Google Updates and SERP Changes - May 2015

[Mentat]

---

System: The following 3 messages were cut out of thread at: https://www.webmasterworld.com/google/3003161.htm [webmasterworld.com] by aakk9999 - 4:35 pm on May 1, 2015 (gmt 0)

---

Just thought I'd join this thread.

Our traffic from old man google has been consistent +-5% since panda 4 in May 2014. Our main site IS mobile friendly and wasn't hit on the 21st of April, however it suddenly lost 10% of Google traffic yesterday. Many our top 10 positions dropped 1-2 spots.

I'm almost 100% certain it wasn't panda (as other pandalized sites in my niche didn't drop positions in the queries I'm watching).
I'm about 95%% certain it wasn't penguin either as I did a massive disavow in mid Feb and not that many scrapers, weak sites have been linking to us since then


So what the hell happened on the 28th (29th Australia time)?

The rank drops occured on both desktop AND mobile.

@ColdRum

I believe it has something to do with that huge disavow file..

My disavow file is ~ 20 000 domain lines, gathered with a lot of work from a lot of sources, but my traffic follows a very strang downward path in the last 2 years, so I've said that I need to try something.
I've used as source only the info from WMT and I've selected only the worst spammers for 2 of my sites.

Guess what? I see increases in the last week!

P.S. All my sites are mobile friendly.

[Nutterum]

Well Google are trying to avoid the link-tied algorithm for some time now as they said on several occasions that it is not a perfect system, but there isn`t a better one implemented yet. Have you considered that what you see are the test bots, crawling and judging websites not by its links but by its Quality Score the same way the same bots judge the Adwords landing pages. I have to agree that I have seen one or two bots ending up in my honeyjar url. Times are changing and its not the strongest who survive but the most adaptive ones. I hate to be the doomsayer, but I suspect that in an Year or two, the web as we know it will obey different rules, not bound by backlink strength as a main factor for relevancy. Hell this is happening already albeit at a small scale.

[nomis5]

not bound by backlink strength as a main factor for relevancy

I think that started a year or so ago and has been the case for several months. So many have gamed backlinks that they are now almost worthless. My opinion is that time on page within your given sector, where people go when they leave your page, bounce rate and other similar factors are the key now and have been for some time.

[EditorialGuy]

Sure, user metrics probably matter a lot more now than they did a few years ago. (Martinibuster started a thread on that topic recently.) That doesn't mean links are out the window, though. Brute-force linking probably isn't as helpful as it once was, though. Even for sites that don't do link-building (such as mine), 95 percent of inbound links are obvious junk that can be ignored or discarded.

[seoskunk]

I think they have been using user metrics for about 5 years now and its more and more influence in the algo. I think user metric data was the main reason Chrome was created. The important question to ask is how good are the filters. Can they discard bots hitting websites or not. Because if there is a way to game the system I'm sure its already beeing done. Does the rise in referrer spam and clickbots indicate user metric spamming of a site? I did open another thread on this but its interesting its become a subject here as well.

[Mentat]

Another low point in traffic for me.
Is there any big hollyday?

[samwest]

Last 4 Friday sessions: 465. 466. 470, 468 - no kidding. Now THAT'S precision throttling. You can't pull that off without either a hard coded set point or some fancy AI. I tuned PID control loops for a living and you were either very good or very lucky to hit repeated accuracy like that. There is always a natural external force affecting your error signal. In the wild, this kind of fixed pattern never occurs or if it does, you should be playing the lottery.

I could tune a 2 million BTU gas burner to hit 500 degrees with great regularity, but there are little "natural" external forces other than ambient air temp, pressure and flow (controlled by a motor drive, so it's not natural). On the web however, there are almost an unlimited number of natural external forces that should be varying traffic week to week. These observed numbers and this analogy shed a lot of light (at least to me) on what's being taken OUT of the equation or intelligently manipulated.

[netmeg]

Most of the sites under my purview are being flooded with referrer spam and ghost referral traffic (which are two different things - referrer spam actually visits your site, ghost traffic just shows up in your stats) But that's not a Google thing. I was trying to filter it out, but almost as fast as you create a filter or block for it, they come up with new IPs and new referrers.

[samwest]

What I am seeing is Phantom traffic that shows up in stats and GA as human visits. I've never seen bots appear in Google Analytics, only real visits...and bots SHOULDN'T appear in GA Real Time view. Likewise, these same bots are appearing in PIWIK and WP Wordfence as human from Mountain View as crawl-66-249-65-52.googlebot.com (and others). My point being that you can't believe what you are seeing in GA is real traffic when a growing portion is now Phantom bots.
It may be just my hallucination and I hate to think that the plex would devise such things, but I can't help but imagine a bunch of eager software engineers around a conference table suggesting that they could make people think they are still getting traffic (now that they've corralled most of if) by creating a human looking and acting bot. It might make some think that "oh, everything's OK, I'm still getting traffic from Google". Problem is, this Phantom bot has yet to exhibit true human behavior, only a very unnatural dwell time than should make my bounce rate be about 2%, but bounce hasn't changed one bit. Just my observation. Take it or leave it.

