Discover's first core update bites, WordPress 7.0 hits Beta 3, and CSS shape() goes Baseline.

The first week of March opened with Google's monthly Webmaster Report and a hangover from the first-ever Discover core update — three-plus weeks of rollout, a confirmed serving bug muddying the diagnosis, and listicle-heavy sites nursing real losses. Meanwhile rankings kept twitching with no confirmed search update, a quiet pilot handed a handful of publishers control over their Discover profiles, WordPress 7.0 reached Beta 3 on schedule, and the February web platform roundup delivered a genuinely useful CSS milestone. If your website is how customers find and choose you, here's what happened between 2 and 8 March — and what to do about each piece.

1. The March Webmaster Report: volatility with no confirmed update

What happened

Search Engine Roundtable's March 2026 Google Webmaster Report, published on 2 March, summed up a familiar 2026 story: ranking volatility stayed heated into March with no confirmed search update behind it. Google's rankings have been in near-constant motion since early January, and the tools tracking SERP movement barely cooled as February turned into March. Google confirmed nothing for classic search this week — the movement is just how the algorithm behaves now, breathing between named updates rather than only during them.

The triage sequence when rankings move and Google says nothing

• Separate rank changes from click changes in Search Console. A position drop and a click drop have different causes and different fixes.
• Check the query class. Losing brand queries is rare and alarming; losing competitive commercial queries is ordinary churn; losing informational queries is often AI answer expansion, not a penalty.
• Compare competitors on the same queries. If the whole niche moved, it's weather. If only you moved, audit your last 30 days of changes.
• Rule out the boring causes — deindexed pages, broken redirects, a slow deploy, an expired plugin — before blaming the algorithm.
• Wait a full week of data before rewriting or restructuring anything.

2. Discover's first-ever core update: listicles lose, and a bug confuses the picture

What happened

Google's first-ever Discover core update — announced in February — took over three weeks to roll out, and its effects dominated early-March conversations. The update deliberately reduces sensational, clickbait-style content in the Discover feed in favour of in-depth, original, timely content from sites with demonstrated expertise in their area. The clearest casualty class: shallow listicle content. Publishers built on low-depth numbered lists reported lasting traffic losses that, per Google's framing, are policy rather than accident — they will not self-correct.

Complicating the diagnosis, Google also confirmed a brief Discover serving bug in the same window. That left publishers staring at overlapping causes: a temporary technical glitch that recovers on its own, and an algorithmic reset that doesn't.

How to tell the bug from the algorithm

• Check the shape of the drop. A sharp cliff that recovers within days points to the serving bug. A step down that holds for two-plus weeks points to the core update.
• Segment by content type. If losses concentrate in list-format and trending-topic posts while reported, original pieces held, that's the update working as designed.
• Treat Discover as a bonus channel, never a business model. It has always been volatile; it's now volatile with an editorial opinion.

3. A quiet pilot: publishers get control of their Discover profiles

The other Discover story from the report is more interesting long-term: starting in March, Google began quietly letting a small, hand-picked set of publishers control their Discover profiles — custom banner images, a configurable links shelf, pinned posts, and control over the order of social links, website and content tabs. There's no application form and no Search Console toggle; it's an invitation-only pilot.

Why it matters even if you're not invited: Google is signalling that a publisher's presence in its surfaces is becoming a managed brand asset, not just a stream of URLs. The obvious preparation is the unglamorous kind — consistent organisation schema, a real about page, verifiable authorship, and social profiles that match your site. When features like this widen, the sites with clean entity signals get them first.

4. WordPress 7.0 Beta 3 lands — your redesign timeline just got a date

WordPress 7.0 Beta 3 shipped on 5 March, keeping the release on schedule, with Release Candidate 1 slated for 19 March. This is the first of the flagship-event-aligned major releases in WordPress's new 2026 cadence, and it's the release most agencies will anchor spring rebuilds around.

