Update: clawed back from ~3-4K to ~8-9K daily clicks after the May Google core update — here's what actually worked
A month ago I posted asking has anyone recovered from a core update hit like this — Chatzyo, a no-login video/voice/text chat platform, had gone from 20K daily organic clicks down to 3-4K almost overnight after the May 2026 update. A few people pushed back on jumping to conclusions too fast, which turned out to be the right call. Wanted to close the loop with what actually happened.
Quick recap for anyone who missed the first post: we'd grown to 350K monthly organic clicks over about 6 months, mostly on long-tail, language-specific chat room pages (Tamil, Telugu, Hindi, Malayalam). The update hit those pages hardest — no manual action, no obvious technical issue, just a broad re-ranking that happened to land on us.
What we actually found once we stopped panicking and started digging:
- Pulled Search Console's Performance report and compared pre/post by page and query instead of just watching the top-line number. That's what confirmed it really was concentrated on the regional/language pages, not site-wide.
- Audited our actual backlink profile using GSC's Links report (free). Turned up something we didn't expect — over half our referring pages were coming from a cluster of ~20 generic "AI tool directory" sites from a bulk-submission tool. Zero relevance to a chat platform, and it turns out that's a pattern Google's own link-scheme guidance calls out directly. We had no idea this had been happening.
- Checked indexation, not just rankings, on the affected regional pages — some of them had drifted into thin, near-duplicate territory across languages/countries, which is an easy trap when you're template-generating geo pages at scale.
What we changed:
- Stopped the mass directory submissions immediately once we traced where they came from.
- Rebuilt out the thin regional pages with real depth instead of near-duplicate templates — Brazil and Tamil are the two that came back strongest.
- Cleaned up existing profile/directory listings instead of chasing new ones (fixed a mis-categorized listing, verified a review-site profile was actually claimed).
- Started pursuing a small number of genuinely relevant editorial backlinks — dev/tech communities, WebRTC-adjacent blogs — instead of any more automated submissions.
Where we are now: back up to roughly 8,000-9,000 clicks a day. Still below the 20K peak, but a real, steady climb off the 3-4K floor, and the pages leading the recovery are exactly the ones you'd hope — homepage, Brazil, Tamil, Indonesia, Telugu, Malayalam all showing up again in the top-pages report.
Not declaring victory — one recovery trend after one update cycle doesn't mean much yet, and I'm wary of overfitting a story to a few weeks of data. But the honest version is: the fix wasn't a trick, it was unglamorous cleanup (kill the bad backlinks, fix the thin pages, stop guessing) done patiently over about a month.
If anyone else got hit by the May update and is still in the "has anyone recovered" phase — happy to compare notes on what you're seeing. And if you want to see what we're building, it's at chatzyo.in.
Four changes in one month, one recovery curve - you can't tell which paid. Page rewrites re-rank on each page's own recrawl; a link-profile change only moves once that referring cluster gets recrawled and rescored, which is slower. Brazil and Tamil came back first and those are the pages you rebuilt, so the content work is probably doing more of the lifting than the directory cleanup is getting credit for.
Before treating 20K as the target - ALM Corp's February 2026 analysis put AI Overviews on 48% of tracked queries, up from about 31% a year earlier. On question-shaped long-tail some of that gap isn't coming back, clean profile or not. Impressions against clicks tells you which half you're fighting.
The next useful step is a recovery ledger at the page-cluster level: change date, thin-content rewrite, link cleanup, indexation state, query class, and weekly clicks. Keep a small set of affected pages untouched as a control where possible. That will help separate the impact of content depth from normal post-update volatility and make the playbook reusable for the next core update.
SEO recovery posts are always more useful when they include what didn’t work too. After a core update, it’s very easy to change ten things at once and never really know what caused the recovery. Would love to hear which changes you’re most confident actually moved the needle
Recovering after a core update is much more useful to read about than generic SEO advice because you can actually compare what changed before and after. Going from roughly 3–4k back to 8–9k daily clicks is a serious recovery. Which change do you think had the strongest effect versus things that just happened to correlate with the recovery?
The recovery being concentrated on specific pages makes the before-and-after especially interesting. Curious which change had the clearest effect once you compared the affected pages individually.