I think SEO still matters, just not in the same way it used to.

What we can prove unequivocally is that the more a page deserves to be ranked well, the more it is penalized.

This something that I have believed for a long time, and the result shifts prove that.

Every page that did rank well, that has since been updated for the latest SEO trends like including schema, etc, no longer appear in the first 100 results!

There is no justification for this and the only solution will be a class action.

I think we’re all saying the same thing in different ways. Google isn’t just ranking pages anymore - it’s routing demand. So “deserving to rank” often doesn’t matter the way it used to.

SEO still matters, but it’s now just one part of the mix. What’s working, I believe, is brand familiarity and signals, direct channels, and concise content that solves the question better than a summary. If you depend fully on Google, you’re exposed. The real game now is understanding the system that decides what gets seen. That's evolving rapidly imo.

@Whitey - Exactly. The game now is finding out what gets seen and WHY. The why is the part that keeps updating, but as always, it must have some sort of pattern, right? Game on.

"finding out what gets seen and WHY" sounds like optimism but at the end of it all, it will come back the same - there is no rhyme or reason for returning search results that are barely remotely related and results that are garbage.

AI has very little to do with the garbage that follows the top of the first page of results - which is usually like:
1. Any drivel that wikipedia has that is even remotely related.
2. Any drivel that reddit has that is even remotely related.
3. Any drivel that slack exchange has that is even remotely related.
4. Spamvertising posted by a content thief.
5. Blog posting drivel seeded to attract traffic
6. Another #4
7. Another not well informed blog
8. Something totally unrelated
9. More of the above
10. More totally unrelated info.

Maybe they are employing seven year olds - because the algorithm for these results could not be less sophisticated.

"finding out what gets seen and WHY"

What gets seen the most? Paid for ADS, Why? Because they increase Googles profits.

there is no rhyme or reason for returning search results that are barely remotely related and results that are garbage.

There is a reason, increasing AD clicks and engagement. In my experience ADS are almost always relevant. The less useful, relevant (and visible) the rest of the SERPS become the more useful and appealing ads become, resulting in more ad clicks, resulting in more profit for Google.

So If the question is like Does SEO matter any more after AI Overview?

Yes, SEO absolutely still matters after the introduction of AI Overviews.

Reason...

The AI Overviews are generated by analyzing the top-ranking results on the Search Engine Results Page (SERP). If your content doesn't rank highly (a result of good SEO), the AI will never find it or use it as a source.

The AI provides links to the websites it used for its answer. These citations are a new form of high-value referral traffic.

AI Overview doesn't satisfy all intent so users often click through to read the full article, see detailed instructions, or browse a website's products/services. Informational SEO (answering questions) often leads to Commercial SEO (driving conversions).

Think of AI Overviews as a new, high-stakes SERP feature. To be included in it, and to continue generating traffic from it, your underlying SEO foundation must be strong.

Cheers!

The toxic backlinks part is the most frustrating. If you have a long history, you’ll inevitably have old links, directories, PAD listings, and dead sites. In my case, another audit flagged almost everything as toxic without explaining why.

When I worked with a SEO agency, they treated it as a triage problem, not a panic one: they excluded dead domains, checked affiliate links separately, and kept only what actually made sense to clean up. It was far more realistic than disavow everything.

[edited by: not2easy at 12:54 pm (utc) on Dec 15, 2025]
[edit reason] Please see TOS [webmasterworld.com] [/edit]

Does disavow actually do anything these days? (off topic)

Audit tools throw twenty warnings at you and most of them do not matter for real users. You want to keep an eye on load time, structure and content, not every tiny icon that shows up in a report. When an agency steps in, pay attention to how they handle the first talk with you.

They should ask clear questions about your goals, your pages and the problems you want solved. That shows they plan around your site instead of following a checklist.

Ask them to show you how they handle revisions and how they measure progress. A good team explains what changes improve speed, what fixes help usability and what content work actually moves the needle. You avoid paying for fluff.

I reached that point after wasting time on pointless warnings.

[edited by: not2easy at 4:47 pm (utc) on Dec 9, 2025]
[edit reason] Please see TOS [webmasterworld.com] [/edit]

Does disavow actually do anything these days?

The problem here is that the lists provided by SEO software are faked. Most use the same resource that is based on listings compiled over more than a decade, lists that have never been updated.

For example, I found that more than 90% of the links on my disavow list had been dead for more than a decade. It took a couple of days, but that is what I discovered after checking each one. In fact most of the sites no longer existed.

So if you are going to submit a disavow list, it might look better if the list was real.

