[chiefalchemist]

"- Say no / only agree to what you know your team can deliver on... and always double your estimate. In the long-term reliability is valued over agreeableness."

Early in my career I read that X (almost) __always__ takes twice as long and cost twice as much as you typically estimate.

I can't count the number of times that's been true. It's also universally true. e.g. Wanna hang a picture. Piece of cake, right? Until you have to find the stud. Likely get a tape measure to get the height right, etc.

Perhaps an over-implied example, but once you start to keep track in your head of your estimate vs actual, the 2x Rule pans out quite a bit.

[fencepost]

That sounds like a good example when getting pushback on estimates: "Here's a 24x36 picture in a wooden frame, how long will it take you to hang it on that wall over there?"

After getting an estimate ("15 minutes?") come back with "OK, let's plan it out. Does the picture have hanging hardware on the back or are you going to need to get some screw eyes and wire and install some? That's a pretty heavy frame, do you have good enough wall anchors? Are you going to need to drill to put them in? Do you need to be going into studs, and do you have a stud finder? How high are we going to put this? What if I decide I want it 4 inches higher? If I come back in 15 minutes and ask why it's not up are you going to want to punch me?"

"That's why an off-the-cuff estimate isn't reliable and why we need to do some review before giving a good one. Even with that review, there's a good chance we're going to miss something at first (like the steel support beam that runs down the interior of that wall)."

See also Yak Shaving and the intro of Malcolm in the Middle Season 3 Episode 6 "Health Scare"

[mharroun]

My suggestion is when asked off the cuff always toss an estimate that is astronomically bigger then what you expect it to be. Then come back a day or 2 later and claim after further research we can do it in (2x your real estimate).

[chiefalchemist]

Agree. __Always__ start high and then you've "reserved the right" to adjust down. Else, most people will think you're trying to rip them off.

One of my TODOs for 2019 is to come up with a checklist (for building websites / basic web-based applications). There's no sense assuming the client is thinking of all the things I'm thinking of, and vice versa.

The little things always add up. For example, does the client expect 1 hour or 3 hours of edu on using the CMS?