estimate is radically optimistic.

Some part of me must want to impress my collaborators with an ambitious timeline. Another part of me thinks that I'm an inexhaustible superbrain with no other commitments. However, when I state a ludicrously optimistic timeline, my collaborators usually nod as if I'm being reasonable. I've just unwittingly set a trap for my future self.

The trap is sprung when I don't meet the timeline I've set for myself. I feel bad, get stressed, sleep poorly, and get tired. When I'm behind, I have a hard time giving myself credit for the progress I'm making. When the date arrives and passes, collaborators remember my public commitment and get irritated. The net result is that I'm more stressed, less happy, and slower than if I'd just publicly stated a realistic deadline.

This outcome is both silly and completely avoidable. I could have said a realistic number to begin with and avoided the trap. All I had to do is accurately estimate my time.

Supposedly, some people are good at estimating time. I doubt this skill is innate, and I'd like to be better at it. So, I've started training.

Strategy 1: Scaled Guestimation

This is more of a heuristic than real strategy. I take the first time estimate that pops into my head and double it.

Usually, my gut objects to this number. It's much longer than I'm really expecting.

I've noticed that if I state this new number, sometimes my collaborators give me funny looks. The time estimates are clearly longer than they are expecting. Maybe this is because my collaborators are also serially optimistic. Or maybe they've secretly been doubling my time estimates too.

Still, I'm pretty sure this technique improves my estimates. Occasionally I over-estimate my timeline. The major problem with this heuristic is that without validation, I still don't really know if my ability to estimate time is any better. It's a quick fix that avoids the hard work of truly knowing my own abilities.

To really get better at estimating time, I'm going to need data.

Strategy 2: Predict and Validate

This approach takes significantly more effort. It involves constantly estimating and testing how long it takes me to accomplish my work. I've developed some rituals.

Every morning, on the commute to work, I estimate/set goals for how much I'll get done in the day. I write down about three to six goals. At the end of the day, on my ride home, I write down what I actually got done (goal or no) and record how many of my daily goals I accomplished.

I do the same thing on the scale of a week. I predict on Monday and confirm on Friday. It's actually a surprisingly enjoyable ritual. Occasionally, I even attempt some month long predictions.

I've definitely gotten better at setting realistic goals. I've learned to adjust my predictions based on the amount of time I'm locked into meetings and other obligations. And most importantly, by writing down my progress every day, I am more able to give myself credit for actually getting things done.

I still wouldn't say that I'm good at predicting my time. I suspect that takes years of training. I have yet to get more than a handful of month-long predictions correct. (The priorities of my research seem to change drastically every two weeks.) Nevertheless, I heartily recommend this approach to anyone who's interested in learning more about his or her own capabilities.

How good are you at making short-term predictions about your own future?