I suspect it's what Robert Charlton said, too - content.
This may seem obvious to some, but it wasn't to me until fairly recently:
I think Google's slowly reached a new confidence threshold, making some of the established SEO techniques superfluous. In other words, Google's increased its confidence in its own reading of value signals. It doesn't need us to tell it what we were telling it before when we strategically optimized for search. So it's started ignoring obsolete signals and looking for new ones, and if those signals are there, it's ranking those pages well.
(Whether or not those value signals come from the user or Google's own definition imposed on the user is another matter.)
What I mean is, it seems like Google better knows what content means and is looking at a different part of it than it was before, like a blind robot that's become suddenly intelligent and proceeds to deactivate some of its own previous feedback sensors because now its eyes can see, and it's pretty new at seeing and rather surprised to find the world so different, and not altogether sure of what to do with what it finds. But now its eyes are opened, and no matter what it does, the world will always look different. (Yes, before you ask, I do think of Google as AI.)
If it were me, after your minor SEO tweaks, I'd first tackle the content, make it better and more satisfying to the visitor, and then if that doesn't do it, only then address the SEO. But content first.
The reason? SEO might get you short term results, but then you'll be at odds with Google more and more, in a constant struggle to get seen.
SEO exists not because it's a game to be played, but because it helps search engines think. As search engines start to think better, SEO becomes less important.
With content that gives your users what they want and what nobody else who's visible gives them, all you have to do is tweak the SEO here and there over time and you're fine. At least, that's what I believe. For what it's worth. I've only been at this a couple of years, and I mostly write for other sites, not my own.
And one rather odd tip: if anybody thinks his content is the best it can be, and never doubts it, there's a very good chance that it's not as good as he thinks, because he never doubts it. Content that delivers is about the reader, not the text. In print, writers have to take into account their readers first and foremost or they won't get read; online it's not much different, especially with so much competition entering the field every day.
Google says that over and over in their help pages - think of your visitor. While website developers are busy trying to get Google to notice them by flashing their arms around, Google's saying, "Wait! I'm getting better at seeing! Give me something worth noticing, and jab me in the ribs every so often to make sure I'm listening, and I'll see you!"
And "good" content doesn't have to be in-depth content, though it often is. It can be quick and easy to read. People searching for "how to replace a widget gadget" don't always want to know every step - sometimes they just want to know if they're capable of it. If some article sites are ranking higher for these kinds of queries, it may not be a mistake - it may be because that's what those queries were really asking. And for those long-tail terms being lost by those same user-generated content sites - vice versa; for some queries, only in-depth will do.
Google's bound and determined to figure out the meaning and intent behind queries - thus their introduction of implicit triggering to definition queries.
So I hope all that doesn't come off as moralistic. I'm not trying to draw the line between good content and crappy content and say anybody's content is one or the other. As I said, it's the reader, listener, or viewer that decides whether something's worthwhile as far as search engines care, and that changes over time. And I don't think Google or any of the search engines are very good at what they're trying to do yet and they're making a lot of mistakes. I'm just talking about a way of looking at content that (so far) might help you.