Google Updates and SERP Changes - September 2016

[NickMNS]

---

System: The following 4 messages were cut out of thread at: https://www.webmasterworld.com/google/4814310.htm [webmasterworld.com] by robert_charlton - 3:01 pm on Sep 1, 2016 (PDT -8)

---

I agree with you that features such keyword, location, browser type and so on (not set) may well be someone blocking the info for privacy reasons.But @ionguy my understanding is that whenever the hostname is set to anything other than your domain name then it must be a bot, this includes (not set). In so far as hostname specifically is concerned (not set) is always a bot and not a privacy issue. Filtering out the hostname (not set)s and other foreign domain from your data is essential to getting accurate stats.

This is applies to piwik and other analytics tools as much as it does to GA.

[highlander888]

Initially I saw a nice jump in traffic and rankings for 3 Penguin hit sites (sept 14 hit) these have been throughly cleaned up for about 12 months and it was the first recovery they experienced, all three have drifted back down now, feels like a "trial run" of Penguin or if not something similar related to backlinks - all 3 sites have diverse backlinks - some very strong, some weak, some spammy. The problem links have been disaowed or long since removed.

Generally I am seeing sites not noticed before, or older competitors who had not figured in the SERPS for ages popping back up. Like others seems to be a good amount of churn going on.

Not sure what to make of it - either it will settle or something bigger will roll very soon

[Jez123]

@highlander888 Fingers crossed it's a good sign. i feel that it might be.

[BushieTop]

@jez123 - what are you expecting?

[Jez123]

@ BushieTop - Expecting is too strong a word :-) I am HOPING that the rise up the rankings will come back into play permanently.

My site was at least partially released from Penguin in 2014 and one sector at least was not released. I am hoping that I will see a full or partial release for that sector. I have seen only ranking changes in the sector and none of my other sectors. So, hopefully, but who knows?

[BushieTop]

@jez123 I'm not going to hold you to any prediction, just interested to see what your thinking - so i assume you think something rolled out ~1st Sept ish and then was rolled back in the last 48 hours? You're thinking it will roll out again and that was some sort of test... am i close?

[renatovieira]

Someone is noticing an unbelievable drop clicks today?

I have on my site now, ZERO active users.

This update is being one of the strongest for me.

[Jez123]

@jez123 I'm not going to hold you to any prediction, just interested to see what your thinking - so i assume you think something rolled out ~1st Sept ish and then was rolled back in the last 48 hours? You're thinking it will roll out again and that was some sort of test... am i close?

Something rolled out for sure but not sure if it's rolled back or has been altered /fine tuned. I am hoping it was a test and some glimpse of the future, for sure.

What are your own thoughts on it?

[BushieTop]

@jez123 I'm hoping the same as you. Whatever rolled out on the 1st September worked much better. In our verticle for competitive terms, we saw the right people rank at the top, that's changed in the last 48 hours. this vertical (ecom) now has content sites littered all over it, including subdomains from highly powerful sites like the telegraph. They are just serving as affiliates.

Although, with both outcomes we're still seeing paid links with EMAT rank top 3

[Shai]

Guys, will you do me a favour please? All those with large drops (2 -> 8 would qualify) will you be able to check the number of server requests your target page is making, size of the complete page and also the time it takes to load it up? You can quickly check using Chrome or Firefox.

1. Clear the cache first
2. open chrome
3. right click choose inspect
4. click on the network tab
5. type in the address of the page ranking
6. right at the bottom you should see n of requests, size transferred and time to load.

p.s if the status code is 304 appears in the status column then you have not cleared your cache. Its vital you do that before posting results.

If you could also specify if the site is CMS (WP or similar) or static/php etc that would be great. Also, if you could specify how terrible the drop is that would even be better.

We think we are seeing a pattern.

thanks,
Shai

[BushieTop]

Custom CMS, with WP blog, drops 6 to 10, 180 requests, 5.1mb, 10.56 s

[Shai]

wow that's quite hefty BushieTop. Might be something to look into improving!

Just as an example (removed the urls) here is the new rankings for one of our clients. Annoyingly, we talked about speeding things up just prior to this update.

