Google Updates and SERP Changes - Jan 2016

[Nutterum]

Continuing from:
Google Updates and SERP Changes - Dec 2015
https://www.webmasterworld.com/google/4780066.htm [webmasterworld.com]


I was off for the past two weeks, but looking back at the traffic patterns at the mid-end of Dec I can safely say something moved in the non US Google domains. Seeing positive changes in the UK, DE, FR and IT traffic. Asia is same but since we do not get much traffic from there either way I can't say for sure there was no changes in the SERPS there.

Too early to tell as for Jan. because it is the first workday of 2016, but I guess I can confirm my observations on Friday.

[edited by: Robert_Charlton at 10:56 am (utc) on Jan 4, 2016]

[RedBar]

I have noticed non-mobile friendly sites dropped a little.

That's been on-going since Mobilegeddon last May for my three remaining non-mobile sites and since the recent update all my responsive sites have all had traffic gains.

[engine]

According to Google's Gary Illyes, he continues, today, to confirm this is a core algorithm update, and not Penguin.

[Nutterum]

Trust me engine, when Penguin hits, you'd know. :) At the very least the blackhat forums will explode.

[Slinkywizard]

Nutturum hit the nail on the head. You never need to ask if it's Penguin or not. That's like asking if the firework's exploded or not – you'll know.

[superclown2]

Looks like a roll-back here in the UK. I'm not surprised.

[engine]

@Nutterum yes, i'm sure I would know. I was just passing on what Google is saying about the update, and they are continuing to say it's a core algo update.

[timemachined]

I don't think it's a complete roll back as my traffic is still 30% higher with 30% lower CTR. Mainly 1000 word content moved up. While site is responsive, the content perhaps isn't hence mobile traffic seeing longer content, users getting fed up so need CTA to be better perhaps. And or G mucking up those anchor link rewrites and users confused and back pressing so end up on adwords.

Is there a scientific term for a wave crashing, removing some of the sand and depositing some? In which case, I think G has several waves one big wave, then smaller algo waves, like earthquake after shocks.

[RedBar]

Is there a scientific term for a wave crashing

There are several different types of waves all resulting with different effects:

Spilling
Plunging
Collapsing
Surging

[en.wikipedia.org...]

[dipper]

semi-roll-back in Australia too

[aristotle]

The web has gotten so big and complex that Google's current algorithic approach just can't cope with it. They need to scrap what they've got now and start over with a new approach that integrates everything from the beginning. The current system made it necessary to add separate pieces like penguin and panda, and the different parts don't harmonize or work togethervery well.

[keyplyr]

I'm not seeing any roll-back, just the normal flux that occurs after most algo updates.

[ecommerceprofit]

We have been getting a ton of traffic - I see that "search partners" was turned on...I had this off - somehow went back on. I turned this off and my traffic is lower now - my cost per conversion was through the roof.

[ecommerceprofit]

Also, these last 6 months our chargeback rate is way up - we never used to get chargebacks...I mean just about zero...now we are constantly having to be paranoid and look at each and every order...cancelling several a day...never had this problem before...

My theory...Google's algorithm is based on "good customers who can afford to pay" - maybe based on "social economic status" - maybe based on "honesty score" - maybe based on "browsers vs real buyers" - sending us the dregs of society...I don't know

On purpose (mix good customers with bad) or not on purpose (algorithm charge that was made by accident) is causing our zombie / fraud traffic.

[aristotle]

the dregs of society

So you're suggestng that google shows different search results to "the dregs of society"?

[timemachined]

Well even though defined inappropriately and comments should be in the adwords thread... Yes G certainly shows different results to the poor, rich, on personal search, based on sex, age, income, routine, previous purchasing history etc. and not just in G but on websites displaying adsense.

Still laughing... the 'dregs of society' do however have the means to increase their income even though ours might go down. I started getting some shared revenue from one such dreg who regged an account two years and one month ago and just deposited a nice sum. Knock the dregs at your very peril ha. They might come good!

Idea: Capture the dreg's email, sell them some dreggy stuff...

[Simon_H]

So you're suggestng that google shows different search results to "the dregs of society"?

That would certainly explain why I don't see the same serps as everyone else.