@Kendo The Web at an IP level is even worse now. Much of the African IP allocation (IPs were initially allocated on a regional basis to Regional IP Registries. The development of the need for IPs differed from expectations.) was "acquired" and ended up being used by companies in Hong Kong and China. Large ranges of IP addresses that once belonged to big companies are often acquired by IP brokers for resale and lease. That means that a range that might have been a US range in the lists published by the registries could have subnets in other country. Then there is the ease with which the country of an IP address can be changed by a web hoster and published in its geofeed (or not). This is based on having identified the web hosting providers of 99.4% of all gTLD websites with 100% coverage of China, Hong Kong, Ireland, Australia and many other countries, and 99.47% of US web hosters. I am not sure that Google has the level of intellect necessary to deal with the problems of the geography of the Web and that feeds into the backlinks problem.

Web geography and the value of domain name ownership data was completely banjaxed as a signal by the well-intentioned incompetence of the European Commission with its GDPR. ICANN's dithering navel-gazing on redacting key location data from WHOIS records compounded the problem. That means that the country data and ownership of domain names is no longer a reliable source of data on website locations and audiences. Some gTLD registrars took to opportunity to completely redact *all* ownership data from WHOIS records after May 2018. So not only is IP locaton data of questionable value when it comes to backlinks, domain name and website ownership data might also be unreliable.

The backlinks problem is made more complex by the fact that the Web is continually changing. Again, I don't consider Google to have the necessary abilites and intellect to deal with this becuse with its concentration on Search, it missed the simple fact that the Web is more like a river than a monolith. The number of .COM domain names deleted in November 2025 was 2,835,739. The non-renewal rate of new .COM registrations from 2024 is approximately 49%. Many of those domain names drop without ever having had a developed website. And those that had been developed might have created some short-lived backlinks. The problem with some long-lived backlinks is that the ownership of the website might have changed and it could be part of a private blog network.

The mobility of websites, even ten years ago, used to be quite slow moving. They might have changed annually as a new web developer took over the website and move the site to a different web hoster. That would have resulted in a change of IP address for the website. With the rise of Cloud hosting (Amazon/Azure/Google/Alibaba/Tencent), CDNs like Cloudflare, tracking websites at an IP level is more difficult. That stuff from W3Techs on market shares of Cloudflare and Cloud hosting is not accurate. This is because measuring such IP/location usage of 195.5 million gTLD websites (December 2025) is much more complex than guesstimates based on small samples. And some web hosting businesses and their websites only exist in the Cloud and have no unique nameservers/DNS other than the Cloud hosting provider's main DNS. Large registrars also use Cloud providers for much of their services. If an organisation like Google with its wonderfully clever people can't deal with backlinks, it might be a bit optimistic to expect SEO heads to be able to make sense of it.

Regards...jmcc

[edited by: jmccormac at 4:30 pm (utc) on Dec 26, 2025]

@jmcc

I love reading this statistical stuff and the messes those supposedly "knowledgeable" people make! Overall though, how do these "errors" affect us all and in what way?

Seemingly, these days, localisation is more important especially to Google than quality of the content information. This is especially evident in the UK in my widget industry where SEO hardly exists these days.

@RedBar The big split in the Web happened in the mid 2000s when the incompetence of ICANN management let the Domain Tasting business get out of hand. Basically, gTLD registrars were using a loophole in regulations to register millions of domain names each month, test them for Adsense/PPC revenue and then delete them within the five day Add Grace Period if they didn't make enough from clicks. That meant that deleting domain names that would have been reregistered and developed were kept out of the reach of ordinary registrants. It led to the "all the good names" are taken claims and kickstarted the evolution of the ccTLDs. By 2009, serious pressure had been applied and a "restocking" fee was added to each transaction by ICANN so that any new registrations over a % of a registrar's typical new registrations had to pay that restocking fee. It helped kill large-scale Domain Tasting. But it was legal action against the main registrars (they were targeting trademarked brand names) and the decision of Google (when it wasn't "evil") to demonitise new registrations within that 5 day period. ICANN had already bought into the whole New gTLD thing and proceeded with that. The end of Domain Tasting resulted in a flood of gTLD domain names becoming available. But it was the shift to ccTLDs from gTLDs that largely went unnoticed by ICANN and even a lot of SEO heads. It has also had more profound implications in that it has effectively killed .COM as the First Choice TLD in many countries.

Any business targeting the UK market would probably choose a .UK domain name first. With the Irish market, it is .IE. With Germany it is .DE and with France it is .FR. Where there is a strong ccTLD, the .COM goes into legacy mode with the new registrations in the local ccTLD being much higher than the numbers of new registrations in that country for .COM and other other gTLDs. The4 .COM still has its place as a global TLD but its glory days of the DotCOM bubble have passed. In many countries it is barely above replacement level in terms of registration volume. It is the the US market that keeps the .COM at its current level because it s the de facto US ccTLD.