1. https://www.competitor1.co.uk/sell-widget/ | 22 requests | 440kb web site size | 0.63s loading time
2. https://www.competitor2.co.uk/sell-widget/ | 24 requests | 345kb web site size | 0.15s loading time
3. https://www.competitor3.co.uk/quick-widget-sale/ | 13 requests | 326kb web site size| 0.10s loading time
4. https://www.competitor4.co.uk/ | 76 requests | 1420 web site size | 1.60s loading time
5. https://www.competitor5.org.uk/en/articles/quick-widget-sales | 48 requests | 1197kb web site size | 2s loading time

8. https://www.ourclient.uk/ | 62 requests | 6900kb web site size | 2.53s loading time

Look at the difference between the top 3 and the rest of the pack. I am seeing 2 further examples of this exact same situation. Seems to be the top 3 results where you see the biggest difference in speed compared to the rest of the pack.

Might be something to check if you have seen lots of movements.

[Simon_H]

I'm going to throw something out there. It might be nonsense, but at least worth ruling out.

Google has just dropped/lowered the position of the image widget in the SERPs. This will have a major impact on the CTR of all results in searches that return images. Google says that CTR has no impact on rankings, but there is (limited) anecdotal evidence that a sudden change in CTR does at least temporarily impact rankings. If this is true, then the image widget change would have a major effect.

Could this be contributing to what people are seeing today? Those seeing big changes... do the keywords seeing the changes usually return images in the SERPs? And do the keywords seeing no changes usually return no images?

[NickMNS]

@Simon
I have not seen any significant changes in traffic and keywords leading to my site generally have no images.

[javadth]

in my own case , i just can say ; i used many links for my first page , natural links but all for my main page and main keyword , without spreading
--
what is it , i am partly sure it is about to OVER SEO , more than normal using on/off- site seo

or

it is new update around penguin and rank and social signals , but didnt work perfectly ( out of control )

[Shai]

I'm afraid I don't see any signs of this having anything to do with dodgy link building. No sign of Penguin either. Lots of competitor pages that are very spammy still there while two clients with very clean, real editorially given links from the likes of the Guardian, Telegraph, local newspapers etc for research dropped substantially.

[NickMNS]

@Shai, what about Simons suggestions. Do the affected sites rank for keywords that normally display images in the serp?

[Shai]

I'm afraid not. No images show up on any searches that come to mind.

[30K_a_month]

@shai
If spammy competitor is still there and has always been there then hes already got enough authority. Your very clean clients need to get some stronger links?. I dont think you can dismiss link building with your information.

[Shai]

If a site is very spammy, it does not matter how much 'authority; you think it has, it is always at risk of going down as soon as Google catch up with it; Algorithmically or manually. As you have no idea what information we hold, I would say your last sentence is an assumption. Obviously our observations are just that, observations and theories but I'm trying to base them on real facts. As opposed to what you seem to believe, we hold quite a lot of data on quite a lot of sites.

[edited by: Shai at 7:07 pm (utc) on Sep 15, 2016]

[Simon_H]

Thanks @NickMNS. Given there is strong evidence that the image widget has dropped down the SERPs, and there is also strong evidence of algo change going on, there are only 3 possible scenarios here:

1. The theory that CTR impacts rankings (even temporarily) is false, and so the algo changes are independent of the image widget move.
2. The theory that CTR impacts rankings (even temporarily) is true, and the algo changes are entirely due to the image widget move.
3. The theory that CTR impacts rankings (even temporarily) is true, and the algo changes are due to a combination of the image widget move and also a separate algo change at the same time.

If I had to guess (and it's purely conjecture), I'd go for 3. I think it's interesting Google should throw that image widget change into the mix right when there appear to be other changes going on and Penguin is due. Almost as if it's providing a smokescreen...

[Shai]

Simon, The CTR effecting results temporarily is well documented and in my opinion proved by several experiments such as this one: [moz.com...]

I don't see how the movement of the image widget would effect organic results though for terms that have nothing to do with images. Both examples we have are the furthest away you can imagine from having images produced. Its a clever theory but I just don't think it fits this situation.

[Simon_H]

@Shai There's a very split opinion on the CTR vs ranking thing. The studies are quite flawed. But, yes, given people have seen movement on keywords that produce no images, that suggests that there is a combination of things happening.