[engine]

The personalised SERPs don't help and are somewhat disturbing, especially when trying to share with friends and colleagues. For the average user, personalised can give tainted views of what's really there, whatever part of "society" you're labelled as.

Importantly, and on the topic of referrer quality, it's got to be worth attempting to segregate whether that's mobile or desktop. It's of no one's interest, including Google's, to send bad referrers, and it's always worth checking, especially when it's an ecommerce site, whether there is a problem with the whole process of checkout, especially when on a small-ish smartphone screen.

[EditorialGuy]

Whatever happened to the War on Dregs?

[ecommerceprofit]

It may be in Google's interest to send bad referrers to keep our cost per conversion higher. Step one...test using on and off zombie traffic..send some from Detroit then send from the Hamptons...send some from Cleveland...then Aspen...watch adwords cost per conversion go up to perfect level...not too high not too low....get it to just the right level...

Then stop on and off zombies and turn on suffocation traffic permanently.

Another dreg is born...me...haha

* not just economic...could be gamers getting results for home decorating sites or home decorating users getting chainsaw reviews.

[ecommerceprofit]

Google allows us to adjust bids for mobile devices (if they did not there would hardly be advertisers). However, tablet traffic is much lower...so why not make a little extra cash here by not allowing us to bid less just for tablets. This is a little hole for us to see though as advertisers that they are out to give us low quality traffic. The zombie / suffocation traffic is impossible to prove until the government starts to regulate,,,I used to think this is just the algorithms...it is the algorithms but they are still written by people or AI that is trained by people.

[Simon_H]

@ecommerceprofit Actually, the zombie phenomenon is not too difficult to prove, certainly on paid. Our graph of cost-per-converted-click over time shows a relatively straight line apart from spikes representing zombie days. We also see a huge turbulence starting around mid September which is the date many others started noticing this and writing about it on WebmasterWorld. Our graph is plainly unnatural and any statistical analysis would conclude that. Google would struggle to show otherwise.

Apologies for talking about zombies on this thread, but we've certainly seen a huge zombie impact from the two recent organic algo updates. However we're seeing this impact on paid (Google Shopping) which suggests some significant 'leakage' from organic updates across to paid. I'm still monitoring results and will write in more detail about this once I've had a few more days of data to analyse.

[dipper]

@aristotle - comment of the day award!

"So you're suggesting that google shows different search results to "the dregs of society"?"

[dipper]

@Simon H - Interesting. Isn't the goal of the zombie to pump up CPC in AdWords? - (excuse ignorance on this topic) - if so why would there be leakage from organic? Surely the zombie would want to be targeting all AdWords customers in that case.

[ecommerceprofit]

To go one further...I can now visualize government regulation of Google and other large search engines. Banks are regulated...there are huge government departments created just to regulate banks, the EPA for environment, the health dept, etc. 25 years ago there was no web...but now information is not quite as important as water or electricity but fast getting there. Absolute power corrupts absolutely so that's why there is government regulation... this is coming from a conservative leaning individual.

I'm not saying Google is doing any of this on purpose...I'm only proposing theories but it's a hidden black box that no one can see into so the government must be allowed access because Google and others are affecting too many lives for them not to see in the black box. I'm not saying the government should see everyone's e-mail or private search information...just to look at the algorithms and how they work...when an update is done the government should be informed.

[edited by: ecommerceprofit at 5:58 pm (utc) on Jan 19, 2016]

[Simon_H]

@dipper The reason for zombie traffic is unclear. What seems to be consistent all the way back to when @tedster first wrote about this is that algo updates/flux seem to result in increased zombie traffic (on organic). One theory discussed is that Google has a test set of sites that it uses to throw test traffic at using test periods, e.g. sites that may be borderline Panda. However, many of us have also been seeing zombie traffic in paid, which really doesn't make sense as the organic algo and the paid algo are allegedly completely different. Which has made an explanation of the zombie phenomenon even more difficult.

