How Google Ranks: Ranking Systems, Core Updates and E-E-A-T

Version note: This post is based on Google's "A guide to Google Search ranking systems" and core update pages as of November 2026, the Search Status Dashboard records, and the Search Quality Rater Guidelines dated September 11, 2025.

Guess First

Bookstore's blog has a long review of the book "Async in Practice". On the second day of the core update that started on March 27, 2026, someone on the team looks at Search Console and raises the alarm: the review's impressions have dropped. The product manager suggests two things: add a "author: 15 years of experience" badge under the post, because "E-E-A-T is a ranking factor", and immediately delete the older posts whose rankings dropped. Which one is right? Neither? On which day should the analysis start?

What Does "Ranking System" Mean?

SEO (Search Engine Optimization) conversations talk about one thing called "the Google algorithm". Google's own documentation uses different language: results are produced by many automated ranking systems that sort hundreds of billions of pages in the index in a fraction of a second. Some of these systems are part of bigger systems, some are specialized systems that kick in for particular query types.

Two details matter. First, the systems mostly work at page level, but site-wide signals and classifiers also contribute to how a page ranks. Second, in Google's own words, good site-wide signals don't guarantee that every piece of content on a site ranks highly, and poor ones don't mean every piece ranks poorly. So the sentence "our site got penalized" is usually too coarse a diagnosis.

The Systems in the Ranking Systems Guide

Google publishes a guide that lists the systems it uses by name. For a developer they don't all matter equally; the table below puts the guide's definition and what it means on your side next to each other:

The guide also has crisis information, local news, exact match domain, reliable information and removal-based demotion systems. Google's note on MUM (Multitask Unified Model) is interesting: it isn't used for general ranking, only in specific applications.

The guide's "retired" section is instructive too. Panda was announced in 2011 and became part of the core ranking systems in 2015. Penguin was announced in 2012 against link spam and joined core in 2016. Hummingbird was a major improvement made in August 2013. The helpful content system, announced in 2022 as the Helpful Content Update, became part of the core ranking systems in March 2024. So after 2024, "we were hit by the helpful content update" no longer points to a separate event; those signals change together with core updates.

Core Updates: What Changes and What Doesn't?

Several times a year Google makes broad changes to its search systems and calls them core updates. Google's emphasis: core updates don't target specific sites or pages; the goal is to show helpful and reliable results overall. A page that drops in a core update may drop not because it "did something wrong" but because other pages are now assessed as better.

According to the Search Status Dashboard records, recent ranking events were as follows (times in Pacific Time):

• December 2025 core update: December 11 – December 29, 2025
• March 2026 spam update: March 24 – March 25, 2026
• March 2026 core update: March 27 – April 8, 2026
• May 2026 core update: May 21 – June 1, 2026
• June, August and September 2026 spam updates: June 24–26, August 18–20 and September 24 – October 7, 2026 respectively

Use these dates not to memorize them but to put them next to your own data. In periods like March 2026, when a spam update and a core update started a few days apart, tying the change you see to a single event gets harder.

After a Drop: The Analysis Google Recommends

Google's core update page spells out step by step what to do when traffic drops. Let's apply it to Bookstore:

First wait for the update to finish. On the Search Status Dashboard, the March 2026 core update ends on April 8. Measuring while the update is still rolling out means looking at a moving target; the teammate raising the alarm on day two is too early.

Wait at least one full week after completion. Then, in the Search Console Performance report, compare a week after completion (April 9–15) with a similar week before the update started (March 20–26). Because the days of the week match, weekend effects balance out on their own.

Look at the Web, Image, Video and News search types separately. The drop might be only in image search; that leads to a very different place than a diagnosis about text content.

Look at the most affected pages and queries. Google recommends not doing anything drastic for small position changes, and for large drops, assessing the site as a whole using its own helpful content self-assessment questions. If possible, get an honest opinion from someone not involved in the work.

Avoid quick fixes and treat deletion as a last resort. The product manager's "let's delete the old posts right away" is exactly what Google doesn't recommend. On the effect of changes: some show up within days, but it can take months for Google's systems to confirm a lasting improvement. If there's still no effect after a few months, the next core update may be an opportunity; smaller, unannounced updates in between can also bring change. And Google says it plainly: no outcome is guaranteed.

