Up and Down: An SEO’s Investigation into Google Search’s Alleged Spam Problem

By Ethan Lazuk

Last updated:

SPAM cartoon image with Google color scheme.

I’ve been covering SEO social media posts, articles, and trends in Hamsterdam for upwards of 40 weeks now, and following them for years.

If there’s one thing I’ve learned, it’s that SEO culture — as my anthropology student-self back at the University of Montana might describe it — has epochs.

If the latter half of 2023 was defined by talk of Google’s search rankings volatility, so far 2024 feels like it’s revolved around spam.

There may be a correlation — as we SEOs are apt to look for — as part of the volatility was driven by ranking system updates, including the third helpful content update in September 2023 and the completed roll out of the hidden gems improvements to the core ranking systems in November of that year.

[Update: Since this article was written, Google rolled out some large updates, including a March 2024 core update (March 5th to April 19th, lasting 45 days), a spam update (March 5th to March 20th), plus new spam policies, including manual actions and a planned “site reputation abuse” policy with planned algorithmic updates that begins on May 5th of 2024.]

In this article, I’ve included several examples of SEOs and others calling out spam in Google’s search results, and not always in the same context.

I’ve also got a hypothesis for why this is happening in part, which I’m going to test in real time with you later in the article!

But first, let’s get our definition of “spam” squared away …

What is spam?

I’m vegan, so I’ve never had Spam the canned meat product. Supposedly, it’s good when grilled Hawaiian style.

Google SGE desktop results for Spam recipes.

I do have an email account, though, so I’ve experienced spam in my inbox, or unsolicited messages, usually with a commercial purpose.

The first kind of spam messages that come to mind are “guest post” pitches, which usually are a euphemism for low-quality, black-hat backlinks. Ethics aside, these aren’t likely to get referral traffic or pass meaningful PageRank anyway, in my view.

Here’s an example of why I suspect that:

LinkedIn DMs are also a hot bed for link seller spam pitches.

Ironically, those usually come from a person with “SEO expert” in their title.

Wayne's World Sh'yeah right GIF.

Google even refers to “links that are intended to manipulate rankings” as link spam and has dedicated link spam updates to help neutralize their impact.

That said, link spam is just one type that falls within Google Search’s broader spam policies.

I’ll quickly go over the different types of spam Google Search Central lists in their documentation, because I think it’s helpful to know them.

Then I’ll clarify the specific types of spam I’ll be talking about in the rest of this article.

Here are examples of spam Google lists in their documentation:

  • Cloaking – where users and search engines see different content, usually to get the former to take an action while tricking the latter into thinking the content is honest quality.
  • Doorway sites and pages – this is when various entry points get created to lead users to the same destination, rather than each site or page having uniquely valuable content, essentially.
  • Hacked content – this is when a site’s security vulnerabilities get exploited to add content or redirects without permission. (I’ve been hacked before, and it’s no fun.)
  • Hidden text and links – this is when sites use content that’s not visible to users while trying to influence search rankings, but it’s old school.
  • Keyword stuffing – this is writing content unnaturally for a reader in order to use more keywords and to try to make content relevant for search engines, but in today’s era of semantic search and machine learning systems, it’s less effective.
  • Link spam – these are inbound links intended to manipulate search rankings, there are many types, and most ways link-building gets done violates these policies, in my opinion.
  • Machine-generated traffic – this is when people use bots to search queries or scrape results without permission.
  • Malware and malicious behaviors – these are sites hosting software intended to harm your computer, such as by installing a virus.
  • Misleading functionality – these are sites promising to deliver a service but don’t and use that as a ruse to get ad impressions or something else.
  • Scraped content – this is when a site takes another’s content, usually from a more reputable source, and republishes it as its own in some form, either verbatim, auto-translated, or superficially revised, like with generative AI.
  • Sneaky redirects – these are redirects that send users to unexpected places maliciously.
  • Spammy auto-generated content – this isn’t necessarily AI-generated content, but it’s programmatically created content that offers little-to-no value to users and is meant to manipulate search rankings.
  • Thin affiliate pages – affiliate links go to product pages and earn commissions, but when those pages offer little unique value, they’re considered thin and spam. Google also has a reviews system that now runs continuously.
  • UGC spam – this is spam added to a site by users, like blog comments or forum threads. (It’s actually sad how much blog comment spam my old site got, but sometimes reading the messages is quite funny.)