What to do this month if you run WordPress

• Stand up a staging copy and install the beta now. The point of the beta window is to find your theme and plugin breakage before your customers do.
• Audit your plugin stack. Every major release strands a few abandoned plugins. Fewer, better-maintained plugins beat a drawer full of freebies.
• Don't upgrade production on day one. Let the .0 release breathe for a week or two, then move.
• If a redesign is on your 2026 roadmap, building it on 7.0 from the start beats migrating a fresh 6.9 build six months later.

5. Web platform roundup: CSS shape() goes Baseline, real columns get fixed

The web.dev February platform roundup, published as the month closed, confirmed a strong four weeks for CSS. Chrome 145, Firefox 148 and Safari 26.3 all reached stable in February, and between them:

• The CSS shape() function reached Baseline newly-available status. You can now define custom shapes for clip-path and offset-path in readable syntax — with real CSS units, calc() and CSS variables — instead of hand-computed polygon coordinates. Responsive angled sections and curved image masks just got dramatically cheaper to build and maintain.
• Chrome 145 shipped full text-justify support, giving designers proper control over justified text — long a print-only luxury.
• Chrome 145 added column-wrap and column-height from Multicol Level 2, letting columns wrap into new rows instead of overflowing horizontally — a real fix for magazine-style layouts on small screens.

What this week means for your website strategy

Three threads worth pulling together if you own a business website or are scoping a build this quarter:

• Google is editing its surfaces, not just ranking them. The Discover update punished shallow formats and the profiles pilot rewards coherent brands. The sites that win both are the ones with genuine depth and clean entity signals — decisions made at design and build time, not bolted on later.
• Platform timing is strategy. With WordPress 7.0 weeks away and CSS capabilities landing monthly, the "when" of a rebuild now materially changes what you get. The same logic applies in commerce: Shopify development partners who track platform releases ship stores that age well; a shopify developer partner who doesn't will have you rebuilding in a year.
• Judge builders on evidence, not proximity. Whether you're searching "web developers near me" or briefing a remote team, the questions this week hands you are sharp ones: how do they diagnose a Discover drop, how do they stage a major CMS upgrade, and does anything they build move revenue for a business website — not just traffic.

Frequently asked questions

Why did my Google Discover traffic drop in early March 2026?

Two overlapping causes: the first-ever Discover core update, which finished its three-plus-week rollout and deliberately reduced sensational and clickbait-style content, and a brief serving bug Google confirmed. If your losses concentrate in shallow listicle posts and have held for two weeks, that's the update — and it won't self-correct without deeper, more original content.

Was there a confirmed Google core update in the first week of March 2026?

No. Rank trackers stayed heated into early March, but Google confirmed nothing for search that week — it was a continuation of the near-constant movement seen since January. Diagnose before acting, and give any change a full week of data before you touch content.

Should I install WordPress 7.0 Beta 3 on my business site?

Not on production. Beta 3 (5 March) exists for testing, with RC1 scheduled for 19 March. Put it on a staging copy now to surface theme and plugin breakage early, and upgrade production only after the stable release has settled.

What is the CSS shape() function and can I use it in production?

It defines custom paths for clip-path and offset-path in plain CSS syntax — units, calc(), even variables. It reached Baseline newly-available in February 2026 across Chrome, Firefox 148 and Safari, so yes: use it now, with a simple no-clip fallback for older browsers.

Do listicles still work for SEO after the Discover update?

Thin ones don't. The update targeted low-depth, sensational list content specifically. List formats still perform when every entry carries real analysis, testing or first-hand expertise — Google punished the emptiness, not the format.

The takeaway

This week's theme is editorial judgment moving into the machine. Discover now has an opinion about depth. Profiles are becoming managed brand assets. Even the CSS pipeline is rewarding teams who keep up. None of it favours shortcuts — all of it favours websites that are genuinely well-made, clearly authored, and built by people watching the platform move week to week.