Clones took ages to make an impact in the home market, but they made an impact in the office environment because of price.
I think what I'm trying to claim is that the IBM PC was already dominant, before clones dominated the market. That is, clones followed the pre-existing success of the IBM PC; they didn't make it successful. And this meant the reason the IBM PC became successful wasn't because it was open, but because it was IBM. For example, as far as I understand, MDA and CGA card clones appeared on the market with the rise of the clones, not because of the 'openness' of the IBM PC. Instead, different third party graphics cards such as the Hercules Graphics Adapter appeared first, but they were targeted at the IBM PC itself, not the clones.
Nevertheless, it's tricky to find actual figures. We could get a flavour for how the market changed by reading back issues of Byte, say every 6 months from when the 5150 appeared in 1981 to the late 1980s: that's 20 issues. I found this article:
Not even Big Blue could keep up with its creation’s success
spectrum.ieee.org
Which said: "IBM's share of the PC market shrank from roughly 80 percent in 1982–1983 to 20 percent a decade later". Subsequently I found an actual graph from this
Research Gate article.

So, rough estimates for when clones were the majority of PC + Compatible sales were, from this graph, the end of 1986 when top-tier clones were at about 4.9%, other clones at 31.66% and IBM at 36.25%. But the PC was already
the computer standard by the end of 1983, beginning of 1984, 3 years earlier.
This is what I mean about tackling computer industry myths, like the idea that the Apollo Guidance Computer had less power than a digital watch or simple scientific calculator. In fact it had roughly the capability of a Commodore C64 (with a slower clock rate). Or the one where people claim users never expected to have more than 64kB, or 640kB or 16MB or 4GB of RAM or 32MB or 524MB of disk space. It's all rubbish, because at the time users and engineers knew perfectly well we'd need more RAM, storage and MHz.
I suspect people don't like the idea that the PC won, because of the name IBM, because we prefer to think that people made rational,
technical, fair decisions and the idea that it won because of the IBM name, seems populist and unfair. But the people who have bought PCs in the past 40 years, because of its 'openness' and 'competition' weren't the ones who made it dominant. The people who made it dominant were ignorant sales and marketing executives. It had to be that way, because knowledgable computing types would have chosen it less frequently for multiple reasons:
- They can cope with retraining or reengineering whereas these executives couldn't, as you say: retraining is expensive for them.
- They can explain to higher management why computer x is better; these other executives couldn't justify other computers on a technical basis.
- They had the technical expertise to truly compare computers; whereas these other executives weren't able to even justify buying a non-IBM PC to themselves, because they didn't have the technical expertise to compare an IBM PC with, say, a Sirius One (Victor 9000). They did have the confidence to understand IBM wasn't going to go bust next year.
It was already too late in 1987 for IBM to regain control with the MCA PS/2s, because they no longer had a majority (27.5% vs 12.5%+34%). But they did contribute the PS/2 keyboard/mouse and VGA graphics - things that could be easily ported to new PCs.
Also, we can see how names matter more than price or technical features from this graph too as the top tier PC manufacturers overtook generic clones by 1994. In fact generic clones peaked at the end of 1980s. Why? Because users who don't understand the technicalities are only able to choose based on a name they have confidence in. It's pretty much the same story, just with different players.
Before the PC every time you purchased a computer for the office you needed to buy all new software and train the office staff in its use and that was big money back then.
Yep, I agree it helped, but it applied to both original IBMs and clones.
Once people realized that each new faster PC would run their old software but faster sales went crazy. Clones and volume drove down prices which also caused an explosion of available software since it was very profitable.
80286 PCs appeared in 1984 with the PC/AT. The first 80286 clone appeared.. when? Compaq beat IBM to the first 80386 PC in 1986.
In most business and home environments though, slow 8MHz or so 8088 or 8086 PCs were dominant until the late 80s. My workplace in late 1989/early 1990 gave me a 12MHz, Intel 80286 Tulip computer with 1.25MB of RAM. My workplace in 1992 to 1995 typified that in that we had a few cheap Amstrad PCs; an 80286-based PS/2; an Apricot F1 (for stores, not even a clone) and several 80386s. When I arrived, I had the fastest computer, a 40MHz AMD 80386 and 2MB of RAM + 40MB HD! By the time I left we had 486's (including a DX2 or the IBM branded Blue Lightning DX3 at 75MHz/25MHz), but I was still using that same computer. We also had a few 386sx's for reception and other staff; while a no-name 386sx ran a Netware server.
Home users very much cared about buying a winner because platforms were abandoned very quickly and then you could not get software or service for an expensive home computer. I knew people who sent back their ADAM computers the second they found out it was being discontinued even if they liked it.
OK, so that's some anecdotal, but valuable evidence.
Sure, if the price point was cheap enough people purchased the machine, but few really did anything with it. How many people actually used their Timex 1000 before buying something with actual software and a real keyboard? Home users tended to jump around with home consoles because of the rapid advances in technology.
I have a different memory, but I'm UK based so it could have been different here. Home users here settled on ZX Spectrums and Commodore C64s by the early/mid 80s and were still using them in the late 80s; while gamers moved onto Atari ST / Amiga. I don't remember anyone in my computer science 1986 intake at UEA who used an actual PC in their room. We just couldn't afford one. At that level, most students still didn't have a computer, but for those who did it was mostly BBC micros and STs. I knew one person with an Amiga; one other with an Elan Enterprise; another with an Acorn Electron (who later bought an Archimedes). I had a Sinclair QL. Until my 3rd year, no-one I knew even had a Mac, despite the Uni having plenty.
Oh, hang-on. I think a few might have bought a PC towards the end of my degree, because I heard about people building up their own clones. At one point I looked into it too; scouring the back of Personal Computer World magazine for cheap motherboards, cards, RAM, disks and keyboards. However, I still couldn't justify the expense, even if I could get it down to about £500. Also, I was already sold on Macs, didn't like PCs and yet used my QL as my main machine until 1993.