Ok, that’s quite a few types of webspam. Speaking of which, how does Google deal with all of it?

Well, Google has automated systems to detect spam mostly, then there’s a spam form users can submit manually, as well as human reviewers on a webspam team who levy manual actions against violating sites.

Google has a list of manual actions, many of which align with the spam types mentioned earlier.

I’ve never worked with a site that had a manual action, but in a different article, I did include an example of a site that used generative AI to rewrite scraped content from a competitor and then got a manual action for copied content.

That’s sort of the type of spam I’d consider “SEO spam.”

My definition of “SEO spam”

I’ve been doing SEO since 2015. My first job was on an agency’s content team, where we created blog posts and webpages at scale.

A lot of that content was low quality, churn-and-burn stuff designed to earn rankings, not satisfy users.

That, to me, is “SEO spam.” However, I don’t prefer that terminology.

It’s a common term in the tech world among non-SEOs, who use it to reference content intended to manipulate search rankings.

I’ve heard “SEO spam” used several times on podcasts like Decoder (which is associated with The Verge, so take that as you will).

But the most prominent reference that comes to mind was in a letter from the CEO of Perplexity AI, an LLM-driven answer engine looking to compete with Google Search:

The context there, to me, is partly strategic marketing speak.

Perplexity gives generative AI answers to users’ search queries. I compared how those look to a featured snippet on Google Search (with human content) in Hamsterdam Part 41.

By suggesting SEO-driven content is spam, the CEO is, in my view, trying to suggest that generative AI text is probably better. Everyone has skin in the game.

However, “SEO spam” to me actually refers to “search engine-first content.”

I wrote more about this in my helpful, people-first content and 11x framework articles, and I agree, it’s lame stuff.

Those churn-and-burn content days are far behind me now, though. I matured and grew, learned to build brands, not chase rankings, and most of all, got experience (partly from auditing sites impacted by ranking system updates) to show manipulative tactics are short sighted.

To succeed on Google Search today, I’d follow a user-first approach, where SEO content is created to satisfy the search intent of an intended audience first. The opposite of that, or content made to manipulate rankings only, is search engine-first content.

When SEOs complain about spam in search results, one type is search engine-first content that could be considered “unhelpful.”

However, there’s another type of spam, which is far more egregious …

Gross spam

Some things on the internet are just gross. No, I don’t mean like Dirty Jobs gross. I mean ethically gross.

Shortly after my old site got hacked, for example, I saw another site had taken one of my articles and added “chance to win” type language:

That’s not a misguided person stuffing keywords in a blog post. That’s something way more sinister, with deceptive intent.

That grossness is the bigger spam problem.

Lily Ray was an early voice on the spam issue, posting about it in late December 2023. Some of her posts also have comments with screenshots from others who found spam, as well.

Then Roger Montti, with insights from Bill Hartzer, wrote an article for SEJ about the spam:

“The spam sites appear to be taking advantage of at least three windows of opportunities that a part of how Google ranks websites. These opportunities are not new and spammers have been taking advantage of them for many years but not to the extent that has been going on lately.

Perhaps the most important reason for the success of the spam is that the search queries the spam sites are ranking for are low competition, which makes it easier to rank.”

Google Search Overwhelmed By Massive Spam Attack, Roger Montti, SEJ

The three windows of opportunity for spam that Roger’s article mentions include:

  1. Google’s local search results (those algorithms being more permissive).
  2. Targeting long-tail keywords (those SERPs being less competitive).
  3. The freshness of the spam site domains (being new, Google gives them the benefit of the doubt).

Remember the discussion of link spam earlier?

It also appears many of the spam websites were interlinked, although, similar to what I posited about spammy guest post links earlier, the authors suspected the link networks to be helping with discoverability (crawling and indexing) more than rankings.

