Typically if they have any business sense to go with their estimating knowledge, they"ll either expect to be able to buy in to the company, or at least expect a share of profits. Failing that they"ll start their own company. But that"s not really any different from a good project manager. That"s just how construction is. Despite hearing about massive companies like Halliburton, PCL, Fluor etc etc they typically don"t do a whole lot of the actual work. They just oversee it. Most of the actual work is conducted by much smaller subcontractors, most of which are privately and/or family owned.
Our company has only had 3 main estimators in the past 15 or so years. Myself, my bro before me, and a guy before him that had bought in to the company and then left to take over another one (we still do lots of work with him as he took over a sheetmetal/HVAC company and he"s easily the best to work with, but he literally taught me and my bro what we know about estimating and he"s good at playing the games inherent in it).
Our estimating is 90% by hand. Most of the time spent is "taking off" the material, equipment etc. We literally draw out isometrically most of the systems that can"t easily be counted on the plans. Once that"s all done the bill of materials is input in to a 20 year old estimating program that applies material prices and labor values to everything. We play around with discounts, input quoted prices for the equipment, fixtures etc, take our best guess at what the labor will be (80% of what the labor book says? 110%? Fuck if I know!), add in a bunch of other misc. costs like accommodations, permits, travel time etc in to the software package and that tells us what the plumbing number is. Then we do a spreadsheet with our plumbing number and the various subtrades with the various markups.
It"s a very manual, paper oriented process. The biggest fuck ups happen within the estimating software, typically from input/keystroke errors. That"s been the biggest thing training a new estimator to take over most of what I was doing, is hammering in to him that a single wrong keystroke in the "right" place can be tens or hundreds of thousands of dollars. As an example a few years ago I botched the pricing for some pipe we were to use a lot of on a job. Instead of $11.80 a foot, I had input $1.18 a foot and didn"t catch it when reviewing the pricing summary. It was about an 80k hit on the job (total contract with subs and markup was 9m). Thankfully I also goofed up on the price of some expansion tanks, with the total being 64k instead of the intended 6400, so that offset most of the error, and it was in the salad days of 2007 when we were marking things up double or triple what we are today.
We"ve got a new estimating package that we bought and I received training for 5+ years ago. We"ve never moved over to it because it attempts to automate a lot of the manual take off process,so it requires a ton of initial implementation and for your own sake you should use it in parallel with your old system for at least a year or two to ensure it"s set up properly and giving you consistent, correct results.I just never had the time to bid every job twice using two very different methods, but hopefully with the new guy learning fairly well we"ll be able to give it a shot, and in the end it should reduce the time it takes us to estimate a job by 1/3 to 2/3. But it does so much shit "under the hood" it"s fairly terrifying to contemplate what could happen if something goes awry.