How Google’s Algorithm Has Actually Evolved (2011–2026)

I’ve been writing about Google updates on this blog since before some of my clients had smartphones, and if there’s one thing fifteen-odd years of this has taught me, it’s that everyone panics about the wrong things every single time. So instead of chasing the next one, I thought it was time to pull the whole story together in one place — properly, not as five separate reactive posts written in the heat of the moment back when Penguin was brand new and everyone thought their site was about to vanish overnight.

I’ll keep this one updated as new updates roll out, rather than writing a fresh “OMG Google changed something” post every time. That’s the plan, anyway.

The early days: brute-force spam fighting

Long before Panda and Penguin got their names in lights, Google was already going after the obvious stuff — keyword stuffing (Florida, 2003), duplicate content across domains (Big Daddy, 2005), and manipulative link schemes (Jagger, 2005). It’s easy to forget how crude search results were back then. If you could stuff a keyword in often enough or get enough links pointed at a page, you could rank for almost anything, regardless of whether the page was actually any good.

Panda and Penguin: the pair everyone remembers

Panda (from February 2011) was the first update that really went after content quality rather than pure spam — it hit content farms and thin, low-value pages hard, and kept getting refined for years afterwards (I covered Panda 2.2 on this blog back in 2011, wondering aloud whether Google would ever get it “finished” — spoiler: they didn’t, they just kept iterating).

Penguin (from April 2012) did the same job for links — going after unnatural link patterns and aggressive manipulation rather than content. I wrote about the 2013 Penguin update at the time, and about recovering from Penguin the year before that, both written in real time while everyone in the industry was frantically disavowing links and trying to work out what had actually happened. With hindsight, the pattern was obvious: sites built on genuinely useful content and natural links mostly rode it out fine; sites built entirely on manipulation didn’t.

By 2016, Penguin 4.0 changed how it worked entirely — instead of penalising a whole site, it started just devaluing the bad links in real time. That single change tells you a lot about where Google was heading: less “punish the site,” more “just ignore the noise and reward what’s left.”

Hummingbird and the shift to understanding language

Hummingbird (2013) is the one I think deserves more credit than it gets. It wasn’t a spam crackdown at all — it was a rewrite of how Google interpreted a query in the first place, moving away from matching individual keywords and towards understanding what you actually meant. That’s the foundation everything since has been built on, from RankBrain (2015) — Google’s first real machine-learning ranking signal — through to BERT (2019), which dramatically improved how well Google understood natural, conversational phrasing. If you’ve ever typed a full question into Google instead of three keywords and got a genuinely relevant answer back, you’re benefiting from a line that runs directly back to Hummingbird.

The mobile and user-experience era

Pigeon (2014) tied local search much more closely to Google Maps data. Mobilegeddon (2015) started rewarding mobile-friendly sites in mobile results, at a point when a lot of small business sites — including plenty I was working on at the time — genuinely weren’t built for phones. Intrusive Interstitials (2017) went after the aggressive pop-ups that made mobile browsing miserable. Later, Core Web Vitals and the Page Experience update (2021) folded loading speed, visual stability, and interactivity directly into ranking — which is exactly the kind of thing I still check for clients today (and exactly why the PageSpeed and Core Web Vitals scores in your Search Console matter, not just as a vanity metric).

The “helpful content” era (2022–2024)

This is where things get directly relevant to how I’d advise anyone writing content today. The Helpful Content Update (August 2022) was Google’s first explicit, named effort to reward “people-first” content and demote material written primarily to rank rather than to genuinely help someone. SpamBrain, Google’s AI-driven spam detection system, got a dedicated link spam update in December 2022 to catch artificial link networks at scale.

By March 2024, Google folded Helpful Content directly into its core ranking systems and announced it expected to reduce low-quality, unhelpful content in search results by around 40% — a genuinely enormous shift. The same period brought a site reputation abuse policy (May 2024), targeting what’s often called “parasite SEO” — low-quality third-party content hosted on the subdomains of otherwise reputable, authoritative sites purely to borrow their ranking power.

AI Overviews, and the updates still happening as I write this

AI Overviews launched in May 2024, putting AI-generated summaries directly into search results above the traditional listings — arguably the biggest visible change to the search results page in over a decade, and the reason “being cited by ChatGPT or Google’s AI answers” is now something worth actively tracking (something I’ve recently started doing for this site, as it happens).

Since then, the core updates have kept coming — August 2024, December 2024, and a December 2025 update that Google rolled out in just six days, the fastest core update on record. The March 2026 update leaned hard into rewarding sites that create original content over those that merely aggregate, list, or comment on other people’s work. The May 2026 core update went further still, specifically rewarding content with genuine topical authority and named expert authorship, while pages relying on thin or unedited AI-generated text lost visibility. An August 2026 spam update followed, continuing the same fight against manipulation that Google’s been running, in one form or another, since Florida in 2003.

Worth noting too: Google’s way of talking about all this has changed. For years Matt Cutts was the public face of Google’s webspam team, then Danny Sullivan took over as the more personal, candid “Search Liaison” voice on social media. Sullivan quietly stepped back from that public-facing role in August 2025, and updates are now announced in a more formal, institutional way through the Google Search Central accounts rather than a single recognisable person. Less personality, more press release — which makes it more important than ever to have someone keeping an eye on this stuff for you, rather than just watching for a tweet that isn’t coming anymore.

So what’s actually changed, in plain terms?

Looking back at over two decades of this, the trend line is remarkably consistent even though the mechanics keep changing: Google started by fighting obvious manipulation (keyword stuffing, link spam), moved on to genuinely understanding language and intent (Hummingbird, RankBrain, BERT), then user experience (mobile-friendliness, page speed), and has now landed on rewarding real, demonstrable expertise and original work over anything that merely imitates it — AI-generated or otherwise.

If you’ve built a site around genuinely useful content, written by someone who actually knows the subject, that loads reasonably fast and doesn’t try to trick anyone — you’ve been on the right side of essentially every one of these updates, going all the way back to 2011. That’s not a coincidence, and it’s not going to stop being true just because the next update has a different name.

I’ll keep adding to this page as future updates roll out, rather than writing a new standalone post each time — so bookmark it if this is useful.