All along Google has taken pride in it's ability to do incredible things, but in the edited words of Chris Rock : "you can drive a car with your feet if you want to, that don't make it a good xxxxxxx idea!"

[RedBar]

Another low point in traffic for me.

Saturdays for European traffic are easily my lowest days now.

[netmeg]

I've never seen bots appear in Google Analytics, only real visits

If bots can execute javascript (and they can) then they show up in GA and other stats programs.

Have you told Analytics to filter out bot traffic from the normal view? That'll at least keep the better behaved ones out.

[Itanium]

I think they just broke Google.DE. For the last 20 minutes I'm getting mostly foreign (mostly english) results for various search terms. And I'm not talking about terms I monitor. I'm talking about my private usage of Google.

[seoskunk]

These bots are executing directly on google's servers, no visit to the site so there is nothing you can do to stop it. You can filter them but they will still be hitting your analytic account you just won't see the damage.

[netmeg]

Actually there's some that visit and some don't.

[Hollywood]

Interesting mentat...

If anyone wants to take a look at mine (disavow file) I would appreciate it and need it, I am thinking we overdid it personally.

~H

[doc_z]

I think they just broke Google.DE. For the last 20 minutes I'm getting mostly foreign (mostly english) results for various search terms. And I'm not talking about terms I monitor. I'm talking about my private usage of Google.

I'm seeing this phenomenon since two weeks. SERPS are often full of non-german results when I'm searching for phrases which are the same in german and in english (e.g. nepal news on google.de with german language settings and location in Germany).

[silentneedle]

I can confirm this, english sites ranking higher in Google DE since some weeks for pretty much keywords. I'm also still seeing russian, chinese (wtf) and those damn unrelated google books results. I never used site: and bing that much before.

[samwest]

All I can say is WOW! In the past 50 days, the traffic on the internet has DIED. Oh, it's still there, or appears to be there, but it is completely non interactive. I'm watching GA all day today and every few minutes there is someone on my signup form. Right this moment I'm watching 3 users on my signup page and not one places an order. It's THREE form fields, fist name, last name, email then submit. I've trouble shot my site till I'm blue in the face. My hosting company saying everything is working fine. They've even check my secure forms over and tested them. This traffic must be so poorly targeted, but how can that be? I'm totally amazed and bewildered at what the web has become. Anyone else experiencing this incredible change? or is it just me?

[dannyboy]

@samwest Is your site wordpress powered?

[samwest]

@dannyboy - yes it is, and I'm afraid it won't be for long. The past year it has performed (conversion wise) poorly even though I've run even diagnostic imaginable on it. Runs fast, no canonical issues, etc. Think I'll simplify the site back down to hand coded html5. Everyone tells me it looks great. pff

[dannyboy]

@dannyboy - yes it is, and I'm afraid it won't be for long. The past year it has performed (conversion wise) poorly even though I've run even diagnostic imaginable on it. Runs fast, no canonical issues, etc. Think I'll simplify the site back down to hand coded html5. Everyone tells me it looks great. pff

@samwest I'm going to throw a couple of ideas out there.

1) Have you tested your site performance at [webpagetest.org...]

If so, do you consistently score well on your time to first byte. If not, do you know where the bottleneck is, such as DNS response time. Also, don't just test your homepage. Test the page where you conversions happen, such as your cart/checkout/registration page.

2) Does your hosting provider provide your DNS, or do you have a third-party DNS provider, such as [easydns.com...] ? If you're not hosting your DNS third-party, you may want to consider it.

3) Do you have a WordPress caching plugin installed and active, such as W3 Total Cache, Super Cache, or ZenCache? I've used all, and would recommend ZenCache over the alternatives.

4) Do you have XML-RPC disabled and/or blocked? If so, consider disabling it as it can drain you server resources. Read more here: [blog.sucuri.net...]

I've recently discovered bots pummeling my WordPress 'xmlrpc.php' file. I had no idea this was going on. I had thousands of hits to this file, many times in a single day.

You can enable logging to see if this is really happening to you:
[saotn.org...]

If you don't need XML-RPC, you can disable it the following ways:

Plugins:
[wordpress.org...]
[wordpress.org...]

But if you know you really have no use for XML-RPC, I would disable at the Apache .htaccess level above the WordPress rewrite section:

# Block XML-RPC
<Files "xmlrpc.php">
Order allow,deny
Deny from all
</Files>

This may be anecdotal, but ever since I blocked xmlrpc.php using .htaccess (after previously trying both plugins above) I've experienced traffic and conversion improvements.

BTW, be sure to also disable the useless XML-RPC header WordPress sends with each response. Just search for "hide xmlrpc.php in HTTP response headers" in the following page:
[deluxeblogtips.com...]