Many of the spam sites were also using sneaky redirects and cloaking, techniques called out in Google’s spam documentation earlier.

Google’s main search results aren’t the only place spam has been sighted.

Google News was recently implicated for surfacing low-quality content, as well.

It’s also worth mentioning that per Danny Sullivan, Google Search Liaison, Google Search is aware of the spam issue(s) and has changes coming in response:

So far we’ve spoken anecdotally about spam on Google, but there was also research on this topic published recently …

Academic study of web search spam

A recent study published by German academics looked at whether Google’s search results (along with Bing and DuckDuckGo) had gotten better or worse over the last year, from October 2022 to September 2023.

The study got a lot of attention from news outlets as well:

Google News desktop results for Google search results study query.

Its conclusion, as summarized by Barry Schwartz in Search Engine Land, was that:

“The study aims to figure out if search results are getting worse. And the study concludes that specifically with Google the search results “seem to have improved” to some extent from the start of the study to its completion. But the study said it requires algorithmic updates, constant ones. They wrote, “that search engines do intervene and that ranking updates, especially from Google, have a temporary positive effect, though search engines seem to lose the cat-and-mouse game that is SEO spam.” …

“Google’s updates in particular are having a noticeable, yet mostly short-lived, effect. In fact, the Google results seem to have improved to some extent since the start of our experiment in terms of the amount of affiliate spam. Yet, we can still find several spam domains and also see an overall downwards trend in text quality in all three search engines, so there is still quite a lot of room for improvement,” they added.”

New research study asks if Google Search is getting worse, Barry Schwartz, SEL

There we have that term “SEO spam” again!

But, basically, the study is talking about an old game in this town.

Spammers manipulate search rankings using loopholes, Google adjusts its ranking systems to close them (or at least tamp down), so then the spammers switch up. It’s a whack-a-mole game that goes around and around.

Meanwhile, the rest of us SEOs are building helpful content to grow brands and support our clients’ long-term business goals. 😉

But my fundamental question is this: is the spam in search results found uniformly throughout the SERPs, or is it concentrated in lower results?

Here’s the hypothesis and logic behind that question …

Hypothesis about the spam issue’s cause

I’ve been researching and writing about Google Search’s ranking systems in my blog posts more lately. While looking into RankBrain recently, I caught a line in a blog post from February 2022 on The Keyword called How AI powers great search results, written by Pandu Nayak:

“Thanks to this type of understanding, RankBrain (as its name suggests) is used to help rank — or decide the best order for — top search results. [Highlights added.]

That mention of “top search results” stuck out to me.

It made me think about how not all results in a SERP are subject to the same attention from ranking systems.

More broadly, I recalled an SEL article from December 2023 by Danny Goodwin called How Google Search and ranking works, according to Google’s Pandu Nayak, where he includes this quote from Pandu:

“”So that’s — the next phase is to say, okay, now I’ve got tens of thousands. Now I’m going to use a bunch of signals to rank them so that I get a smaller set of several hundred. And then I can send it on for the next phase of ranking which, among other things, uses the machine learning.””

I also caught Episode 320 of Marie Haynes newsletter from January 17th, 2023, where in a section called, “My updated thoughts on how ranking works,” she mentions how:

“The top 20-30 results for any search are re-ranked by RankBrain.”

My hunch is that spam is getting into Google’s search results using the loopholes already mentioned earlier, but I also wonder if by adjusting some of the ranking system dials associated with its AI-based systems to surface more “hidden gem” content, Google also created a lackadaisical way of letting in spam.

One reason I suspect this is because I recently saw a screenshot of a SERP replete with spam, however, when I checked the query myself, the top 10 or more results were all valid, and the spam was below them.

I also suspect Google may have turned back its “hidden gem” dials a bit recently, based on rankings declines I’ve observed to smaller sites that had more visibility in late 2023.

That said, let’s test our theory by manually reviewing some SERPs for spam. I’ll record the findings in real time below, so you can learn more as I do! 🙂

Spoiler alert: I’m not sure if Google has a spam problem so much as a “showing too many results” problem.

