This is The Biggest Google Search Algorithm Leak in History. Here's What it Means for Bloggers

She Knows SEO contains affiliate links and is a member of the Amazon Services LLC Associates Program. If you make a purchase using one of these Amazon links, we may receive compensation at no extra cost to you. See our Disclosure Policy for more information.

“In the last quarter century, no leak of this magnitude or detail has ever been reported from Google’s search division.” – Rand Fishkin, CEO of SparkToro and founder of Moz.

I thought I’d get a quiet Tuesday morning with a coffee and some puppy snuggles. Instead, I got 2,500+ pages of Google algorithm factors dropped in my lap.

Today, Rand Fishkin and Mike King, iPullRank’s CEO, released articles about a leak of Google Search’s Content Warehouse API. They got the data from a source on May 5th, which Rand spent weeks reviewing, corroborating, and verifying with other SEO and former-Google sources.

The source was initially anonymous, then revealed themselves to be Efran Azimi of EA Eagle Digital.

I did my best to review as much of the 14,014 attributes (think of them as “ranking features”) that were leaked and put in a Github repository, as well as reading the original articles by Rand and Mike. But I’m just one woman, on her second day back from holiday, with minimal tech SEO knowledge.

But this is too important to leave waiting for weeks.

So today I want to review what this leak means for bloggers, specifically, and how to proceed.

UPDATE: Google confirmed the leak on of March 29th, 2024.

The Sky Isn’t Falling

Before we get started, I’ll take a page from Mike’s book and note some things y’all need to keep in mind as we go through this review:

I’ve had 1 day with this material – Much like every other SEO (aside from Mike and Rand), this was an unexpected bombshell dropped in my lap. I was busy working on an analysis of the 211 sites accepted to Journey by Mediavine before my therapy appointment when this popped up on my Twitter feed. I’m sure we’ll have more information come out soon as more SEOs review the material.
I am NOT a tech SEO – Contrary to the very incorrect popular belief, I do not know everything. And if there’s anything I really don’t know, it’s tech. I’m using my understanding of basic tech, my interpretation of the dataset and Mike’s tech SEO analysis, and some context from smarter tech SEOs than myself from Twitter. I’m largely drawing on my understanding of how Google works from years working in SEO, and from what I’ve seen work/not work with my own tests over that time.
Existence does not guarantee efficacy – Just because this data exists, doesn’t mean it’s up to date. It also doesn’t mean that every attribute existing in the API data is in use. How many of us have half written blog posts taking up space in our Google Drive? It could very well be similar. Rand noted that since this document is for internal use, and it seems to show attributes that are no longer in use, that it’s likely it’s pretty up to date. I’m not discounting that, but we can’t be 100% sure what’s being used.
Don’t be hateful – Bloggers and SEOs have been especially upset lately, with Google showing pretty blatantly that they don’t care about small publishers. Between the March Core Update, the Google I/O 2024 Keynote, and the shitshow of AI Overviews, most people are rightfully pissed. But that doesn’t mean you should send hate to anyone. Remember that employees of Google are employees and likely have to abide by certain rules of what they can/can’t say. You can be critical, but don’t be a bully or hateful just for hate’s sake.
There are no “quick hacks” here – I’m sure many people expected a leak like this to result in a bunch of quick hacks that can be exploited to hit 1 million pageviews in a day. That’s not what this article is about, and not what any of the reviews I’ve mentioned so far offer either. Ultimately, we only have a piece of the Google puzzle and it has a lot of gaps. I’m also not in the business of scammy hacks that work for 5 mins then get your site shut down. So that’s not what you’ll find in this article. I’m sure there are others out there if you prefer that.

Google API Leak Cliff’s Notes

When: Rand received the information on May 5th, 2024 and released it publicly May 27th, 2024. However the API information seems to have been leaked on March 13th, 2024 on Github.

What’s Inside: 2,596 Github modules with 14,014 attributes (think of them as named ranking factors).

Who: Rank Fishkin, Mike King (also referred to as Michael King, but I’m going with the author name from iPullRank), and Efran Azimi shared this information.

For legal purposes, I’m not going to share a link directly to the Github. But I will share one to a tweet that may or may not link to the Github with the leaked data.

(If, like me, you don’t spend a lot of time in Github repositories, you’ll find it pretty incomprehensible. I had to spend a while re-familiarizing myself with the platform and the syntax.)

Interesting Ranking Factor Finds

I’m not going to go too in depth on the individual ranking factors in the document. There are a LOT.

