Google's March 2026 Core Update: Who Won and Who Got Crushed

Here's the headline most people missed in the noise around the March 2026 Google core update winners and losers: the single biggest loser was YouTube. Not some no-name affiliate farm. YouTube. Google's own platform shed more visibility in two weeks than any domain we can find in the recent record.

If that doesn't tell you something interesting is happening, nothing will. So let's go through what actually moved, who gained, who got flattened, and what we genuinely think it means. No hand-waving. We track this stuff across client sites every day, and this update was a proper shake-up.

What Google actually did, and when

The first thing to clear up is the timeline, because half the internet got it wrong (including, we'll admit, an earlier version of this post). The March 2026 core update started rolling out on 27 March and Google confirmed it finished on 8 April, taking 12 days and change. Search Engine Journal reported that Google described it simply as "a regular update designed to better surface relevant, satisfying content for searchers from all types of sites," and pointedly did not publish a companion blog post or any new guidance.

That matters. When Google stays quiet, they're telling you the playbook hasn't changed, only the strength of enforcement has.

The second thing to clear up: the Discover changes were a separate update from February, not part of March. We had previously framed them as simultaneous. They weren't. The February 2026 Discover core update ran from 5 February to 27 February and was the first core update Google ever aimed exclusively at Discover. We've broken that one down separately in our look at the February 2026 Google Discover update and its traffic impact, so we'll keep our focus here on the broad search core update.

How violent was it, really?

Plenty violent. This was the most volatile core update in over a year, and the numbers from SE Ranking's comparison with December 2025 are the cleanest way to see it.

• 79.5% of top 3 URLs changed position, up from 66.8% in December. Only one in five top-three results held its exact spot.
• 90.7% of top 10 URLs shifted, up from 83.1%.
• Most brutal of all, 24.1% of pages ranking in the top 10 fell clean out of the top 100. In December that figure was about 15%. So nearly a quarter of page-one URLs didn't just slip a few spots, they vanished.

That last stat is the one we keep pointing clients to. Slipping from position 3 to position 6 is a bad week. Falling out of the top 100 is a structural problem with how Google now reads your site. Those are very different conversations, and if you've watched your rankings swing wildly through this period, the gap between "demotion" and "deletion" is the first thing to diagnose.

The losers: who actually got crushed

Now the fun part. We're only naming names we can verify from real visibility data, because every other "winners and losers" post on this topic is recycling the same vibes. The numbers below come from named, fetchable analyses.

YouTube, of all things

The story of this update. In Aleyda Solis's US analysis using SISTRIX data (measured 26 March to 11 April), YouTube posted the single largest absolute visibility loss in the dataset: down 24.7%, more than 1,058 visibility points gone. For a domain at YouTube's scale, that's almost unheard of.

Our read? This isn't Google punishing its own child. It's Google deciding that for a chunk of informational queries, a video result was never the best answer in the first place. When someone wants a quick fact, a written source that states it plainly beats a 12-minute video with the answer buried at minute nine. The update simply stopped pretending otherwise.

Dictionaries, encyclopaedias and reference sites

This is where the pattern gets loud. Solis recorded Collins Dictionary down 30%, Wiktionary down 21.3%, and OneLook down a savage 52.8%. The UK-focused SISTRIX radar analysis told the same story from a different angle, with Britannica off 9% and Collins Dictionary off 12.87%, and summed the update up neatly: "Authority beats interchangeability." Their blunt explanation was that these sites "offer content that is increasingly covered by AI-generated answers in search."

That's the quiet killer here. If your page's entire job is to define a word or restate a fact, Google's AI Overviews now do that inline. The click never happens. We dug into how brittle that traffic has become in our piece on when to prune thin content rather than defend it, and this update is the clearest evidence yet that "correct but commodity" content is on borrowed time.

Aggregators, directories and comparison sites

The middlemen got squeezed hard. PPC Land's write-up of the Solis data found losses "more concentrated among aggregators, directories, and comparison-driven sites" than in December. The specifics are eye-watering: in jobs, ZipRecruiter fell 36.6%, Glassdoor 36.3%, SimplyHired 43.2%. In travel, Travelocity dropped 44.3%. In property, Apartment Guide cratered 56.2%. Finance comparison sites took a beating too, with Credit Karma down 40.6%.

The thread connecting all of them: they sit between the searcher and the actual source. Google decided it would rather send people to the source.

Broad consumer health publishers

Healthgrades down 43.5%, Verywell Health down 26.3% in the same dataset. If you operate in a Your Money or Your Life niche, this should focus the mind. Generalist health content written to rank is losing to specialist, clinically grounded sources. More on the winners in a second, because they prove the point.

The winners: who Google rewarded

If you want the one-sentence summary of who won, it's this: the source, not the middleman. The original, not the summary.