Investigating Google’s spam problem, one SERP at a time

Following my hypothesis — that spam will be predominately lower in Google’s search results because they’re ranked differently than top results — I’ll now look at a handful of SERPs manually to identify a) if spam is present, b) what kind, and c) where it appears in the ranked results.

To view these SERPs, I’ll use an incognito tab in Chrome on desktop, so we won’t have SGE snapshots or personalized results.

Which queries should we look at?

The list of spam tactics from Google above mentioned “thin affiliate pages.” The study referenced earlier also focused on “low-quality, mass-produced commercial content” from “affiliate marketing” in the “product review genre.” So I think it’s fair to start with product review queries.

The study specifically calls out the Amazon Associates affiliate network. Since Amazon can be a great place to find product review keyword ideas, I’ll use its “Best Sellers” page from today (1/27) to select a few random products and search for “review” keywords related to them.

Amazon best sellers page.

Additionally, the SEJ article by Roger Montti mentioned before had a list of keywords for one of the spam domains. These are all over the map:

List of keywords for a spam domain from a Search Engine Journal article.

But we can examine a few of them, as well.

I’m not a huge fan of basing SEO decisions on keyword metrics from third-party tools, but I’d imagine if spammers tend to target long-tail, low-competition queries, they might be using keyword research data.

To that end, I’ll use Ahrefs Free Keyword Generator to report the volume and difficulty level for each keyword.

Ok, let’s get this party started!

Keyword 1: “stanley tumbler review”

The Stanley Quencher is the #1 best selling kitchen and dining product on Amazon today.

The query [stanley tumbler review] has a difficulty of 6 and a monthly volume of 150.

In the top results, I found a few webpages with affiliate links to Amazon, but most results were large publishers, ecommerce retailers, or influencers.

Google desktop search results for "Stanley tumbler review."

Other SERP features included videos, discussions and forums, and images.

I scrolled the whole SERP and found nothing that would count as spam to me.

Even the affiliate sites seemed like legitimate reviewers.

Maybe that product is too well known?

Let’s try a more obscure one.

Keyword 2: “casekoo iphone case review”

The CASEKOO iPhone 13 Pro Case is the #20 most popular item in cell phones and accessories on Amazon.

The query [casekoo iphone case review] has no registered keyword difficulty and a volume of 10.

This is definitely more obscure.

Let’s check the SERP.

The first results are videos, but the top web results below definitely included more of the lower-quality, search engine-first type affiliate content.

Google desktop SERP for CASEKOO iPhone case review query.

Check out the 8th web result, for example. While these aren’t affiliate links, you can just tell from the exact match keyword anchor text and number of links that this is SEO-forward content:

Paragraph from a phone case review page with exact-match keyword anchor text.

The SERP also had quite a few branded results for CASEKOO, and a lot more video content.

Also, the lower in the results we go, the more sites had ads.

By the bottom of page 2, we start to hit borderline gross spam.

This website is called phoneskins.org. (A .org domain name was originally for nonprofits, but its uses have expanded. This seems a bit too commercial, though.)

The site’s menu only has three items: home, phone case recommendations, and a blog.

Phone Skins phone case review page and menu.

The page content itself is formulaic and likely AI-generated or -assisted, including with artificial customer reviews.

FAQs and customer reviews for a phone case review affiliate site.

The page was also recently published on December 25, 2023 (Christmas in the West), and it has no favicon.

Phoneskins.org website result on Google desktop.

By around page 4 of the SERP, we start to get review sites like Judge.me, which is clearly showing fake and likely AI-generated reviews.

Judge.me phone case reviews.

I mean, how many people in the U.S. do you know named Marvin who write like this: “Because of this I’ve already tested it’s robustness and it didn’t let me down but now I have many scratches on the corners where it always seems to fall on.”

Even lower in the results, we got stuff like this beautiful blank page from review.insightzine.com, which was published in September 2023.

Insightzine webpage that's just blank.