[Shai]

I don't think its ever possible to hold meaningful yet watertight experiments in such a varied and volatile environment. There are far too many variables and the more you focus on the subject matter, the better the results are but the less they tell you. We recently carried out an outgoing link experiment examining the effect of outgoing links on rankings in a manner that was as controlled as possible. To make it as controlled as possible, we had to forgo checking the effects of other variables such as NOFOLLOW/FOLLOW outgoing links. Even that, although as tight as possible, can be argued to be flawed from a scientific perspective. Its like trying to weigh yourself accurately on a moving train.

The thing is that even if the results of some of those tests are not 100% reliable, we have to accept that in our field 80% or even 70% reliability as good enough.

[Simon_H]

@Shai I can't call whether it's true or not. Yeah, very few studies are going to be flawless in that space. My personal view is the same as yours. And if rankings are indeed impacted by CTR, even temporarily, then the image widget change will have caused some chaos in the SERPs, on top of whatever else is happening at the same time, making the whole thing very confusing indeed.

[sench]

@Shai regarding the speed and requests, I compared this and they are either relatively the same as the top 10 results or even better. That's for 1 site that had the biggest drop these days. It's also not pengiun-related. Still no big solid indicator like that

[liamkk]

@Shai

Website speed has no impact on ranking. Sites that come in top 10 results are generally faster because they get regular traffic and that keeps their caches warm.

Most CMSs have built in caches which means if someone accesses page A every x minutes, it will always stay in cache and hence will have higher speed.

On one of our sites, we reduced average crawl time (which actually matters the most) to less than 100m (from 1.2 seconds) using sophisticated caching algorithms in addition to getting 100 page speed score. It had no impact on rankings on over 4,000 keywords we tracked for the site.

[edited by: liamkk at 2:24 am (utc) on Sep 16, 2016]

[liamkk]

On CTR issue, I think CTR has significant correlation.

I tested 50 pages with a control and added a 'click bait' title on experiment pages. They went up 3 places on average in 2 months while control pages had no movements.

As a developer, I think Google search algo is limited to just taking following into account. (Thinking from resource point of view)
1) Links
2) A quality rank (Panda based)
3) CTR, use of back button (This mostly happens on big search terms)

I believe that first two factors also determine the most traffic available to a site. If your site traffic is highly consistent and predictable, it's likely that Google has applied a quality penalty on your site and they won't send you huge amounts of traffic even if a million people started searching your #1 term.

The rest is all fiction and overanalysis (mainly to justify the need for having SEO professionals),

[hopepro]

The first time that I need to register with webmasterworld to post on this thread. I'm seeing ranking like I have never seen before in the past 8 years and it started to rocked on Sep 12th in my country.

I normally ranked on 3rd to 4th and most of the thin content sites are either down from #7-#10. But now I'm down to #7-#8 while they're going up on #2-#3

Some say this happened last time with Panda. Crappy sites go up first and being slapped later....anyone any ideas?

[Shai]

@liamk the site you mentioned that you improved crawl time to less than 100md. Did it have any drop in rank over this update. If you actually read what I said you will note that my thinking is that this particular update had a speed element to it.

Also, I'm well aware of browser cache but we are talking about visiting sites for the first time.

You say speed has no effect on ranking but you really do need to qualify that with "in my opinion" unless you have done some extensive testing on the subject which you can share with us?

I for one believe UX, which speed is a component of, does have a role to play. It may be a small one at the moment, but we have seen some very limited evidence to suggest it does exist.

[liamkk]

@Shai

I am not talking about browser cache. I am talking about server side cache. To generate a page, a CMS has to run 100s of queries and it will always be slow. Most CMSs have built in caches which would store the entire generated page for a specified period of time and all non-logged in users will be presented with that version without querying the database again. The more a page is visited, the more chances of it being found in a pre-generated condition rather than CMS having to run those 100s of queries on each pageload.

The above mentioned site so far has had no impact in current update. A few queries have changed positions but that's mainly because of other sites moving up and down.

On page speed issue, there is just one study which shows a correlation between crawl time and SERP rank and warm server side caches are the only reason for quicker crawl times.

Browser side speed doesn't matter at all. If Google started penalizing browser side load times, they will be killing their own Adwords/Adsense business which is #1 obstacle in the way of sites scoring higher on page speed scores.