Government and institutional domains

Solis recorded Census.gov up 30.2%, BLS.gov up 26.8%, and HUD up 36.2%. The official body that owns the data outranked the dozens of sites repackaging it. For anyone publishing original numbers, that's encouraging. For anyone publishing other people's numbers with a thin layer of commentary, less so.

Specialist destinations over generalist intermediaries

The jobs vertical is the perfect microcosm. While the big aggregators bled, specialist and first-party destinations surged: Amazon.jobs jumped 242.7% and myworkdayjobs.com 115% in the same SISTRIX dataset. People searching for a job at a specific company got sent to that company, not to a directory listing it alongside ten thousand others.

Authoritative, specialist publishers

Health makes the case again, just flipped. While broad health publishers fell, the New England Journal of Medicine rose 107.3% and Nature climbed 41%. The signal could not be clearer. Demonstrable, specialist authority is the moat. Generic "we cover everything" content is the casualty.

Strong brands holding their ground

The SISTRIX UK radar showed Amazon.co.uk as the biggest winner of the major domains, up 5.6%, with IMDb and Instagram also gaining. When a query has a clearly dominant brand answer, Google leaned harder into it. Brand strength, the kind you build over years, paid off.

Our honest take on what this update was really about

Strip away the individual domains and one philosophy is doing all the work: Google is collapsing the layers between a searcher and the truth.

Reference sites lost because AI Overviews now answer the question on the page. Aggregators lost because they're a tollbooth on the way to the real destination. YouTube lost where a video was a worse format than text. And the winners, almost without exception, were the people who originate the thing everyone else summarises.

We've been saying for a while that the old keyword-gap model is running out of road. Find a gap, fill it with competent content, rank. That worked when "competent" was rare. It isn't rare anymore, and AI made it free. So Google moved the bar from "is this competent?" to "is this the source, or a copy of the source?" If your content survives the question "could an AI Overview replace this entirely?", you're on the right side of this update. If it can't, no amount of clever internal linking or technical polish will save it for long.

One more thing, because it comes up constantly. This was not an "AI content penalty." Google has been clear it doesn't penalise content for being AI-assisted, and we've covered the evidence in our breakdown of whether Google actually cares about AI detection. What got hit was commodity content, the interchangeable stuff that adds nothing a model couldn't generate on demand. AI just happens to manufacture that at industrial scale, so the casualties skew heavily AI-produced. The cause is the sameness, not the tool.

What to do if you got hit

No quick fixes. There never are with core updates, and Google explicitly tells you not to chase them. Their own core updates guidance is worth quoting because people skip it: pages that drop "aren't necessarily 'bad'," and you should "avoid doing 'quick fix' changes" in favour of changes "that make sense for your users and are sustainable in the long term." Google also warns recovery can take "several months." So patience is part of the strategy, not a cop-out.

Here's the order we actually work through it with clients.

1. Separate demotion from deletion. Pull your lost queries. A page that slid a few positions is a competitiveness problem. A page that fell out of the top 100 is a quality or relevance problem. Treat them differently.
2. Run the source test on your top pages. Read your 30 best pages and ask: could an AI Overview replace this completely? If yes, you're commodity content. Add the thing only you can add. First-hand testing, original data, a named expert's actual opinion.
3. Check who replaced you in the SERP. If a more specialist or first-party source took your spot, that's the pattern this update rewards. Don't copy them. Work out which of your pages can credibly become the source rather than the summary.
4. Make your expertise visible on the page. Named authors, real bios, credentials, genuine experience. The reference and aggregator losers almost all share one trait: anonymous, interchangeable content. Don't be interchangeable.
5. Do not panic-publish. The worst move after a core update is dumping out 40 fresh posts to "recover." We've watched sites bury their good pages under a pile of mediocre new ones. Fix the assets you have first.

If you got caught in this one alongside the related March 2026 spam update, untangle which is which before you touch anything, because the fixes are completely different.

Where this leaves your strategy

The trajectory has been pointing the same way for years, and March 2026 just turned up the volume. Google wants to reward the origin of information and skip everything in between. As AI Overviews keep eating commodity clicks, the only content worth investing in is the kind a model wants to cite, not the kind it can replace. That's exactly the thinking behind our work on AI search visibility and getting brands quoted in AI answers, because the qualities that earn an AI citation are the same ones this core update rewarded.

For most sites, that means fewer pages, deeper expertise, real authorship, and original material that genuinely couldn't have come from anywhere else. It's slower and harder than the old gap-filling game. It's also the only version of SEO we'd put our name to in 2026, because it's the only one that survives the next update too.

If your traffic took a hit through March and April and you'd like a straight answer on whether it was demotion, deletion, the spam update, or just AI Overviews quietly siphoning your clicks, that's the kind of diagnosis we do every week. Tell us what's happened and we'll take a proper look. We'd rather give you an honest read than sell you a panic.