Yet, by the time I reached the end of the SERP, I didn’t find too much that was egregious.

If I was genuinely looking at the top results, not including the videos, my best options would be either Amazon, Reddit, or the CASEKOO site.

Google desktop search results for CASEKOO iPhone case review.

For that type of query, those results are reasonable, in my view.

The rest was largely affiliate content that borders on search engine-first content, and sometimes even gross spam.

Let’s try an even more generic query.

Keyword 3: “3 pack usb cable review”

The 3-Pack 60W USB C to USB C Cable is the #19 best selling musical instrument product on Amazon.

The query [3 pack usb cable review] had no information found in Ahrefs.

Sounds like a good one to test!

Here, the first two results were Amazon, followed by a couple of New York Times results, Reddit, some larger brands, like PCWorld and Best Buy, and also GearLab, an affiliate site. The SERP also had a product filter side panel.

Google desktop search results for 3 pack USB cable review.

The GearLab affiliate site had a lot of content. My feeling is that it probably wasn’t written with a reader in mind, judging from the wordiness, but it does provide lots of information and clearly had effort behind it.

GearLab product review excerpt for USB cables.

Lower down in the SERP, I saw Wired, Walmart (product page), and some ecommerce tech retailers.

I also started to see more search engine-first-type sites, like ZeroLemon.

ZeroLemon product page.

It got weirder still.

I’m not sure what type of E-E-A-T The Jerusalem Post has for “Usb” cables, for example.

The Jerusalem Post USB cable review page.

But, if we’re going to call them out, we also have to ask what does CNN know about USB cables? And surely a Lightning cable isn’t a USB cable, so why are they there?

CNN article for the best Lightning cables.

I’d say the first is search engine-first content, while the second is just “Goog enough,” to quote AJ Kohn.

I also saw New York Magazine, for instance, while the rest of the SERP were smaller tech retailers (mostly legitimate) and niche publications (mostly affiliate spam).

Conclusion about product review SERPs

Three keywords does not a comprehensive study make. 😉

That said, there is clearly a lot of search engine-first content (or “SEO spam”) for these queries. Some of it even borders on gross spam, especially for the more obscure, lower volume product review queries.

We also see evidence of big brands seemingly getting preferential visibility, which is more an issue with marketing incentives (and, arguably, ethics) than spam.

However, for the average user, I don’t really see them looking too far beyond Amazon, Reddit, or some of the big retailer sites to find the product content.

As Thom Yorke reminds us in the lyrics to There, There, one of my favorite Radiohead songs, “Just ’cause you feel it doesn’t mean it’s there.”

Now let’s look at one of the weird keywords from that other list.

Keyword 4: “tacoma lug nut torque specs”

My first job in digital marketing was writing “optimized” website copy for automotive dealerships, so right away, I knew this keyword was about a Toyota pickup truck.

The query [tacoma lug nut torque specs] has 0 keyword difficulty and a volume of 90.

That’s interesting.

If I were a spammer, I suppose those are the types of metrics for a low-competition yet modestly searched query I might target.

Now let’s check the SERP.

Right off the bat, these look like good results.

Google desktop search results for Tacoma lug nut torque specs.

We have a forum (real experience) from a site dedicated to Tacomas. Then we have two other sites that seem to be from gear heads, as well as a dealership result.

Oh wait …

That Tacoma Explorer site seems pretty legit. I mean, just look at all that blog content.

Tacoma Express webpage with blog menu extended.

But the content for the Sparky Express site ranking third seems a bit off.

Sparky Express Tacoma lug nut page.

That page is a few years old (predating publicly available generative AI tools), and the content is quite short and to the point.

I could see this being perceived as a “hidden gem” by an algorithm.

However, a discerning human would probably tell this wasn’t created for them. It’s got “target keywords” and “rank on Google” written all over it, in my opinion.

Lower down the SERP, there are quite a few forums and helpful videos and images.

Lower SERP results for Tacoma lug nut torque specs.

But by early in page 2, I started to find some borderline gross spam, like this 21 Cylinders website.

