That Was the Plan: The World Did Not Cooperate
Vista was ill-timed, it should have been 12-24 months later, but they thought they had to do something and its delay occurred because of mis-steps and internal concerns being amplified into revenue threats.
Vista started out as Longhorn, which would have a filesystem with relational database overlay so as to help users find their stuff (and related stuff) quickly. Also, Apple had implemented its graphical interface in Display Postscript, meaning the screen and the printers were fed the same data, PDF saving of any printable element was a side effect that customers liked. Microsoft, and I don't think it was copying, it was just the obvious way to go, does something similar (using a pdf knock-off called xps) and gives it a brand name.
ILuvYou breaks in 2003, and Microsoft has an effective year-long security introspection. Longhorn's team focuses on security and XP SP2. 2004, otherwise the expected year for Longhorn, passes without its release.
In 2005 or so, Google has made a huge business out of search services and it is most definitely not using relational techniques to produce quick results from searches of that dynamic database called the internet. Adios the relational file database system. Hello, background indexing, metadata, and utilizing mutli-cores and parallel processing.
As Apple discovered when it went Quartz/Aqua, the nice graphics extracts a large cost in speed. Apple let the processors improve and did some optimizations (farewell pinstripes?). and had nearly annual os updates. (Performance improvements, that's why we paid paid for our service packs, they added value!) Microsoft will put the optimizations in service packs and the follow-up to Longhorn.
2005, Microsoft discovers that the NT code base is such a hair-ball that something has to be done. A hero is born as one of the engineers leads a team that makes the os more modular and which means that there will be a successor to XP, something the unreconstructed codebase was not going to enable. Wall Street, meanwhile, is saying that Microsoft is clearly having problems with shipping a key product. It's right and wrong. Microsoft was having problems, there was a tiny bit of erosion of share to Apple, every speaker at a tech conference saw more Apple logos glowing back, and web developers invested in LAMP decided that the best platforms for development were Linux or OS X, so farewell cutting-edge users. Wall Street was wrong in that Microsoft still put money in the bank if the pc sold was running XP.
At the end of 2005, Bill Gates said Vista would ship at the end of 2006, unless it wasn't ready, because Microsoft was committed to getting it right. I suspect that meant it was coming at the end of 2006 no matter what.
Turns out, the "what" was driver support from third parties. Many folks excited by Vista's release quickly restore XP as their computer, even the ones with the Microsoft approved "Vista Capable" sticker, didn't have the graphics horsepower to work. Some found that key peripherals wouldn't work and could not work until the vendor provided a driver. Enterprise waits a year for the first service pack and generally decides that Vista is not for them. (They use their license fees to continue to run XP, Microsoft's net loss for the choice, 0.)
In 2007, OEMs started to put out Netbooks, small, light, single-core Intel processor powered, and inexpensive computers that took off like gangbusters. Mostly ran Linux because it brought retail costs down. Linux had 70% of the share in that sector. That was a big problem in Redmond—Wall Street are not the only ones with misguided perspectives—and so Microsoft needed to spend some money to get that share number turned around.
But, Vista was too big and needed too much processor power, so XP's life was extended because it could run on Netbooks.
Microsoft also lost revenues because the Netbook XP license was discounted in order to help the OEMs choose Windows without having zero profit. There was a Win7 Starter Version (remember the brouhaha pre-release because it would only allow three applications to be opened at a single time?) which was discounted, but could not be used by an OEM except the device truly was small and low-powered.
Netbooks were a short-lived phenomenon. Something people bought because Microsoft and its OEMs didn't figure out light-weight, portable computing the way Apple did with the iPhone, to some degree, and the iPad, most definitely.
So missing search, codebase sprawl, next-gen graphics now requiring better hardware, advanced security that changed the rules for drivers, mollifying Wall Street, moving faster than their platform affiliates, missing netbooks, and caring about netbooks (and I have no doubts that there were dissenters to upper management who focused elsewhere and missed opportunities or didn't apply the stitch in time), these are the goofs that led to Zombie XP.
Oh, and Vista, Win7, and Win8 did not provide utilities for seamless migration of applications from XP. When it was clear to me, an Apple-using idiot, that Microsoft's first task was to GET FOLKS OFF OF XP, they didn't spend the engineering resources to make it easy.
Ah well, big company, big problems. Besides, my large-caps nonetheless, Microsoft's real problems aren't Win8 or XP, it's that people are holding on to old pcs, because the hardware advancement has a smaller delta YOY than 10 years back, the negative growth in new computer sales, though this may be temporary, and where there is sales growth, someone else's operating system is on the device. Easy migration is one more mitigation, but really, if the computer turns on and the applications works, that's as much as many folks care about. Besides, Microsoft still makes money, so XP migration revenues at this time would be the frosting roses on top of a nicely frosted cake. Once support stops, the cost for having XP users goes to 0.