Google's May Quality Updates

Google's May 2017 quality update is an algorithm update seen on 17 May 2017 that hit sites with weak content quality and user experience. I wrote this article on 28 May 2017, so its dates and observations belong to that period. Update (October 2026): Google announces broad changes as core updates and publishes their start and end dates on the Search Status Dashboard.

2017 has been the year we have seen Google algorithm updates most often. There was one at the start of January, the 7 February update, Fred on 7 March, then updates at the end of April and the beginning of May. Most recently, on 17 May, we witnessed a big one.

The 17 May update was first reported by Barry Schwartz on the Search Engine Roundtable blog.

Since 17 May I have been digging into the update and analysing the sites it touched, both the ones that gained and the ones that lost. Non-disclosure agreements mean I cannot share the raw data, but I can share everything I found: what was done wrong, what was done right, and how the update played out.

What was the rule again? Build for people, optimise for robots. Remember that real human beings visit your site, and treat them with respect. In my reviews I saw plenty of aggressive advertising, UX obstacles, interfaces that irritate the visitor, deceptive ads and low-quality content. A site that committed those mistakes aggressively took the hit and lost visibility.

After the analysis and testing we improved that site and brought it back to where it was. Of course I cannot claim these are 100% of the factors at play.

So let us pull it together: here are eight things you can do to avoid being hit by the latest update.

• Review your site objectively in terms of content quality and user experience.
• Improve your low-quality content. Find what competitors are doing more of than you and apply it to your own site.
• If you cannot improve a low-quality page, use noindex. Pages that are not indexed cannot hurt you.
• Avoid aggressive advertising. If your ads disturb users, Google is able to work that out.
• Fix every significant UX problem on the site. Walk through your own site as a visitor and try to reach the information you are meant to find.
• Look for technical problems that cause content or UX issues. Page performance and technical SEO matter.
• Keep adding quality content. What you publish should be unique, relevant and genuinely answer a user need. If you cannot do that, do not publish at all.
• Check the queries that lead to your content and analyse the user intent to meet those expectations. Expand your content for the different queries it can serve.

Updates have been frequent in 2017, and I expect Google to increase the pace of quality updates further. We will wait and see 🙂 Update (October 2026): Google states that it makes broad core updates to its search algorithms several times a year.

What is the Google May 2026 core update?

The Google May 2026 core update is the broad core update that Google released on 21 May 2026 and finished rolling out on 2 June 2026. In October 2026 the newest entry on the Search Status Dashboard is the September 2026 spam update, which completed on 8 October 2026. Google states that core updates do not target specific sites or individual pages. A page that drops is not necessarily bad, because other pages may simply help the searcher more.

What should you do after a core update drop?

After a core update drop, Google recommends measuring the change with the right dates before you touch any content. The check has four steps:

1. Wait until the rollout has finished.
2. Wait at least one full week after the update completes. Then compare that week in Search Console with the week before the rollout began.
3. Review your top pages and queries to size the drop. Google calls a fall from position 2 to 4 small, with no need for drastic action. A fall from 4 to 29 is large and calls for a self-assessment of the whole site.
4. Rule out other causes. Google's guide to traffic drops lists technical issues (server availability, robots.txt, a misplaced noindex), security issues, manual actions and seasonality as separate causes.

Google says some changes take effect within days, but a full recovery can take several months while its systems confirm that the site now produces helpful content. When a few months pass with no effect, the site may need to wait for the next core update.

How does Google judge content quality today?

Google judges content quality today by whether a page is helpful, reliable and made for people rather than for search rankings. The helpful content system, announced in 2022, became part of Google's core ranking systems in March 2024. Those systems look for signals of experience, expertise, authoritativeness and trustworthiness (E-E-A-T), with trust as the most important.

Google's self-assessment questions ask whether content shows first-hand expertise, and whether it mainly summarises what others say without adding value. That second question describes what SEOs call thin content. The helpful content section of my SEO guide covers E-E-A-T in more depth.

What do the Google leak and the DOJ case show about site-level quality?

The Google leak and the US Department of Justice (DOJ) case show that Google's quality signal has a site-level part. In a February 2025 call, Google engineer HJ Kim spoke with Google's lawyers and expert. The DOJ published the call notes as an exhibit in its search case against Google. The Kim call notes define Q* as page quality, "the notion of trustworthiness", and call it "incredibly important". They add that Q* is "largely static and largely related to the site rather than the query".

The leaked Content Warehouse documentation has attributes at both levels. Its CompressedQualitySignals module, which holds compressed per-document signals, contains these four:

• siteAuthority: its description says it is applied in Qstar, the leak's spelling of Q*.
• pandaDemotion: the encoding of Panda fields from a structure named SiteQualityFeatures.
• babyPandaV2Demotion: a newer BabyPanda demotion, applied on top of Panda.
• pqData: encoded page-level PQ signals, kept next to the site-level ones.

After a quality update, review the whole site, not only the URLs that lost positions, which matches Google's advice for a large drop above. My guide measures the quality spread across your pages with a unique-word count. Kim's notes say the leaked documents named components of Google's ranking system but "don't go into specifics of the curves and thresholds". The weight of each attribute and its role in a core update are unknown.

How do AI Overviews and AI Mode fit into a core update check?

In a core update check, AI Overviews and AI Mode fit in like the rest of Google Search. They follow the same SEO best practices as the rest of Google Search, and Google lists no extra requirements for them. Search Console counts traffic from both features in the Performance report under the Web search type. The before-and-after comparison above already covers both features. My guide to measuring AI search visibility covers the other data sources.