I run eleven sites. Seven were hit, four were not. Six months of data, so the dust has settled enough to say something useful.
Recovered (3 sites)
All three had the same profile: a real brand, a visible author with credentials that check out, and traffic spread across hundreds of pages rather than concentrated in a handful of money posts. What I changed was mostly subtraction — I deleted or merged about 30% of the pages, all of it thin stuff written to catch a keyword with no reason to exist.
Partially recovered (2 sites)
Both are back to roughly 60% of pre-update traffic and have been flat for three months. These are the sites where I cannot honestly claim the content was better than what already ranked. I think 60% is the new ceiling and I have stopped spending on them.
Did not recover at all (2 sites)
Programmatic pages built off a dataset with a template around it. Thousands of URLs, no editorial value, nobody would miss them. I was always going to lose these eventually and I knew it while I was building them.
What I would tell myself a year ago
The sites that survived were the ones a human would bookmark. That sounds like a platitude and I resented it when people said it to me, but it is the only variable that separated my winners from my losers. Everything else — schema, internal linking, page speed, publishing cadence — was identical across all eleven, because I run them all the same way.
Deleting pages was the single highest-impact action, and it was terrifying every time.
Building boring sites that print money. Case studies in my journey thread.