But here are some that I think are especially interesting and important for bloggers:

There’s an attribute that specifically marks small personal sites/blogs. It’s unclear what it’s for
Google has a siteAuthority score
NavBoost is almost entirely made up a click data, such as longest click from the SERPs (the search result the user spent the longest time on) and last good click (the last time someone went to your site and stuck around). It tracks clicks over 13 months
Google tracks Chrome visits across the web to determine user behaviour
Twiddlers are re-ranking algorithms that run between major updates and shift rankings in the SERPs
Over optimized anchor text – especially when used constantly for links to your site from 3rd parties – seems to trigger a spam demotion for those links
Poor navigation and exact match domains can lower your rankings
Google stores at least the last 20 versions of your web pages. So to have a “clean slate” you probably need to update it more than 20 times. It’s unclear how significant of a change the page would need to count as a new version
Font size for links and weight for text are attributes Google tracks. Having larger links seems more positive and Google seems to read bolded text differently than normal text (it’s also better for accessibility!)
Google has a keyword stuffing score
Google tracks when domains expire, so they either already can or soon will be able to find expired domain abuse
If a site has videos on more than 50% of it’s pages, it becomes classified as a video site. It’s unclear if these videos need to be indexed, be in a certain area of the post, or need to be natively uploaded to the website
YMYL content has its own ranking score
There is a “gold standard” attribute that seems to demarc human generated content, but it’s unclear how it’s triggered
AI Overviews are never mentioned in the document
There is no clear mention of internal links as an attribute
Three topics have “whitelists” (aka they need to be approved to be shared). These are travel, Covid, and politics. It’s unclear if the travel sites are for the general SERPs, Google’s “travel” section, or even the widgets they pull. But it’s interesting travel is the only non-YMYL niche covered here. I do wonder if it’s just a left over from the Covid lockdowns and if it may be defunct

What Does the Google API Information Mean for Bloggers?

1. Create For Users, Not For Google

The #1 takeaway I have from this documentation is that Google is looking for the best content and isn’t really sure how to find it.

To the best of their ability, they’re trying to rely on users by focusing on click data and dwell time.

I sent a recent email about my time at Affiliate Gathering 2024, where the main undertone of the presentations was “build a brand”.

Building a brand doesn’t mean having 20 social media channels with 10k followers, like some people thought post-HCU.

It means building a site so good that your audience find you no matter what.

I often liken this to the Vine stars from back in the day. Once Vine closed, many people became obsolete. But the stars who had a community were able to transition to other platforms like Youtube and now Tiktok.

You need to make your brand platform-proof.

And a key way of doing that is to build a really strong community where you help your audience.

“But, Nina, my content is amazingly helpful but no one can see it because Google isn’t showing it. What do I do?”

Diversify!

If Google isn’t serving you, find another platform.

I’d even recommend diversifying much earlier – around the 10,000 pageview/month mark now versus the 30,000 pageview/month I used to believe in.

This needs to be a traffic-attraction platform. Newsletters are great, but they don’t naturally attract an audience, so they don’t count towards this.

My two prime methods for traffic acquisition for new users are SEO and Facebook Groups.

But others include Pinterest, Flipboard, Instagram, Twitter, Facebook Pages, Paid Ads, Tiktok, and Youtube. (Youtube is steadily becoming a third prime method for me, especially for this site.)

2. User Experience is King

Google clearly values user signals.

Other than demotions for keyword stuffing, keywords weren’t very prevalent in the API documentation.

But user-based attributes were.

Clearly, Google cares about users using your site and actually enjoying it.

We often talk about user experience (UX) in the SEO world, but usually more to do with things like page loading time.

It comes up once in a while for UX issues relating to ads or newsletter pop ups.

But too often people, SEOs in particular, forget about users getting what they need out of the content.

If Google is tracking – no matter how weighted these attributes are – the clicks to your site, time on your site, and longest time spent on a result from the SERPs, bloggers need to make sure we’re doing everything we can to give the users a great experience.

That means getting the point immediately, having freshly updated content with accurate information, making it clear why they can trust you to help them, and providing unique content that’s not the same as everyone else in the SERPs.

A lot of this will come down to a strong introduction to keep the user from clicking back to the SERPs in just a few seconds – which gives you a terrible bounce rate and shows Google you weren’t helpful to that user.

Before you ask, yes, the technical SEO components still matter for UX as well.

Having a site that loads in 12 seconds isn’t going to be great for the user. Making the content impossible to read because it’s buried in ads or the page keeps shifting to load more ads will send them away immediately. Never actually getting to the point, or having clickbait nonsense, will make them leave.

3. Get the Right Type of Links

Within an hour of waking up, I saw 10 link building agencies/service providers tweet about how links really do matter for rankings.

Again, these are potential ranking factors, not guaranteed ones.

The leaked documents do suggest that certain links may help, but are much clearer that spammy links will hurt you.

It seems like Google is favouring links from established sites, with high PageRank, using proper anchor text, and on popular pages that get traffic.

Basically, DA 90 links from a constantly updated news site that gets lots of viewers each month are amazing, but from a DA 90 site that hasn’t been updated in 10 years and gets 1 viewer a month aren’t.

