Let me start by saying that I don't see this as a Microsoft-only problem, but I'm going to use Microsoft as the example to make my point. Companies often make repetitive mistakes because they focus too much on assigning blame and shooting the executive who made the mistake, rather than making sure whatever caused it isn't repeated.
I'm going to highlight a series of repetitive Microsoft mistakes and the outcomes to showcase how the same mistakes were repeated, largely as a result of a policy that focused too much on blame and not enough on understanding what happened so that the same mistake wouldn't be repeated.
Windows 95 Launch Failure
Windows 95 was Microsoft's most successful product with respect to image. It is the only product Microsoft has had with lines around the block to buy it. However, Microsoft made two critical mistakes - one it hasn't been able to repeat and the other it has repeated regularly in the launch.
The first mistake was that given how much product was being installed, the number of problems overwhelmed its call centers. Instead of sending out a notice or bringing on board additional resources, support had callers get a busy signal to keep the wait time down. This improved the metrics of the time on hold dramatically and pissed off buyers at a near-legendary level. I actually ran into this trick at IBM one year and the folks who got caught doing it were mostly fired. That didn't happen at Microsoft.
In addition, Microsoft stopped marketing Windows 95 post launch and the end result was that negative news surrounding the product overwhelmed the new services, and pictures of lines at stores were replaced by pictures of unsold pallets of Windows 95 products. While a lack of demand and a change in executive management prevented the first problem from recurring, the second problem has recurred up until the launch of Windows 7 where Kathleen Hall, hired from outside of Microsoft, killed the practice. In other words, the mistake was repeated launch after launch for nearly 15 years and remains a practice for other Microsoft products.
Ironically, the guy who drove the success of Windows 95, Brad Silverberg, and his team were managed out of Microsoft.
Office 97 Compatibility
Office 97 was a landmark product for a number of reasons. One of the biggest being that Microsoft changed the file format, creating severe compatibility problems. The issues were highlighted early on and flagged, but the issues were internally concealed. In other words, executives at Microsoft and analysts were told the problem had been corrected when it hadn't been. The end results were massive problems at launch and the termination of many of the employees who had made the mistake.
Since that time, products from Windows Millennium Edition (Windows Me) to Origami to Windows Vista have been brought to market incomplete or with unacceptable aspects to them that senior management was unaware of. The folks making the mistakes were generally moved to other jobs or asked to leave the company, which ensured that the same problem occurred over and over again. A similar problem resulted in the only firing of an IBM CEO in the early '90s and the near-failure of that company. At the core of this problem is the apparent belief that the messenger will be shot and the inability to recognize that no matter how bad it is, the actual discovery of a covered-up problem is often far worse.
Windows NT was slated to be a Microsoft replacement for UNIX for workstations and servers. Both customer sets were unique and the winning approach that had given Microsoft its massive market dominance was "embrace, extend, extinguish." That was Microsoft's winning formula and it tossed it out with Windows NT, which seemed be the primary goal, making life easier for Microsoft. It was OS/2, a product it had initially co-developed with IBM, and updated and released. It didn't embrace or extend anything but Windows.
This was a double mistake in that it also put Microsoft's need for something easy in front of the customer's needs, and it made it vastly harder for Microsoft to determine whom the customer actually was. Ironically, it appears to not have been easier either as the linkage has often slowed both the desktop and the server side to market and created complexities in security and updating that otherwise could have been reduced or avoided altogether.
In effect, in order to create a simpler product set, it complicated the desktop product unnecessarily and made the server and workstation products harder to sell (recall Microsoft "Scalability Day"), likely adding years to the ramp and significantly fueling Linux's success as the alternative customers used to vote against Microsoft's direction. It repeated this mistake in parts again with Zune (didn't embrace iTunes) and with Windows Home server (tried to remission an enterprise product - badly).
Wrapping Up: Lessons Learned
One of the things that defines Apple's success is that it does something right once and then it repeats it over and over again. Microsoft, and it is hardly alone, is often defined by the mistakes it repeats and the inability to repeat successes. This showcases a lack of institutional learning. The first CEO I ever reported to on my first day said something to the effect that I'll make mistakes and get into trouble, but if I make the same mistake twice I'll be fired. His point and something that has stuck with me since is that mistakes happen, they are part of life, but repeating them is stupid and no firm can afford stupid.
The lasting lesson is that if you focus excessively on blame rather than understanding a problem as an institution, the mistake is likely to be repeated over and over again. If, instead, you focus on improvement and learning, the company will get smarter. In Microsoft, Kathleen Hall's Windows advertising team is actually doing this and Windows 7 is performing vastly better as a result, suggesting that the firm is finally putting in place a process for institutional learning, which is something more of us should emulate.