By consistently and continually molding Firefox to be as similar to Chrome as possible instead of offering something different enough to present a meaningful choice.
Of course most users took the path of least resistance if the choice was between Chrome and pretty-much-also-Chrome.
I think FFs loss of market share of course has been affected by advertising, but mainly by their own mistakes. I don't think the similarities to Chrome (which were not really so great for a long time) were the main driver.
I tried to hold onto FF for as long as possible for ideological reasons, and then for some years when I couldn't justify it because of just terrible performance, I used Chromium. They had a much, much better plugin market / platform for a long time and that too was a larger part of the draw that differentiated them. I switched back to FF as soon as I heard they had dealt with most of their speed issues in an update but, extremely reluctantly, I have been considering whether they are any better than Chromium at this point with the privacy / ad shenanigans that really seem to insist on reappearing in different ways with great consistency lately. The recent monetization debacle has frankly been disheartening.
So the main points for me are that they failed on speed (which drove most people I know who switched to Chrome) and now on culture/honesty.
I still use Firefox and only Firefox, as I still see it as the lesser evil, while being performant enough (I never really noticed any significant slowness that people keep harping about), but Mozilla is making it harder and harder every year.
Still, I am nowhere near wanting to switch to a Google product - Firefox would have to become really bad for me to consider that - and I see no other browser even remotely close to replacing Firefox for me. It's still "good enough".
I even keep an ancient Firefox version installed in /opt, because I need it to run a small proprietary Java applet for work, but that's another topic. :)
> I never really noticed any significant slowness that people keep harping about
> /opt
From what I understand, the slowness was primarily on Mac (a high-dpi rendering issue) and sometimes Windows (don't recall if there was a specific issue or just several small things that added up). On linux these issues tended not to exist, or at least be nowhere near as bad.
Is it even possible to be "different enough" from Chrome without also being a significantly worse browser? Becoming "as similar to Chrome as possible" is also the thing most browser users want. e.g. Multiprocess broke the XUL add-on model, but it also means dramatically better browser stability.
Multiprocess was a step in the right direction, of course - nobody likes it when one crashed tab brings down everything else with it.
But they kept on removing various cool and useful features just because their telemetry showed that they were only used by a small percentage of users, claiming that they need to save money and focus on things people use the most. Naturally, that leads to the world's blankiest blank of a browser, to borrow a bit of Futurama vernacular.
Of course most users took the path of least resistance if the choice was between Chrome and pretty-much-also-Chrome.