Around 2003, many of the ccTLD registries stopped providing access to their zone files. This meant that search engines no longer had regularly updated lists of new domain names in various ccTLDs. That clobbered search engines with inefficiently designed detection because without the feed of new domain names, they became completely dependent on following links. About 14 years ago, SEO heads were swallowing the Google Koolaid about "toxic" backlinks. With the rise of the the ccTLDs, Google and other SEs found it much harder to find new websites because owners were being scared into not linking to other websites due to the FUD propagated about "toxic" links. The link model of the Web changed. That did create an opportunity for SEO heads. It is hard, at times, not to think of them as a congregation of witch doctors trying to read the entrails of Google's latest update.

The localisation issue has become much more important since then because most countries have changed from the outward phase where businesses were mainly trying to sell globally to the local phase where most e-commerce is local. That shift also results in an impact that should have been detected by any SEO head worth their salt.

Google was always very good at the Dead Web. That's the static websites that rarely changed. It completely sucked at Social Web. It tried to compete with Facebook and failed. It tried to compete with Twitter and failed. The Social Web has become much more important as a means of promoting websites. At a ccTLD level, it is essential because communities find out about new websites through word of mouth via Social Web (the Live Web). This kind of hyper-localisation builds communities for websites and Google is no longer a member of these communities. That's where much of SEO needs to be focused much like the PR industry of old. The importance of community has been mentioned upthread. It is likely to become much more important.

Apart from the geographical new gTLDs, most of the 2012 round of new gTLDs have been a disaster in terms of renewals and longevity. Some of them can see more than 80% of new registrations not being renewed on their first renewal. The new gTLDs that have become successful, like most of the geographical gTLDs that focused on cities or regions, have targeted well defined communities. These geo gTLDs also resemble the ccTLDs in terms of blended renewals in that they have around 75% of all domain names coming up for renewal in a month being renewed. These geo gTLDs are very different from the Potemkin Village of the .EU which has a Web Usage rate (developed websites) of around 18%. Genuine ccTLDs can have Web Usage rates of up to 40% or more. Most of .EU domain names are registered for brand protection reasons and are often used as redirectes. The problem with .EU was incompetent politicians, "experts" who hadn't a clue about the domain name industry and awarding the contract to operate what was really a gTLD to an inept registry with no expertise in running a gTLD. At a domain name level (and this is incredibly important for SEO), people begin to identify their local ccTLD as being *their* TLD and the extension becomes "invisible" because people expect that a website targeting them has the extension of their local ccTLD. That's the ultimate community. The way that people register domain names also changes because the domain names become semantically driven with business names and geographical keywords becoming more relevant than generic domain names. People beging to develop a kind of map of their locality of commonly used websites much like their real-life mental map of the local pubs, shops, and supermarkets etc.

Regards...jmcc

One of the topics that comes up a lot with Domaining is the age of a domain name. The natural renewal/deletion life cycle of gTLD domain names was broken by large registrars moving expired domain names to auction websites for resale. When a registrar does that within the time before a domain name is completely deleted and the domain name is resold, it retains its original registration date. When a domain name is deleted, that historical registration date is lost. When it is reregistered, the date on which it was reregistered becomes the new registration date. However, some domain names will still retain their backlinks profile from other websites and that makes some reregs more valuable in financial and SEO terms. This may be one potential reason for the FUD about "toxic" links. There is also a natural attrition rate with websites due to their domain names being deleted so it is not uncommon for the websites providing backlinks to have been deleted. Some websites actively block services like Ahrefs/Semrush etc.

Some search engines may not be able to tell the difference between naturally acquired backlinks to an old website and links to a reregistered domain name potentially with content scraped from Archive.org. It is actually very easy to check at a small scale. Registries will limit such queries and that makes it difficult to do at a large scale. Perhaps some posters more familiar with SEO could explain if domain name age became a less important signal for SEs.

I checked the last new appearance year for the deleted November 2025 domain names in .COM. It is a zone file based process that checks the last time that a domain name was marked as "new" in a zone file. These are the figures for 2025 to 2015. The databases here have coverage back to December 2000. Out of over 900 million domain names, only 276 million active domain names (gTLD and ccTLD) were in the December 2025 web survey.