E-E-A-T: What It Is and What It Isn't

E-E-A-T is the initials of Experience, Expertise, Authoritativeness and Trustworthiness. According to Google's "Creating helpful, reliable, people-first content" page, its systems look at a variety of factors that help identify content demonstrating these aspects. Of the four, trust is the most important. Not every piece of content has to show all four: a book review can stand out through the experience of an author who actually read and applied the book, a post about a compiler through expertise.

And the critical sentence, from Google itself: E-E-A-T itself isn't a specific ranking factor. The Search Engine Optimization Starter Guide also answers the question "is E-E-A-T a ranking factor?" with a direct "no". That's why the badge the product manager suggested is not a signal switch; it only makes sense if it is genuinely accurate and useful information for the reader.

On the same page Google suggests evaluating content with Who, How and Why questions. Who: who wrote the content, and is that clear? A byline leading to the author's name and background is recommended; fabricated author profiles are explicitly called deceptive. How: how was the content made? If automation or AI was used substantially, disclosure is recommended where readers would reasonably expect it. Why: why was the content made? Google calls this the most important question; does the content exist primarily to help people, or to attract search traffic?

The term E-E-A-T actually comes from the Search Quality Rater Guidelines. Google uses external quality raters to understand whether its results are good, and these people are trained on E-E-A-T. But Google states two things explicitly: raters have no control over how pages rank, and rater data is not used directly in the ranking algorithms. The raters' job is to produce feedback showing whether the systems work well. The latest version, dated September 11, 2025, added examples for rating AI Overviews and widened the YMYL (Your Money or Your Life: topics such as health, money and safety that affect a person's life) scope to include election and voting information. Google writes that it gives extra weight to reliability on YMYL topics.

The Unknowns: Weights, Page Experience and "200 Ranking Factors" Lists

Google publishes the names and purposes of its systems, not their weights. There is no public, Google-sourced table of how much each signal matters for which query. That's why none of the "200 ranking factors" lists circulating online is a Google document; they are third-party compilations. Google's third-party SEO guidance, published in June 2026, touches exactly this point: claims presented as "Google says" should be verified if they don't link to Google's official documentation, and third-party tools have no access to Google's internal ranking data. If a claim isn't in Google's documentation, it should be read, as in this post, as "Google doesn't confirm this".

The situation is similar for page experience. Google's page experience page says there is no single page experience signal, that the ranking systems look at multiple signals, and that Core Web Vitals are used by the ranking systems. But the same page also says Google always tries to show the most relevant content even if the page experience is sub-par, and that good Core Web Vitals don't guarantee a top position. The technical improvement of Core Web Vitals is the subject of the upcoming Web Performance series.

Production Notes

Tie the Search Status Dashboard to your calendar. The only way to separate the effect of a deploy, a content change or a core update is to put the dates side by side. Keep a dated log of your own deploys and content changes.

Always compare ranges of the same length that contain the same days of the week, and separate the search types. This prevents the most common misreading in Search Console.

Keep author information real. The byline, the author page and the update date exist because they are useful to the reader; a fabricated author profile is what Google explicitly calls deceptive.

Common Pitfalls

Deciding while the update is still rolling out. An analysis done before the rollout completes and a week passes measures noise.

Turning E-E-A-T into a checklist. There is no "add an author box so the E-E-A-T score goes up"; Google says plainly that E-E-A-T itself is not a ranking factor.

Thinking quality raters decide rankings. Raters have no control over the ranking of individual pages.

Blaming every drop on a core update. A robots.txt change made the same week, a canonical bug or the site getting slower is a far more ordinary and far more frequent cause. Rule out technical causes first; we'll build the way to do that in the sixth post of the series.

A Small Experiment

★☆☆ Find the start and end dates of the last two core updates on the Search Status Dashboard and write down the two one-week ranges you would compare using the method in this post.

★★☆ Evaluate the three highest-traffic pages of your own site against Google's Who, How and Why questions. For each question, point to concrete evidence on the page that the reader can see.

★★★ Put your own site's deploy and content change dates in a table and add the ranking events from the Search Status Dashboard next to them. Can you tie a change in Search Console to a single cause by looking at this table, or do two events overlap?

We've seen how the ranking is built. In the next post we'll look at the ranked result itself: where the title and description Google shows come from, title links, snippets, meta descriptions and site names.