The May 2026 core update is the last of its kind

The May 2026 core update has been brutal. Saturday the 30th was the worst day of it — Semrush, Sistrix, Mozcast, every tracker spiking together, the kind of synchronised volatility that only happens when Google is rewriting weights at the index level. Glenn Gabe is posting gambling-niche carnage. WebmasterWorld is the usual mix of "we did everything right" and "biggest hit ever."

I've been watching core updates for eighteen years. The pattern of this one is normal. The context isn't.

Because while everyone is staring at their Sistrix charts trying to work out whether they got hit, a different question is becoming unavoidable: what exactly are these rankings *for* anymore? Tom Capper's pixel research says position one is invisible on mobile two thirds of the time. Google's I/O demos show a checkout flow that never touches a merchant's site. AI Mode just crossed a billion monthly users. The thing the core update is reshuffling is a thing that increasingly the end user never actually sees.

This is going to be one of the last core updates the SEO industry treats as a primary event. Not because core updates are ending. Because the surface they update is shrinking inside the search experience, and we haven't built the vocabulary to talk about that yet.

What the update is actually doing

Let's deal with the update on its own terms first.

Google announced it on May 21st. Volatility kicked in the following weekend, settled mid-week, then erupted again Friday the 29th and especially Saturday the 30th. Barry Schwartz is right that this looks like the tail-end spike that usually precedes a "rollout complete" announcement. The two-week window closes later this week.

YMYL categories — gambling, finance, health — are seeing the largest swings, which is the standard core-update fingerprint. Sites that had partial recoveries mid-rollout are seeing fresh drops. Some of them will recover again when the rollout completes. Some won't. This is how core updates have worked since the medic update in 2018.

The advice for sites that got hit is the advice it has always been. Don't panic. Don't make sweeping changes during the rollout. Wait until it completes, then assess. Look at what changed at the page level, not the site level. Check whether the queries you lost are queries an AI Overview is now answering. That last bit is new, and it matters.

The thing the update doesn't show you

Here's what a Sistrix chart doesn't tell you in 2026.

The rank tracker is measuring a coordinate on a page. The page is no longer the experience.

It doesn't tell you that the keywords you "gained" position on might be keywords where position one is now buried under an AI Overview that quotes a competitor. It doesn't tell you that the keywords you "lost" might be keywords where the user got their answer from an agent before they ever saw a SERP. It doesn't tell you whether your traffic decline is from the core update, from AI Overview expansion, from AI Mode adoption, or from agentic flows that bypass the SERP entirely.

The rank tracker is measuring a coordinate on a page. The page is no longer the experience.

I wrote about this last week after Tom Capper's SEO Week presentation. The median organic position one sits 635 pixels down on desktop, against an 800-pixel viewport. On mobile, the first organic result isn't visible at all about 60% of the time. Add AI Overviews on informational queries — they consume nearly a third of above-the-fold space on their own — and the picture gets worse.

So when a core update reshuffles organic rankings, it's reshuffling a layer of the SERP that the user often doesn't reach. That doesn't make the reshuffle unimportant. It makes the reshuffle harder to translate into business outcomes.

Why this one feels different to long-time operators

I've been emailing with clients all weekend about this. The conversations are different to a year ago.

A year ago, the question was "did we get hit and what do we do." This year, the question is "I think we got hit but I genuinely can't tell because our analytics is so broken now." Server-side tracking gaps, Google's data redactions in Search Console, AI Overview impressions counted as impressions but generating no clicks, agentic flows that never land on the site — the measurement infrastructure that used to make core updates legible is degraded enough that operators are flying on instinct.

The core update is doing what core updates always do. The environment around it has changed enough that the old playbook for responding is partially obsolete.

That's an awkward sentence to write because the *fundamentals* of how to recover from a core update haven't changed. Content quality. Topical depth. E-E-A-T signals. Technical hygiene. Brand strength. All of it still works. What's changed is what "recovering" means. A full recovery in the old sense — getting your rankings back, getting your traffic back, getting your conversions back — might no longer be the same event. You might recover your rankings and still lose traffic because AI Overviews ate the click. You might lose rankings and still keep traffic because brand search held up. The signals are decoupling.

