The February 2026 Discover Core Update

If you get meaningful traffic from Google Discover, you have probably seen volatility over the past few weeks. Google has now confirmed and completed the rollout of its February 2026 “Discover core update”, which it described as a broad update to the systems that surface articles in Discover.

Before you rewrite your whole content strategy, there are a few checks worth running and a few changes worth not making.

What Google announced (and what they did not)

What we know

• Google released a February 2026 Discover core update.
• It is described as a broad update to how articles surface in Discover.

What we do not know (and should not assume)

• That a single “issue” explains all drops.
• That the fix is a technical tweak.
• That changes that help Search rankings will automatically help Discover.

Discover is not “Search without keywords”

A useful way to explain Discover to clients and stakeholders:

• Search responds to an explicit query.
• Discover predicts what someone might want to read next.

That matters because Discover tends to reward:

• strong topical relevance
• clear packaging (headline, imagery, summarised value)
• perceived trust and reader satisfaction

The 5 checks to run this week (Search Console + site review)

• In Search Console, pull Discover performance separately from Search performance.
• If overall “organic” is down, confirm whether it is Discover-only, Search-only, or both.

Use 3 windows:

• a pre-update baseline (for example, 4 weeks before rollout)
• the rollout window
• a post-rollout window (at least 7 to 14 days)

Make a quick list of:

• top gaining URLs
• top losing URLs

Then label each by:

• format (news, opinion, evergreen guide, listicle, how-to)
• topic
• author
• freshness (new, updated, old)

For the top losing URLs:

• Does the headline promise a clear benefit or angle?
• Is the opening paragraph doing work, or meandering?
• Is the primary image high quality and relevant?

For content that represents expertise:

• Is it obvious who wrote it?
• Is there a short author bio with relevant credentials or experience?
• Are claims supported, and is the page kept current where it matters?

What not to change yet (common self-inflicted damage)

• Do not delete large swathes of content based on a week of data.
• Do not mass-edit headlines sitewide.
• Do not “AI refresh” everything with light updates and generic copy.
• Do not change templates, internal linking, and navigation all at once.

A simple test plan for the next 21 days

Pick one hypothesis at a time.

Test A: Packaging refresh for a small set

• Select 5 previously strong Discover URLs that have dropped.
• Improve headline clarity, the opening “why this matters” paragraph, and image relevance/quality.

Test B: “Explain it like I’m busy” structure

• Add a short summary box near the top: 2 to 5 bullets, key takeaway first.
• Keep the rest of the article intact.

Test C: Freshness discipline

• For evergreen pages that still get Discover visibility, add a “Last reviewed” date.
• Update genuinely outdated sections.
• Add one new, concrete example.

A client-ready explanation (2 sentences)

Google updated how it selects articles for Discover, and Discover can swing more than Search because it is prediction-based rather than query-based. We are separating Discover from Search, identifying which content types moved, and running small controlled tests so we can improve performance without accidentally making things worse.