February 2026 Google Search Observations

[jmccormac]

@Christianz That is is a fascinating insight. All of the SG maggots I have seen are just scraping raw html. There have been a few Starlink attempts on other pages but those pages are rate limited. What I have seen in some of the monthly gTLD Website/IP surveys is that some gTLD websites are hosted on Starlink customer IPs. Given the rise of residental proxies, there may also be a connection there with some of the SG maggots (crawlers and spiders being from search engines, maggots being from AI/scraper botnets).

Regards...jmcc

[edited by: not2easy at 11:48 am (utc) on Feb 11, 2026]
[edit reason] New month, new thread [/edit]

[ichthyous]

I'm seeing a big increase in AI Overviews here in the US. Maybe that's to blame?

I'm also seeing some fairly massive drops on my highest traffic pages...I checked and the ranking has been scrambled. It seems that now each request the ranking changes considerably.

Now there is a huge block of images at top, and the a stupid instagram account at #1 for some important searches. I am fairly certain that most people don't want to see some rando Insta account for these searches. Then we have the endless clone how-to articles. Anything else really getting pushed down a lot, which is why my sales are basically at zero I think. I had an eight day span with not one inquiry coming in from anywhere in the world. Then when it did it was just a competitor pretending to be a customer to find out my pricing.

Let us not underestimate how the current economic and political climate is also putting a damper on spending. That is a huge factor at play right now because people don't feel secure to spend or make plans...especially not if AI is coming for their jobs, which it is.

[Fluff_Nutz]

Looking at Analytics once again. I'm seeing a big push for other traffic sources that seemed dead 15 months ago. Traffic being converted into other sources. Whilst sources that was leading the pack now dropping, daily. This is definitely a conversion of traffic sources and manipulation. But why did it take 15 months to complete the reversal? Why are, once dead, traffic sources reappearing? Why is traffic converting into other traffic sources? I first noticed this like 20 months ago, when this once dead source was leading, then suddenly died. I did nothing to cause it to happen. I barely touch Analytics. I, then, concluded that the traffic from this source got converted into the, then and, now, dying traffic source. Which led the pack for the past 15 months until now. Does this even make sense?

Because of this revenue is, once again, dropping. So I stop the Singapore bots. Which also revived a, once dead, traffic source. Only for G to hit another similar plan to drop both traffic and revenue again. With, yet another, once dead traffic source. This is all a bit much but what I can confirm is that traffic sources do not tell the full story and that traffic from other sources do get converted into them. Whichever way to make sure publishers still loose out regardless.

The origins of my views are all normal. Nothing out of the ordinary. Just a sudden revival of, once dead, traffic sources and conversions. I also really dislike ads needing traffic from Big G, just because its considered organic. G are the reason we are getting a lot of bots. So its not entirely honest. Organic traffic can originate from other areas but ads do not seem to agree, for some reason. They just keep falling into G's trap. Whilst us publishers keep paying for it.

I also can confirm ranking issues here too. Though not big, there are still noticeable fluctuations.. So I can kind of see my issues here, but again, its out of my control.. G really do want us dead..

[Dooku]

@ichthyous,
Sorry I did not reply to your question, I do not check this forum on a daily basis.....as you might have guessed :-)
But I will reply here in order for others to benefit from this.
Yes, I did the same in the beginning, use Chatgpt to analyze the nginx log files on the server and the Cloudflare Analytics log file.

But you need to feed it as much detail as possible. If you just copy and paste a bunch of nginx log file entries then chatgpt and Cluaude will see all technical information, but they will not pick up trends or other relations.
For example, chatgpt will see a bunch of hits coming from some location, ip range and user agents. But it will not see that in that sample the majority comes from samsung galaxy S5 or a similar very old iPhone model...and WHY that is important! As NOBODY uses these anymore except phone farms and bot farms with spoofed user agents you need to tel it that these are completely obsolete hardware. So that info can also be used when creating a WAF rule to block bots.
Once you start experimenting with this and checking log files that way you will start to see more and more trends and relationships......yes there is MUCH more to see than one initially thinks. And then you can feed chatgpt more and more info. This also allows to see the difference why a bot is not mitigated in cloudflare and does appear in Statcounter because it was able to run the initial JS enough to appear as a visit.

After a few weeks of doing this you will even be able to block bots coming from residential IP's because you have more insight in their behavior.
It really is worth your initial time investment and will save a lot of time, effort and resources later.

[RedBar]

I'm not even bothering looking at other country SERPs since my UK widget results are utterly deplorable, pages with hardly any information other than keywords plus a couple of utterly rubbish images.

HOWEVER my main reason for posting is that G is now achieving its clear-out / localisation of its US SERPs.

Why do I say this? Two weeks of February and two of my global sites have not had ONE genuine US visitor this month whatsoever. Ten years ago I suspected this was the direction G Classifieds would be heading and, for me and my bulk-supply widget competitors, this looks about complete.

I understand G's reasoning for this, there are probably a minimum 100 more potential searchers for retail widgets than bulk widgets, in fact if G said 1+K it wouldn't surprise me, but what it does mean for me and most other bulk importers and exporters, is that a bulk purchaser, especially in the US, could find searching extremely much more difficult.