21 Cylinders website for Toyota Tacoma.

That too is older, though, published in December 2019.

Then lower down still, around page 4, I found an obvious spam intruder, a domain called car-know.com with a WordPress favicon and a publish date in the URL slug of 1/15/2024 (just two weeks ago).

Car-Know.com result on Google desktop.

Dare I click it?

Ok, it’s for the sake of science!

It actually wasn’t as scary as I thought. But obviously, hello ChatGPT content!

Car-Know.com website content.

And just a little lower down, around page 5, we get to the good stuff.

I mean, the gross stuff.

Spam ranking in Google for Tacoma lug nut torque specs query.

I won’t be clicking those. And what does a 2019 Ram 3500 have to do with a Tacoma, anyway?

It got way worse even lower down, around page 6.

Gross spam ranking in Google around page 6 for Tacoma lug nut torque specs query.

Yet in between, I saw local dealerships and other legitimate car sites.

The rest of the SERP, or rather about 90% of it, aside from some YouTube videos, was all spam, most of which had nothing to do with a the truck in question.

Gross spam around page 7 of Google search results.

So, yes, for the query, Google has a spam problem, of the gross kind.

But let’s also hold on for a second and think.

What’s the search intent of that query? It’s to find a specific answer.

If I wanted to know the specs, would I honestly scroll down to the 2nd, 4th, or 7th page of Google’s search results?

Probably not.

I’d likely click that first forum result, or maybe the blog below it. And even that SEO-forward result in the third position gave an answer.

Is it fair to say Google has a spam problem, or maybe it just has a “too many search results” problem?

After all, look what SGE has to say about this.

Google SGE snapshot for Tacoma lug nut torque specs.

As a user, we wouldn’t even need to expand that answer. (Assuming it’s correct.)

[Update: Something else I’ve noticed since writing this is article is the expansion of Google Explore for mobile results, specifically on the Google app on iOS. Explore introduces related topics with three results (including webpages, videos, social media, etc.) for users to explore. In my experience, Explore tends to appear after the 20th to 25th search result (or between pages 2 and 3). For this query we’ve been looking at, it appeared after the first 10 results!

Google Explore showing for Toyota Tacoma lug nut query.

As we can see, this helps solve the issue of too many search results showing, eliminating the likelihood of spam surfacing. I also noticed far more image, video, and forum results showing. Prioritizing forums here also helps eliminate spam domains, which tend to be blogs:

Google mobile results showing video and forum results.

Combined with the March updates mentioned in the introduction to this article, Google has taken steps to reduce spam since this article was first published.]

Should we bother looking at more keywords?

I’m going to assume most of the keywords from that spam domain have a decent search volume and a low difficulty.

Here are three I randomly pulled:

  • “medium short hair cuts” has a 28 difficulty and volume of 80.
  • “eso survey maps” has a difficulty of 8 and volume of 20.
  • “sundolphin boat” has a difficulty of 1 and a volume of 60.

Let’s quickly search each of these and scan the results.

“medium short hair cuts”

This SERP was very visual. It had images, Pinterest results, discussions and forums, and lots of brands with image previews (like a product category page).

I started to see SEO spam-style content around page 2, but even then, the content was decent and the pages were mostly just overly monetized with ads.

I did find some gross spam, but it was only a couple of results and near the very end of the SERP.

Gross spam lower in Google desktop results for medium short hair cuts query.

Suffice to say, a user probably wouldn’t find a spam problem with those results. If anything, Google is just showing too many of them.

“eso survey maps”

I’m not sure what this means, so let’s find out!

Ok, it looks like it’s related to The Elder Scrolls video game.

Honestly, some of these domains look kind of sketchy.

ESO survey maps Google desktop SERP.

But they’re all seemingly legit.

The Elder Scrolls Online Wiki page.

There were also quite a few YouTube results around pages 2-3.

Then around page 4, we start seeing gross spam, mostly with .shop or .online domains.

ESO survey map results showing spam around the fourth page.

Again, about 90% of the remaining 20-30 results at the bottom of the SERP were gross spam.

