A software lesson I learned in 1984. 

First a bit of chronology. 

1. In 1983 my company shipped ThinkTank for the Apple II. 

2. Then we ported that codebase to produce ThinkTank for the IBM PC, shipped it early in 1984. 

3. Later that year we released our first Mac product. It was called ThinkTank 128. It was a completely new codebase, based on a new simpler data structure. I hoped to get a fresh start, with a completely memory-based product. It would be fast, and eventually support more features because the new foundation was cleaner and simpler. However this had the unfortunate side-effect that it appeared to users as if we had removed features from the product. Because ThinkTank 128, which was only available on the Mac, came after #1 and #2, it was called ThinkTank and it had less features than the previous releases.  

I didn't expect there to be any overlap in users. I don't know why, but that was a big mistake. Many of the people who bought our Mac product in the first few months were people who had already bought our Apple II and IBM PC products.  

And they were pissed. They paid the same amount of money, later, and got less.  

And then the product just died. Sales disappeared, until we could get out a version of the product that had most of the features of the other two, and added some new twists. Even then the product didn't get on a good footing until a year after that, when we came out with a product that did a lot more than any of our previous products. But I learned a lesson I would never forget. Even when you aren't actually taking features out of a product, you can't appear to the users to be doing that. They will make you pay for it.