Skip the scammy links with high DAs.

Go for relevant, helpful links.

Basically, proper digital PR is the way to go, with links from news sources. HERO, HARO, Connectively, and other similar platforms will help you with this.

Or outsource to Linkifi like I did to have them do the hard work for you.

I’ve already stopped doing link swaps and collab posts a ways back (time constraints mostly), but I’d definitely avoid them now with this information.

I’d focus on super high quality link opportunities that are meant to put you in front of your target audience on a popular page/site where the website is incredibly active.

I’m not 100% sure if guest posts are a good or bad idea anymore. A lot of people thought the manual penalties in March were a sign they were done, but I haven’t seen clear indications that they hurt sites.

I definitely wouldn’t do them to “game” the system. I’d treat them more like networking or freelance writing to build a portfolio (but you get paid in links, instead).

Be discerning about where you’re building links, if you’re still doing so.

4. SEOs Got it Wrong

The API revealed that a lot of things SEOs like myself thought were ranking factors actually aren’t.

For example, there is nothing in the documentation specifying a word or character count (beyond one for a snippet Prefix). So page/post titles, meta descriptions, and URLs can kind of be as long as you want.

EXCEPT that it’s bad for click through rate.

So in an indirect way, I would still argue that things like page title, URL slugs, and meta descriptions do have a maximum length to be helpful to users.

If a user can’t read the full page title, they’re unlikely to click through to it.

The leaked documentation does suggest that including the keyword in your title is scored by Google, but anything else about titles and metas really comes down to what’s best for the user.

Don’t obsess over these numbers, but do keep them in mind to improve your click through rates.

Similarly, EEAT is kinda real, kinda not.

The attributes never refer to EEAT, or explicitly to any of the individual elements.

Rand and Mike both note that they couldn’t find any specific attributes related to things SEOs associate with “showing EEAT” – like original images or personal experience.

The author-related attributes seem to be the only connection to EEAT.

Essentially, Google tracks authors across the web and within a website. With sameAs schema, Google can also tell if an author is the same person across different websites.

Make sure you have an author bio and author archive set up on your site for each of your authors, and connect it to their social media or other relevant bios.

“I’m a bit worried that E-E-A-T is 80% propaganda, 20% substance,” says Rand. And I’m inclined to agree.

But I’ve long said that I don’t care of EEAT is “real” to Google, or not. I care that it’s useful to readers.

Readers will trust you more with original images, clear personal experience, and the reason you are capable of helping them (i.e. any qualifications).

If Google doesn’t care, screw em! I know my users do.

For example, you probably wouldn’t care about this post if I hadn’t mentioned I reviewed all the documentation up front, right? I wouldn’t either. That’s why I made it clear.

5. Testing Matters Most

This leak has made it clear that what bloggers have been told by Google’s spokespeople, like Google Search Liaison and John Mueller, is not 100% accurate.

I don’t know whether they had this information and were withholding it, or whether they weren’t informed. But either way, the Google teams have been actively spreading misinformation over the years. And many bloggers make the mistake of blindly listening to them.

My SEO knowledge has never come from them.

It’s come from my own tests or the tests of other SEOs that I trust.

That means we get things wrong sometimes, like when I indexed 20 posts to prove John Mueller wrong about the HCU and trashed my site’s rankings (but did succeed in proving him wrong that content written with no understanding of SEO would rank and help sites recover – this is a reductive summary of what happened, but you get my point).

But it also means we discover a lot that Google may not want us to know. Like Rand figuring out that Domain Authority, the Google sandbox, and dwell time affect rankings in the SERPs.

Test test test! And keep testing!

Doing comparative analysis and running real time tests are the only ways to learn how algorithms work.

It’s like a brand new video game: you have to play it to figure out the map and all the potential cheat codes.

6. Track Your Content Like Google Does

The leaked API data is all about user interaction with a website. So we need to track users’ interacting with our site.

If dwell time, clicks, and internal link clicks matter to Google, we should understand those metrics on our site.

Unfortunately, G4 is kind of a nightmare to do that with.

I recommend installing Microsoft Clarity. It’s a free tool that runs live heatmaps, tracks user interactions in video-form, and captures way more info at a glance than G4.

Figure out where users are leaving your content, and try to make them stick around.

See what links are getting clicks, and try to recreate that elsewhere to increase time on your site.

Find broken links or javascript to make user experience better.

Basically, run a content audit of your site and a user experience audit. Make sure users can find content easily, get the helpful information they want, and keep them on your site by serving them even more helpful information in other posts.

7. Age Matters

Domain age, content age, and the age of sites linking to yours all matter.

But in different ways.

The older and more established your site is, the more Google seems to trust it. This comes both from the attributes in the leaked documents and from what I’m seeing in the SERPs time and time again.