Spam results on lower Google SERP for ESO survey map query.

But, like before, the top results were all quality, at least for the first few pages.

Does this mean Google has a spam problem, or a “too many results” problem?

Wonder if our last query will be the same?

“sundolphin boat”

This looks like it’s actually a navigational query for “Sun Dolphin Boats.”

It’s also the type of misspelling RankBrain would likely help adjust for.

The search results included the brand’s homepage with sitelinks, a product listing grid, images, retailers like Amazon and Walmart, videos, what looked like an old domain for the brand on HTTP that wasn’t redirected, and some more retailers.

I didn’t find a spam result until like page 5, and it stood alone.

Spam result on Google desktop SERP for Sun Dolphin Boat query.

Then, like with the other queries’ SERPs, the last couple dozen or so results were all gross spam.

Gross spam at bottom of Google SERP for Sun Dolphin Boats query.

It’s kind of the same deal each time.

The top few pages of results are usually all quality and relevant, with maybe one or two gross ones snuck in around pages 3-4, while the bottom of the SERPs have the most gross spam.

But is a user really going to scroll that far?

I’d suggest that if Google is looking to surface more “hidden gem” content, maybe it should also surface fewer results by default.

So, does Google really have a spam problem?

I think, yes, Google has a spam problem. But we should qualify how we describe it.

For product reviews, the spam problem is more search engine-first content with affiliate links, or “SEO spam.” We could mention the “big brands seem to rank for anything” idea, as well.

For lower difficulty queries with moderate search volume, Google tends to surface more gross spam, but it seems concentrated in lower results, around positions 50-70+.

We can find instances of search engine-first content and gross spam higher in SERPs, yet to really uncover it, a user would need to scroll deep down.

In reality, most user’s would have their search intents satisfied by the top results.

My conclusion is that Google has less of a spam problem than it does a “too many results” problem.

Some mobile SERPs will show a Google Explore feature, for example. That would typically intervene before most of the spam surfaces.

Explore usually doesn’t show for me on my iPhone Google app, but it will sometimes on my mobile Chrome browser. (Update: as mentioned above, I’m now seeing Explore on the Google app.)

I searched that [eso survey maps] query using Chrome on my phone, as a test. Explore didn’t appear, yet by around page 3-4, when the spam started to creep in, the default SERP stopped.

Lower Google mobile SERP results for ESO survey query.

I think that’s partly the answer.

On desktop, Google is showing more results than are necessary for most queries, in my opinion.

If Google limited its search results to maybe the top 30 or 40, a lot of the gross spam wouldn’t be found (unless someone clicks “more results”).

Such a change might hurt the visibility of legitimately quality and relevant sites, but if the user could instead be given Explore or related searches to refine their search journey, they may find those other sites anyway in a more qualified context.

This is just a thought experiment, though.

I do think our hypothesis was confirmed about spam being predominantly lower in the results.

As for the cause, it may be that Google’s search ranking systems are allowing in more search engine-first content masquerading as helpful content to the top results. Meanwhile, more gross spam might be getting in lower down because Google’s AI-based systems are focusing on top results.

The “spam issue” also seems unlikely to be felt by most average users. If they’re just looking to satisfy their search intent, they probably wouldn’t scroll down far enough to find most spam. They may even ignore most search engine-first content that sneaks into top results, focusing instead on Amazon, Reddit, and other trusted brands.

“It’s only round and round, and round”

I’ll keep an eye on how things develop with news of Google’s spam problem and any updates pushed live to counter it.

As mentioned, I suspect the “hidden gem” dials might already be turning back a bit, based on rankings changes I’ve observed for smaller sites.

I’ll also update this article as we learn more!

Until then, enjoy the vibes:

Thanks for reading. Happy optimizing! 🙂

Editorial history:

Created by Ethan Lazuk on:

Last updated:

Need a hand with SEO audits or content strategy?

I’m an independent strategist and consultant. Learn about my SEO services or contact me for more information!

Leave a Reply

Discover more from Ethan Lazuk

Subscribe now to keep reading and get access to the full archive.

Continue reading