Every core update produces the same two reactions: panic from the sites that dropped, and silence from the sites that didn't. The August 2026 update deserves more attention than either, because it is the clearest signal yet about what Google is prepared to rank now that AI Overviews resolve a large share of informational queries before anyone scrolls.
The pattern across the sites we monitor is consistent. Pages that summarised what other pages already said lost visibility. Pages carrying something Google could not find anywhere else — original data, first-hand testing, named expertise, genuine local knowledge — mostly held or gained. That principle is not new. What is new is how little middle ground is left.
This was a broad core update rather than a targeted action against a specific tactic. Google re-weighted how it assesses quality across the whole index, which is why the movement looks so uneven from site to site. Three effects stood out.
The "complete guide" assembled by reading the current top ten and rewriting them took the hardest hit. When an AI Overview can synthesise that answer in a sentence, a page that offers nothing beyond the same synthesis has no remaining reason to exist in the results.
Pages trying to serve three intents at once — part explainer, part comparison, part sales page — fell behind pages that committed to one job. Search intent has become less forgiving because the alternative answer is now instant.
Unlinked brand mentions, branded search volume and consistent presence in the places your customers already read appear to carry more weight than they did a year ago. If people search for you by name and reputable sites mention you without linking, that counts.
Change nothing for the first fortnight. Core updates roll out over days and rankings move erratically while they do; sites that panic-edit in week one routinely undo work that would have recovered on its own. Use the time to measure properly instead.
That last question is uncomfortable and it is the one that matters. Most of what we see labelled as an algorithm problem is a content problem that only became visible when the threshold moved.
Recovery is rarely a technical fix. It is usually a decision about what you are willing to put on the page that nobody else can.
Numbers from your own client work, screenshots of a real result, the objection you hear every week and how you answer it, a named person with a job title standing behind the advice. A single paragraph of genuine experience outperforms a thousand words of competent summary.
Thin pages published to fill a content calendar drag the rest of the site down. Merge them into stronger pages, redirect them, or delete them. A smaller site of pages you would happily show a client usually recovers faster than a large one you quietly hope nobody reads.
If your page disappeared tomorrow, would anyone notice anything was missing? If not, Google reached the same conclusion first.
Decide whether a page is there to explain, to compare, or to sell, then commit. Split the ones doing all three.
Core updates do not punish sites so much as re-sort them against a higher bar. The August 2026 bar is evidence: something on the page that could only have come from you. Diagnose calmly, fix the thin pages first, and put your actual expertise on the page rather than a polished description of it. That work survives the next update too, which is more than can be said for most recovery tactics.
We audit what actually caused the drop before touching a single page — then rebuild the content Google now rewards.
Talk to us