However, based on what I've seen with the recent algo updates, I think there is substantial leakage between organic and paid. I don't just mean the indirect stuff (e.g. more above-the-fold paid links means less traffic for organic), I mean something more. Last weekend when the first algo update hit, our paid conversions dropped massively but traffic was unchanged. Then this weekend when the second algo update hit, our paid conversions seem to have returned. I'm also conscious that mid-Sept was when people first started reporting the zombie phenomenon on paid, but lines up with when some big organic algo updates were happening. So I don't think these are two different things. For example, if Google has a network of testers running test queries on the algo updates or a botnet doing similar, perhaps some of those are inadvertently (or intentionally?!?!) hitting paid results as well as organic ones. Or, as previously suggested, perhaps the testing is pre-query processing (Rankbrain stuff), so real users are getting duff paid results as well as duff organic results. Not sure yet but will try to post more soon.

[dipper]

@Simon H - thanks for that insight. Would make sense for Google to have programmed test bots (instead of humans) that have learnt from humans in the past on what they click on, and in what volumes - testing their result set would make perfect sense. Although, the developer in me says, surely this would occur in a test environment, and not live. The live result probably has enough "testers" with humans naturally doing searches that it doesn't need it. Sounds a bit strange - never seen it myself. Thanks again for your thoughts!

[Simon_H]

@dipper I'd agree and I'm sure Google must have some kind of test environment. But the fact they regularly release changes live and then can be seen to fully or partially roll them back suggests that they don't have a test environment that fully mirrors live, potentially because they can't justify investing in the huge infrastructure required. Let's face it... users aren't going to stop using Google because of the odd duff day of search results when Google is testing on live, and Google probably doesn't care a whole lot if they upset a very small subset of site owners.

[dipper]

@Simon_H - I agree in lots of ways BUT in regards to bots. I am sure you would agree, test environments can never 100% fully test software. The results from a live set of users would be far superior to bots. Bots could be good in a dev environment to test in mass, and based on prior modelling, but not as robust as a live set of users.

So if they do as you say then why have their own bots for the live environment then? - wouldn't those bots simply screw with their result set from the live results and make their data off? - it would, and no scientist would ever do that to themselves - they wouldn't be able to fully trust the results they receive. Is it possible these bots are from competitors trying to reverse engineer results, or from spam bots? .. hmmmm

[Simon_H]

@dipper This is all guesswork but... If Google had a test environment that was a genuine mirror of live, then they would logically be able to get real user searches from the live environment to run in parallel on both live and test. Hence, Google would then be able to 100% see exactly how their algo update would perform for real. They could also get any human or bot testers to work on this test environment rather than live. However, they plainly don't do this, which is why I believe that they can't have a test environment that mirrors live.

In terms of human or bot testers screwing up the live environment, that's precisely the point. Let's assume that Google has to do at least some testing on live; that's a fair assumption. And it's also a fair assumption that some testing must involve clicking on results in the serps - perhaps multiple times by different testers - and measuring the impact of doing so, the impact of different dwell times, etc. And it's also a fair assumption that Google will want to identify a set of test sites to use rather than just randomly hitting any site during testing. So when you say that these bots/testers will just screw up the live environment, that may be the very definition of zombie traffic, as these testers are indeed just screwing around with the traffic of a subset of sites.

In terms of whether this could just be competitors, the answer is that it can't be. Because the total traffic per day is unchanged with the zombie phenomenon, whereas if this were competitors or spam bots, then the crappy traffic would be in addition to normal traffic rather than instead of it.

[dipper]

@Simon_H

I'm not sure your logic quite works there though for testing software. The actual perfunctory process of performing the search is not what they would be testing - as that requires no testing, it simply happens. They would be testing the quality of the performance of those results - this means, how users interact with the results and what results perform the best. Do users enjoy a certain result set more than the previous set? - after-all this is the goal of search improvement right - to make users happier. Unfortunately these things can be cursory checked with modelled bots, but not with the same accuracy as the live environment unhindered. They wouldn't want to taint those results with a bot, or with internal human clickers as it fights against their goal - improving search results for users, right?

Interesting that you're observed zombie traffic replacing regular traffic - kind of zombie traffic shaping? What kind of volumes? In low volumes there is potential for there to be lower conversions and for that to happen within reason I guess.

Sounds strange though - Maybe my logic is sound, as what you describe just doesn't sound plausible (in all due respect) .. but maybe this whole thing is not sound ! Has JohnMu or anyone from Google ever made comment on zombie traffic shaping? You're not alone, Tedster(RIP) observed this too right?