Check this page with an assistantOpens a chat asking it to summarise this article and name the evidence behind each claim.
Claude opens with the prompt on your clipboard: Anthropic does not support prefilled prompts on the web, and we would rather copy it than ship a button that drops it.
Three Kinds of Update, Frequently Conflated
Named updates were systems built to catch a specific abuse. Core updates are broad reassessments of quality with no particular target. Spam updates refine detection of policy violations. They produce similar-looking declines and require completely different responses.
The distinction determines whether you have something to fix. A spam update decline means you were doing something the policies name, and the remedy is to stop. A core update decline means the assessment of quality changed and your pages compare less favourably than they did, which is not a violation and has no appeal.
Google has been consistent on this: core updates are not penalties and there is nothing wrong with pages that decline. That is frequently read as evasion and it is mechanically accurate. Treating a core update decline as something to appeal wastes the quarter that should go into whether the pages genuinely deserve to win.
The Updates Worth Knowing
Most of the historical named updates no longer run as separate events, having been absorbed into the core systems. They remain useful as a description of what each behaviour was, because the behaviours are still assessed continuously.
| Update | What it targeted | Status |
|---|---|---|
| Panda | Thin, low-value, duplicated content | Absorbed into core systems |
| Penguin | Manipulative link building | Absorbed; links are now largely discounted rather than penalised |
| Hummingbird and BERT | Understanding meaning rather than matching strings | Permanent; see semantic search |
| Mobile-first indexing | Assessing the mobile version as the page | Complete; see mobile SEO |
| Helpful content system | Content made for search engines rather than people | Folded into core ranking |
| Site reputation abuse | Borrowing a domain's standing for unrelated content | Enforced algorithmically; see the reputation crackdown |
| Scaled content abuse | Mass-produced pages made to rank rather than to help | Active; see does Google penalise AI content |
| Discover updates | The Discover feed specifically, from February 2026 | New category; see Discover SEO |
The Penguin row contains a change people miss. The modern handling of manipulative links is largely to discount them rather than to penalise the target, which is why most sites should never touch the disavow tool and why inbound spam links are rarely the cause of a decline, as covered in negative SEO.
The Cadence Changed, and That Matters More
Core updates used to arrive with months between them, leaving a stable window to make a change and read a clean result. That window has closed. 2026 has run at a noticeably faster pace, with core updates in March and May and spam updates alongside.
The practical consequence is that assessment is close to continuous. You can no longer ship a change, wait, and attribute the result cleanly, because something else moved in the meantime. That raises the value of a control group and lowers the value of before-and-after comparisons, which is the argument in SEO split testing.
It also changes the operating posture from reactive to maintenance. Rather than recovering from updates, the workable approach is keeping a site that has less to lose from them: fewer thin pages, accurate content, and a tail that is pruned on a schedule rather than after a scare. That is the argument in page-level authority.
Using the History as a Diagnostic
The reason to know this is not trivia. When traffic declines, the first question is what changed and when, and the update record is one of the two places to look. The other is your own change log, which most teams do not keep.
- Find the exact start date of the decline.
Not the month. Daily data in Search Console shows whether it began sharply on one day, which points at something shipping, or drifted over a fortnight, which points at a rollout.
- Check the Search Status Dashboard.
Google records confirmed updates with start and end dates. A decline beginning inside a confirmed window is most likely that update; one that does not correspond to any is probably yours.
- Check your own deploys.
Template changes, plugin updates, migrations, and robots changes. A sharp single-day decline on a deploy date is an engineering incident wearing an algorithm costume.
- Check whether it is seasonal.
Compare year over year rather than month over month. A large share of alarming declines are the same shape as last year, per seasonal SEO.
- Only then diagnose content.
If it was a core update, the work is identifying which pages lost and whether they genuinely deserve to win, per core update recovery.
What the History Tells You About the Future
One pattern holds across fifteen years: every named update targeted a tactic that worked by exploiting a proxy rather than by being genuinely good. Keyword stuffing exploited term frequency. Link schemes exploited citation counts. Content farms exploited publishing volume.
The tactics changed and the shape did not. Each generation found a measurable proxy for quality, optimised the proxy past the point where it correlated with quality, and was eventually caught. That is a reliable enough pattern to plan against, and it predicts what happens to the current generation of proxies without needing to name them.
The practical implication is a filter for tactics. If something works because it exploits a measurement rather than because it makes the result better for the person searching, it has a shelf life, and the shelf life has been getting shorter. That filter would have correctly flagged nearly everything on this timeline in advance, which is the most useful thing this history offers.
Questions About Algorithm Updates
- What is the difference between a core update and a named update?
Named updates such as Panda and Penguin were systems built to target a specific abuse: thin content in one case, manipulative links in the other. Core updates are broad reassessments of how quality is judged across everything, with no specific target. A decline from a core update is a change in relative assessment, not a sanction.
- How often does Google update its algorithm now?
Constantly in small ways, and several times a year in ways it confirms publicly. The confirmed cadence has accelerated: where three to four core updates a year was typical, 2026 has run considerably faster, with core updates in March and May and further spam updates alongside them.
- What is a Discover update?
A newer category, first appearing in February 2026, scoped to the Discover feed rather than to Search results. It matters because a site can see substantial Discover movement while Search is untouched, and diagnosing that as a Search problem sends you looking in the wrong report entirely.
- Do the old named updates still matter?
As history rather than as active systems. Panda and Penguin were eventually absorbed into the core ranking systems and no longer run as separate periodic updates. Knowing what they targeted is still useful, because the behaviours they were built to catch are now handled continuously rather than in occasional sweeps.
- How do I tell if an update hit me?
Compare the start date of your decline against the confirmed update windows on Google's Search Status Dashboard, then check your own change log for anything you shipped. A decline beginning on a deploy date is an engineering incident; one beginning inside a confirmed rollout is most likely the update.