Year - Domains - %
2025 - 136,574 - 4.816
2024 - 1,144,844 - 40.372
2023 - 295,433 - 10.418
2022 - 146,437 - 5.164
2021 - 111,345 - 3.926
2020 - 80,283 - 2.831
2019 - 63,902 - 2.253
2018 - 39,445 - 1.391
2017 - 35,521 - 1.253
2016 - 24,375 - 0.860
2015 - 28,219 - 0.995

The domain name "year" between registration (or anniversary of registration) and deletion is longer than 12 months because of the post-expiry grace periods during which a domain name can still be renewed. Domain names registered in December 2024 are going through their renewal/deletion phase and that will complete by April 2026 for most of them. In both domain name terms and SEO terms .COM is not a single market. I don't think that I've heard or read any SEO head refer to it as being a single market.

In December 2005, the first renewal rate was approximately 70%. The markets for some countries were still on dial-up Internet access and their infrastrcture was not as developed as it is now. As a result, their domain name and website footprints were much smaller. Large gTLDs like .COM are not single markets in terms of SEO. They are a composite of many country level markets with a small global market. The registration, renewal and development patterns of these country level markets reflect the dynamics of their countries.

Regards...jmcc

@jmcc

Aha, I remember reading your posts about this in the domain name forum however did not understand the relevance at the time! You make some very worthwhile observations, especially communities of all kinds, that reflect my experiences too. Which way now Batman ?

@RedBar Same as it has always been. Build for people, create communities so that the site is not reliant on SE traffic, adjust for SEs only if necessary and only if it does not affect user experience.

Regards...jmcc

The common thread here is that Google isn’t “ranking the web” so much as arbitrating visibility across ads, platforms, brands, and communities. @jmccormac points on web churn, localisation, ccTLDs, and broken link signals explain why classic SEO signals feel incoherent - while my earlier post, @Juniya, and @Kendo seem right that “deserving to rank” is no longer the deciding factor.

This aligns with where I think it’s heading: demand routing + trust-by-proxy (brands, platforms, communities), not page-level optimisation -expanded in my other OP here, which aligns with @martinibuster's article:

[webmasterworld.com...]

Build audience, community, and direct demand first; SEO now follows that gravity, not the other way around imo

Just on that churn idea as it relates to backlinks. I wrote a few queries to calculate the numbers of domain names that have had no deletions recorded and the domain names that have had deletions (effectively re-registrations). It is running on the main legacy gTLDs and the results from the .COM 'a' domain names (10,850,083) are 67.1974% and the percentage for the ones with a drop (deletion) recorded is 32.8026% The database covers all transactions from 2000 to 2025 and is based on monthly zone files. The new/deleted transactions are compared against the 01/December/2025 zones. Should have some solid percetages later tonight or tomorrow along with a breakdown by the year of last new transaction. Not all domain names have developed websites. The linkrot problem (backlinks disappearing) could be affecting SEO backlink lists especially if they do not track domain names.

Regards...jmcc

ALT attributes primarily provide crawlers with contextual signals about an image and help accessibility layers interpret visual content. They are not a direct ranking factor.

What matters is how the image fits into the page structure, the relevance of the surrounding text, and whether the content block supports the topic. Tagging decorative or space filling graphics usually has zero SEO impact and just inflates audit reports.

I came across this on a SaaS platform where a comprehensive audit flagged thousands of “issues”, including ALT tags and massive lists of toxic backlinks. After manually reviewing the link profile most of those domains turned out to be old directories, mirrors, or scraped pages that search engines had already discounted.

Later I ended up working with TESSA in Northern Virginia, and the analysis focused on things that influence indexing, crawl depth, internal link paths, and which external links were still passing signals. Much less theoretical noise, more structural fixes that affect how the crawler processes the site.

ALT attributes primarily provide crawlers with contextual signals about an image and help accessibility layers interpret visual content. They are not a direct ranking factor.

What matters is how the image fits into the page structure, the relevance of the surrounding text, and whether the content block supports the topic. Tagging decorative or space filling graphics usually has zero SEO impact and just inflates audit reports.

I came across this on a SaaS platform where a comprehensive audit flagged thousands of “issues”, including ALT tags and massive lists of toxic backlinks. After manually reviewing the link profile most of those domains turned out to be old directories, mirrors, or scraped pages that search engines had already discounted.

Later I ended up working with TESSA in Northern Virginia, and the analysis focused on things that influence indexing, crawl depth, internal link paths, and which external links were still passing signals. Much less theoretical noise, more structural fixes that affect how the crawler processes the site.

I think SEO still matters in 2026, just not in the old checklist way.
Things like keyword stuffing or obsessing over every audit warning don’t seem to help much anymore. What’s working more is clear site structure, good internal linking, and content that actually matches user intent.
Also feels like search engines are deciding what gets seen, not just ranking the best page. Relying only on search traffic seems risky now.
Are you seeing stable results from SEO alone, or using other channels too?