What Mustafa Suleyman and Jensen Huang are actually arguing about

Buried under the core update chatter this weekend was something else worth paying attention to. Microsoft's AI chief said most white-collar work will be automated inside 18 months. Nvidia's CEO told Carnegie Mellon graduates to consider becoming electricians. Both made headlines. Both got read as predictions.

They're not predictions. They're framings of the same underlying question that this core update is also asking: what's the unit of value when the surface layer of your work has been automated?

Huang's distinction is the useful one. "The task and the purpose of a job are not the same." Most tasks in SEO have been getting automated for years. Keyword research, audit checklists, content briefs, technical crawls, schema generation — the task layer has been collapsing into tooling since at least 2020 and into LLMs since 2023. The *purpose* of SEO — making sure the right people find the business at the moment they're looking — hasn't been automated and may not be automatable.

The core update is a task-layer event. It reshuffles ranking outputs based on quality signals. The purpose-layer question — *did the right people find the business* — isn't actually answered by checking Sistrix on Monday morning.

What the next playbook looks like

I don't want to be glib about this because the industry is genuinely struggling with the transition, and the glib answer ("just build brand!") isn't a playbook.

Here's what I think changes in how a serious operator approaches a core update from this point onwards.

You stop treating ranking volatility as the primary signal. You measure SERP pixel visibility, AI Overview citation rates, branded search trends, and direct/referral traffic to the same dashboard as your ranking data. You weight them roughly equally. None of them tell you the whole picture on their own; together they triangulate something closer to reality.

You stop assuming a traffic drop has a single cause. The May 2026 core update might explain 40% of your decline. AI Overview expansion in your niche might explain another 30%. AI Mode adoption among your audience might explain another 20%. The remaining 10% might be seasonality. Treating any of these as the single cause leads you to fix the wrong thing.

You stop optimising for queries and start optimising for being the answer. The query layer is increasingly Google's to mediate. The answer layer — the structured, citable, authoritative content that AI systems quote when they answer — is yours. The core update affects how Google ranks pages. AI Overviews affect which pages get *quoted*. These are related but not identical optimisation problems, and most teams still only have a workflow for the first one.

You stop treating recovery as a single event. Some of what you lost in this update will come back. Some of it won't, ever, because the SERP that used to deliver it has been replaced with an AI surface that delivers the answer without the click. The honest assessment after a core update now requires asking which of your lost rankings are recoverable in principle, not just whether your content can rank again.

The honest limits

I want to be careful about a few things.

First: this isn't the end of organic search. Total search demand is still growing. AI Mode at a billion users is impressive but Google Search is still well north of ten billion queries a day. The SEO playbook works. It just works for a smaller portion of the discovery journey than it used to.

Second: pixel visibility and AI Overview citation rates are still imperfect measures. We don't have an industry-standard way to weight them. Most tools that claim to measure them are early and partial. I covered the four-layer measurement problem a few weeks ago — the gap is definitional more than technical, and it's going to take another twelve months to close.

Third: some of the panic about this is overcooked. Mike King's "death of the open web" piece is provocative and worth reading, but the conclusion most operators should draw isn't to abandon owned media. It's the opposite. I argued this last week. Owned channels — email list, community, direct relationships, branded distribution — become *more* valuable when discovery is mediated by agents you don't control, not less.

The May 2026 core update will finish rolling out this week. People will write recovery analyses. Forums will fill with case studies. The industry will move on to whatever the next update is.

But the bigger transition won't be on anyone's chart. The core update is the last familiar event in a search environment that's becoming progressively less familiar. The playbook that worked for the March 2024 core update won't work the same way for the next one. Not because the fundamentals changed — they didn't — but because what we're measuring, what counts as recovery, and what a ranking even *means* are all in motion.

The teams that come out of this period in good shape will be the ones who stopped treating ranking volatility as the headline story and started treating it as one input among several. The teams that don't will keep optimising for a SERP that increasingly isn't the experience.

That's the loop. And the May 2026 core update is the last one most of us will run the old playbook on.