I was listening to Jay Pierson from Pierson workholding yesterday and he said "Don't buy the machines for where you are at, buy the machines for where you are going"
One of the biggest issues you have is buying your capital equipment. If you have a job that comes along that you really want to take on but you don't have large enough machines, or the parts require 5 axis, or a 4ax horizontal, you can't take that job! And if that job will be a repeat order from a large company that could theoretically take you to the "next level" well you won't get to that next level.
You want to forecast out 3-5 years and buy your machines for that purpose. You may not have the work right now to justify a larger VMC or whatever it is, but with that extra capability you not only increase your edge over competitors, you increase your capacity. Maybe that big VMC table can run palletized work that basically prints money.
Similarly, I would not buy a 2 axis lathe again (same thing Pierson said)...why buy a 2ax lathe when you can get a Y axis lathe with twin spindles? You massively increase your thruput. Add a bar feeder and you have a fully automated cell that prints money.
And don't rely on one client to feed your business. I've seen it time and time again, company has a big contract with a Prime manufacturer or whoever, contract runs out or the deal falls through, company is shit outta luck. Diversifying is good. My first boss has been hanging on by a thread for this reason for many years.
+1 on not bidding too cheap. You can leave dollars on the table that way. Profits are already thin enough, don't do yourself the disservice!
Another thing would be tooling. Investing in tooling is good. Your return on investment is seen very quickly. So you want to buy good tooling.
One last thing...it may have worked in the past with bidding and quoting, but don't just estimate or guess. Run your bids through quotation or ERP software to generate real, accurate quotes. Your profit margin will be much more consistent in this way.