The first 20 years of The Net were interesting and mostly enjoyable, the last 10 years have been a rocky and fairly unpredictable ride, the next 5 years alone for many companies like mine, could be a very decisive period of whether to even bother expanding or not.

[mosxu]

You guys are using cloudflare dns ? What sort of percent passes cloudflare challange ?

[Micha]

If this “Discover Core Update” is really only being introduced in the US, I'll eat my hat. The fluctuations were terrible for days, with Semrush and others showing extremely high fluctuations. Currently, my news site is performing well in Discover again and the ranking has improved significantly, so money is finally coming in again through advertising. On the other hand, the shop is currently struggling.

Interestingly, all the bots have given up. I haven't had any noticeable traffic for two days, and there's nothing unusual in the logs either.

[morpheus83]

Feb 16 was a bit hopeful with discover and news picking up, but the traffic has again gone down. G isnt surfacing new content. I sincierely hope the update is coming to an end.

[jmccormac]

Connectivity outage for 4 days here. Be interesting to see what happens with Google traffic on the site after the outage.

Regards...jmcc

[jmccormac]

If Google is simply relying on geofeeds and IP addresses for US geolocation/localisation, it has a problem. A lot of IPs from other countries are designated as US IP addresses despite the range owners being in other countries. It is a very complex thing to solve and I am not sure that the Googlers have the expertise to do it reliably. The GDPR/WHOIS idiocy hasn't helped matters either.

Regards...jmcc

[ichthyous]

You guys are using cloudflare dns ? What sort of percent passes cloudflare challange ?

Yes, and CF doesn't give you that information. You used to be able to see it per request but I can't anymore. I would say a high percentage of requests make it through their 'managed challenge', where they determine via various methods whether it's real traffic or not. I have switched to interactive challenge for some traffic when I see it spiking in a suspicious manner from India, Singapore, etc. That stops it cold and traffic drops to a fraction. Am I losing some real visits, possibly. Those countries are not super important to my business so I don't really care. China is blocked completely.

[ichthyous]

Has there been an update announced? Traffic has been demolished lately...USA is off, but UK is down 50%+, EU and Canada also much lower. It's been getting worse for over a week now. Yesterday my site completely vanished for terms I have ranked at the top for over a decade. Gone completely. I also see this with some of my competitors and brand new crappy blog sites and instagram profiles at the top...insane.

Also...am I the only one seeing big reductions in backlink count at search console? Google seems to not be counting links anymore...

[Martin Ice Web]

Very, very big drop yesterday at about noon.
Serps are complete mismatched. Every entrie is a different site with a non matching item. Very wierd.

[RedBar]

@jmcc

If Google is simply relying on geofeeds and IP addresses for US geolocation/localisation, it has a problem.

I have no idea precisely how G would remove non-US sites from its US SERPs for some / all sectors however is G still a domain name registrar? If so removing non-US registrations would be an easy task.

Filtering results simply based upon locality / State would be simple since this is already done and can be seen when using a VPN by State and seeing the results. No matter how it is done the fact is that non-US businesses in my widget sector are filtered-out / removed from its US SERPS.

I can also confirm this also happens in the UK too which is now making me consider reinstating my 90s / 00s strategy of separate sites for each target country or, basically, leave everything as-is but concentrate more on trade fairs again. I do need to point out that this "issue" is most prevalent in the West and English language first countries however this may only be a matter of time.

[Micha]

You can indeed see a significant reshuffling at Google. In my niche, websites are appearing that I’ve never seen there before. At the moment, it feels as if news sites are being pushed. My news site has been receiving Discover traffic again for about a week now, and it’s increasing day by day. But I can also confirm what Martin is saying, it really is strange.

[jmccormac]

@RedBar A geofeed file allows the rangeowner to designate an IP or range of IPs as being associated with a country and a location. Whether the range owner is being truthful is another issue. The geolocation services use these files and search engnes probably do so as well.

Millions of IP addresses that were intended for the African region were "acquired" and ended up being used by Hong Kong and Chinese companes. There are also businesses that buy or rent blocks of unused IP addresses and rent them. Some of these businesses publish geofeed files of where their customers are "located".

Targeting a market for business purposes is one reason for flagging the IPs as being in a different country. The other obvious use is for VPNs for streaming and media services. Some of the IP location services can have incorrect location data for IP ranges because the WHOIS data from the IP registries can be decades out of date or even wrong. Designating a website's IP as being associated with a country for business purposes can work. The smaller the number of IPs in a a range associated with a country other than the rangeowner's country, the more likely it is for all IPs in the rangeowners set of IPs to be associated with the rangeowner's country rather than the designated country.

On the web hosting side of things, in-country hosting can be better than reflagging an IP address. There are lots of domain name and web hosting resellers who use templated websites and false address information on the websites. German websites have an impressum (effectively a copyright notice) which is more reliable.