Google is kind of like a law firm: they trust you more about you once you have some grey hairs.

So older sites tend to rank higher. They’ve got a history for Google to draw on (and usually, going back to point #1, have built that branded community that seek them out).

But old isn’t always good.

Google wants fresh, new content.

Keep your content updated, and make it clear that the content is new. Google’s attributes look at content dates when evaluating the content.

You also want to make sure the date is clear and consistent. Don’t have an 2019 date in your URL but the blog post says it was published in 2024. That’s gonna confuse the algorithm and your user.

8. Get to the Point

I rewrote this article a few times to get to the point as quickly as possible.

I know it’s still a bit rambly, but I’m still jet lagged from Italy, so give me a break!

The attributes show that Google’s tokens prioritize the first part of a blog.

There have been theories since the March Core Update that Google can only read the first 1,000 words (sometimes even 1,000 characters) of a blog post.

Even if it can read more, it does seem like the limited number of tokens referenced in the attributes and highlighted by Mike show that the first information on a page is the most important.

So get to the damn point, y’all!

And use your intros wisely.

Don’t spend 500 words talking about the history of Italy if the post is about where to find the best cappuccino in Bari (honestly, I didn’t have a bad one. So go anywhere. They’re all better than Tim Hortons!).

9. Niche Down, Authority Up

The “siteFocusScore” figures out how on topic a website is. We don’t know how that gets weighed, but really why would it exist if Google didn’t care about a site being pretty on topic?

A 10/10 score (I’m making up the scoring system) might seem a bit spammy, but a 0/10 score would seem equally weird since nothing on it is at all related.

It seems like having closely connected content in a specific niche is a positive. Even if it’s not a positive for Google, having close knit topic clusters that provide users with all of the information on a subject will be beneficial to your user.

But when you niche down, don’t forget to build your authority.

Anyone can write 1,000 blog posts on golden retrievers, especially with all the one click AI writers out there. You need to build your authority and your site’s authority on the subject.

That’s where I think the author signals, especially sameAs schema from off-site profiles/author pages will come in handy.

I also expect it’s where the high quality backlinks have an impact, although Google probably hopes they come organically instead of with some digital PR outreach.

10. We’ll Never Know Everything

Ultimately, we’re not Google.

I don’t work there, and neither do any of the SEOs I’m familiar with.

We don’t have the whole algorithm. We don’t know how up to date this is, if some of these factors are being tracked but have no bearing on rankings, or even how the factors are weighted against each other.

Bloggers can only do our best at the moment to create helpful content for our users and build something we’re proud of and passionate about.

We’ll need to keep running tests, doing case studies, and publishing new content to see what works with Google’s systems – and then do it all again when an update changes everything.

I think Rand said it well: “Google no longer rewards scrappy, clever, SEO-savvy operators who know all the right tricks. They reward established brands, search-measurable forms of popularity, and established domains that searchers already know and click.”

So to succeed with SEO in 2024, you need to become an established brand, or find another traffic source. (In my opinion, you should do both!)

How I’m Adapting My Blogging Strategy

March 29th Update: Google Verifies The Leak

Google responded to the alleged leak of their algorithm by having a representant – as yet unnamed – speak to Barry Schwartz of Search Engine Journal.

The spokesperson said:

“We would caution against making inaccurate assumptions about Search based on out-of-context, outdated, or incomplete information. We’ve shared extensive information about how Search works and the types of factors that our systems weigh, while also working to protect the integrity of our results from manipulation.”

It’s a pretty vague verification of the leak, but if it wasn’t actually Google algorithm attributes I expect they would’ve denied it outright.

Many on SEO Twitter, Threads, and Facebook are rightfully wary of listening to this response. I mean, the leak has disproven explicit statements from Google spokespeople. There are cries of “why should we listen to you, when you’ve just been outed for lying to us for years?”

I’ll definitely be taking the spokesperson’s caution with a grain of salt, but I’ll equally be taking the alleged “ranking factors” with a grain of salt.

We still don’t know their weighting in the search results.

As a data girlie, I’ll be letting my case study data and that of other SEOs influence my strategies more than statements from Google reps. It’s proven to be more valuable and reliable than their advice has been.

#1 Blog Course

Profitable Part-Time Blogger

Ready to grow a blogging business that fits your life?

The Profitable Part-Time Blogger Program helps you attract your ideal audience, master SEO, and make $5K/month—all in just 20 hours a week.

Whether it’s a side hustle or your dream job, it’s time to work smarter, not harder. Let’s make your blog work for you!

Grow Your Blog Today!

Sources

I have so many more thoughts on this, and after spending 14 hours reviewing material and hyperfixating, I could very easily keep going for eons.

But it’s gotta end sometime. So I’ll pass you over to the sources where I got great information for this article:

This is The Biggest Google Search Algorithm Leak in History. Here’s What it Means for Bloggers