AI Growth Hub
← AI SEO

Google May 2026 Core Update: E-Commerce Volatility Again, and How to Diagnose It

google-core-update seo-strategy ecommerce-seo Google Search Console Semrush

What this update is, and why it isn’t March or April

Google started rolling out its second core update of the year on May 21, two days after the I/O Search redesign. That makes it the fourth confirmed ranking update of 2026. The official line is the usual: a broad update aimed at surfacing more relevant, more satisfying content. Full rollout takes roughly two weeks.

The reason I keep saying “not March or April” is that those two had clear targets. March went after content quality, and affiliate sites dropped about 71% on average, with unedited AI content punished hardest. April raised the weight of user-engagement signals, so dwell time, scroll depth, and return visits suddenly counted for more. Each gave you a specific thing to inspect.

May doesn’t. There’s no published angle beyond “relevant and satisfying.” That matters for how you diagnose it. In March you could audit which pages were thin or AI-stuffed. In April you could check whether engagement metrics cratered. This time there’s no obvious handle to grab. What’s dropping in the early trackers is the familiar profile: generic, over-SEO’d content with no real expertise behind it. What’s gaining is topical authority, genuine expertise, and solid UX.

If your site already took hits in March and April, treat May as another cut on the same wound, not a fresh battlefield. The fix isn’t a new tactic, it’s the work you’ve been deferring for two months.

Why e-commerce is exposed this round

Early reports name the same volatile sectors every time: finance, healthcare, e-commerce, SaaS, local services. E-commerce isn’t filler on that list. Two things stack up against it.

The first is the update itself. E-commerce sites carry a structural weakness that lines up exactly with this update’s aim. Category and product pages run on templated copy, descriptions get scraped or batch-generated, and your page looks nearly identical to three competitors selling the same SKU. Under a “relevant and satisfying” yardstick, that content loses by default. The sites pulling ahead have original reviews, real use-case breakdowns, and accumulated genuine Q&A.

The second problem is messier: your data is contaminated. This update landed on top of expanding AI Overviews and the new AI Mode infrastructure. AI Overviews now appear on about 16% of all queries, up from 6.49% at the start of 2026. AI Mode runs on Gemini 3.5 Flash by default, has passed 1 billion monthly users, and is roughly 93% zero-click.

Put those together and the headache is obvious. You open Search Console, clicks are down, and you cannot tell whether the core update demoted your rankings or AI Overviews just answered the query and the user never clicked. Both happen in the same week and produce the same-looking curve, but the causes are opposite and so are the fixes. That, not the size of the drop, is the hard part of diagnosing this one.

How to diagnose: ranking loss versus zero-click cannibalization

Pull the data by week, not as a lump sum. In Search Console, take clicks, impressions, and average position from about May 14 onward, broken out by week. Then segment by page template: product, category, and blog/comparison separately. Aggregated, everything blurs into one flat line and tells you nothing.

Now make the distinction that matters. One, a real ranking loss: impressions fall, average position slides down the page, and clicks follow. The core update moved you. Two, zero-click cannibalization: average position barely budges, impressions may even rise, but clicks and CTR drop. That usually isn’t a demotion. AI Overviews handed the answer to the user and they left satisfied. Three, the mix: you slid a few positions and lost some clicks to AI summaries at the same time.

Getting this wrong costs real effort. Treat zero-click loss as a demotion and you’ll go rewrite product copy and rebuild category pages for nothing, because your rankings never moved and copy edits won’t bring back clicks that AI intercepted. Go the other way, write off a genuine demotion as “AI’s fault, can’t help it,” and the wound keeps bleeding while you do nothing.

Cross-check before you trust a single source. Third-party volatility sensors like Semrush tell you whether the whole sector is shaking this week. If your category is lighting up across the board, that points to the update doing the moving, not your site alone. Then hand-search your top 20 or 30 keywords and note whether a big AI Overview sits above your result, which confirms zero-click siphoning directly.

What to do, and what not to do, during the rollout

While the rollout is still active, don’t make changes. Rankings bounce before an update finishes, so a drop today and a recovery tomorrow are both normal. Change something now and in two days you can’t tell whether your edit worked or the update simply settled on its own, which wastes the only clean read you get.

Things not to do: don’t panic-rewrite large batches of copy, don’t bulk-delete pages, don’t restructure the site, and don’t ship any major redesign this week. During the rollout those are all noise that makes later attribution harder. Above all, don’t tear down every product description overnight because you heard “this one targets AI content.” You haven’t even confirmed whether you were demoted or just cannibalized.

Things to do: build the monitoring. Stand up the weekly, per-template report described above and log it daily. Hand-search your most important keywords and screenshot your current position plus whether an AI Overview sits on top. Assemble a candidate list of pages whose copy really is templated and category pages that do nothing but list specs. Once the rollout ends and you’ve confirmed it’s a ranking problem, that list is your first batch of fixes.

This week is for collecting evidence, not for prescribing medicine.

Timeline and recovery expectations

The cadence runs like this. Rollout started May 21 and finishes in about two weeks, so call it early June for stabilization. Google’s own guidance is explicit: wait one full week after the rollout completes before reading Search Console data and drawing conclusions. Stack those two windows and the first point you can make a reliable call lands around mid-June.

The reason for the wait is simple. Data is noisy during the rollout, and it still hasn’t settled the moment rollout ends. The extra week lets rankings reach a new steady state. Act early and you’re building on quicksand.

Be honest about recovery. Core updates don’t have a recovery switch. Edit something this week and it won’t snap back next week. Even if you confirm a demotion, thicken the thin pages, and turn templated category pages into something real, those changes typically only count at the next core update or whenever Google reassesses your site. That cycle runs in months, not days.

One more thing people miss: fixing the ranking problem won’t bring back the zero-click losses. AI Overviews on 16% of queries and an AI Mode that’s 90-plus percent zero-click is a structural reallocation of traffic, on a separate track from whether this core update hit you. Fixing rankings is one job. Responding to zero-click is a different game. Don’t expect one edit to solve both.

阅读本文中文版: Google 2026年5月核心更新：电商站又一次震荡，这次该怎么诊断

Related Articles