When it comes to Cloud hosting and services like Cloudflare, the location indicators are limited. The language of the web pages might help as an indication. approxaimtely 9% of the gTLD websites are on Cloudflare (W3Tech's "figures" are wrong). Amazon is also a huge operation in terms of hosting and many large operations use Amazon and Google Cloud for their websites.

A site per country-level market makes sense. If the country has a strong ccTLD, then using the ccTLD domain name is the logical choice because people are more likely to consider a site in their ccTLD to be more trustworthy and local than a .COM site.

Regards...jmcc

[mosxu]

@ichtyous

Not sure if you use paid plan but for a country one can see the challange percentage! If you are from USA must have a rule for USA "managed challenge or interactive" and technically only 1% do not solve the cloudflare challange out of humans!

[christianz]

Record low traffic, record low revenue. Everything is worse than ever.

[saladtosser]

Highest non stop volatility I have seen in a decade, constant massive shifts several times a day back and forth. Sites never seen before pumped to top then dropped a few hours later over and over. Not that position matters that much nowadays with AIO thanks to the corporate Robber Barons and Epstein class we governed, owned and ruled by!

[Taran]

The rate limiting on Starlink pages is getting aggressive. I’ve noticed the same thing with gTLD surveys lately. It’s a cat-and-mouse game between residential proxies and bot detection.

[RedBar]

Ten hours into my Google day with traffic at 20% ... Cliff edge on a Friday!

[Sonika Khatri]

I’ve also noticed fluctuations across local service keywords this month.

Pages with stronger topical depth and clearer content structure seem to be gaining visibility.

Has anyone observed whether entity relevance is playing a bigger role recently?

[christianz]

Pages with stronger topical depth and clearer content structure seem to be gaining visibility.

Only spam and slop is gaining visibility. I don't have any spam or slop sites on my own, so nothing is doing well, but even among my own sites, the less value it provides, the less authority and history it has, the better it is doing.

[Micha]

@christianz I wouldn't say that my content is spam or inferior, but I've still had a good week, an amazingly good week in fact. But you're right, you see more and more nonsense in the results. Maybe Google is pushing news for once, who knows.

Shop topic: The week was terrible until Friday, but since then things have been picking up again.

[RedBar]

Welcome to WebmasterWorld Sonika Khatri

Has anyone observed whether entity relevance is playing a bigger role recently?

For those of you, like me, who haven't a clue what this means DDG Search Assist says:

Entity relevance refers to how important a specific entity is within a piece of content, which helps search engines understand its significance and context. This relevance is determined by factors such as the frequency of mentions, the sentiment associated with the entity, and its relationships with other entities in the content.

Do I assume this means the subject's widget word relevancy count?

If not please explain.

[Micha]

What a strange week: the rankings are a rollercoaster ride, as is traffic from Google. The news site was doing well, now slightly better than the week before last, with a lot of Discover traffic. But according to Sistrix, the ranking fluctuates practically every hour. The shop was dead until Friday, then suddenly exploded and had very good sales until Sunday. Today: slow again.

Search results here in Germany: chaos. I constantly see different results when I refresh a search, but there are fewer spam sites.

[Martin Ice Web]

@Micha,

yes sales have gone down the drain again.
User engagment is down the drain too.
Serps are highly mismatched.

[RedBar]

The Lithuanian UAB Baltic bots have been very busy the past 24 hours on several of my sites trying the wp-login.php probe. Some sites had 1100+ page attempts others 2200+ pages. All were 403d with my total montly unauthorised attempts at many tens of thousands of pages and very little bandwith.

Will UAB and MSFT ever realise their failed attempts and remove my domains or is this someone's "job" just to be a pita?

Yes, I do realise it's not a person, I''m just wondering if they were sane enough to write a removal instruction after XXX many failed attempts?

[jmccormac]

An uptick in MSFT probes. Those ranges have turned into a right dump and now get blocked at range level by default. A lot more attempts at scraping deep content. It seems to be for AI slop rather than for recycling content. Usual countries (CN/SG). There has also been a lot more from compromised boxes on ISPs. Traffic levels are back to normal after that 4 day outage.Some mobile ISPs are asl attempting to scrape. It looks like a distributed botnet using pre-paid/pay as you go SIMs. That LT operation (Baltic net) also has LV ranges. Ironically, there were more queries on Google for the site when it was offline. I suppose that's a good thing.

Regards...jmcc

[ichthyous]

Everything snapped back into place as if nothing happened for the last ten days. I had lost a ton of top ten ranking terms, almost all back now. Traffic back to normal globally. These shakeups literally kill off business for weeks at a time so these days you get the brief bursts of business from Google search and that's it. The only bright spot is that the nonsensical tariffs have been struck down, and they can only be extended for 150 days now so looks like they will be a thing of the past. For those of us in the USA, the declining dollar hasn't helped much...

[jmccormac]

Spent about 15 minutes trying to work out what Google AI slop was talking about with a question about mod_rewrite. It seemed that it recycled some rubbish from someone who didn't understand basic logic. Had to revert to a paper book to find the correct answer. Generally, Google has become a lot less usable due to AI slop. Traffic levels do seem to be normalising though.

Regards...jmcc