5) Is your site all SLL/HTTPS? If so, do you have the SPDY protocol enabled and activated on your server? You can test your server here: [spdycheck.org...]

6) Is your site susceptible to the Heartbleed bug (unlikely at this point, but worth throwing out there). You can test here: [filippo.io...]

7) Is your SSL certificate chain complete? Meaning, do you have the Intermediate CA properly installed. You can test here: [geocerts.com...]

It may also be worth testing your site's overall SSL configuration: [ssllabs.com...]

8) Have you tested your site against the Poodle vulnerability? You can test here: [poodlescan.com...]

9) When you test your own site using Chrome or Firefox, do you use private/incognito mode, so as not to pollute the environment with your own cookies and localStorage browser settings?

10) You mentioned using GA and how it seems your traffic is being shaped or artificially manipulated. If you're suspicious of such activity, then you may want to consider replacing them with another third-party analytics script. If your open to switching, I recommend [clicky.com...]

You can compare them here:
[clicky.com...]

They're free up to 3000 daily page views, then shift to $9.99 monthly for 30,000 monthly page views. You can have multiple sites per account. If you're unsure, you can run both simultaneously to see how they stack up. It wont' slow down your site as they're JavaScript code is loaded async.

11) This probably won't tell you much, but you may want to install the Theme Check plugin to see if there's anything dysfunctional going on with your site: [wordpress.org...]

12) Are you monitoring your site's uptime? Uptime monitoring is part of Clicky, but there are also free resources that provide this. Not just notifying you of ping failures, but actually downloading your page to test for text that should appear for a successfully rendered page, such as the copyright text in your page footer.

13) Is your site mobile responsive? If so, are you hiding your menu behind a toggle hamburger/menu icon? Do you have the word "Menu" as part of your icon? Here's a site discussing testing of the navigation icon:
[exisweb.net...]

By the way, I've tried multiple variations of the mobile icon and found that page views increased and bounce rates decreased when I opted to not use any toggle mobile menu and just showed the full site menu as a stacked menu below the logo. If you need to squeeze as much interactivity out of your visitors, then you may just want to show the entire menu, screen real estate waste be damned.

14) Are you sure you have no canonical issues? For example, if you use the Yoast SEO plugin, have you confirmed that the following tags have the correct protocol on http and https pages: canonical, prev_rel_link, next_rel_link, json_ld_output? For instance, if your page is on http, but someone is visiting https, your canonical may point to the https URL instead of the http URL. This can be especially problematic if you have a caching plugin generating these pages.

15) Are you sure you have no mixed content errors on your https pages?

16) Do you have visible ways for visitors to contact you? Email, phone, at the top of your pages? You may even want to consider getting an 800/888 number through a service such as [virtualpbx.com...] that can forward calls to another number, such as your cell or a pre-paid phone. This may instill trust from a user perspective.

17) Is your registration/cart/checkout page on SSL? If not, can you change it? If I were to register on a site, the following must absolutely be in place:

1) SSL-enabled page
2) No mixed content warnings/errors
3) Prefer a blurb of text reassuring me I won't be spammed

18) Do you have a mechanism in place to make sure your sign up form is actually functioning properly? For instance, if your form is on a HTTPS page, are you sure you're also posting to a HTTPS page? You may want to consider logging errors or form submissions via AJAX.

19) Have you ever considered changing hosts or moving to a VPS/dedicated server?

[JesterMagic]

I have experienced unusually low conversions since the beginning of April. Traffic was pretty good in April until about the 30th when we lost about 20% due to the Phantom update. (We gained about 10% during the mobile update so it is a net loss of about 10%).

[samwest]

@dannyboy - wow thanks! that's a pretty comprehensive check list, I'll bet many others could benefit from it too. I've certainly covered most of that list except the XML-RPC stuff. I just moved to the site to a host with SSD drives and that along speeded the site up by a LOT! Gonna go over the whole list again for the new hosting account. Thanks again! (BTW - yes, I've always been on a dedicated & SSL box until today...sales are so bad I can't afford the $220.mo fees, but went to a shared SSD account with IP and SSL)

@Jester - Thanks for the observation, there is a commonality across the board. Especially ecoms. MFA's probably notice no difference in conversion or even see an increase due to mobile accidental clicks.

Tomorrow is another day.

[Martin Ice Web]

@Samwest,

since begining of April we see the worst converting traffic we have erver seen. We saw two! Phantom updates in this time. Each of them caught some more of the buyers.
Traffic is, like yours , steady like a rock. My niche is flodded with big brands, amazon + ebay. Even if they donīt have the item you search for, they outrank all other sites that match the query 100%. It 100% brands biased. The funny thing about it, is that GSA are laser sharp on target.
I see this pattern on realy all ecoms if have the opportunity to look at. And all of the do not have any issues with checkout or ajax....
Finaly google found a solid way to give non converting traffic to smaller ecoms while routing all buyers through their ad system.

Search brings up foreign sites that might be connected to the search, mostly is because one keyword in query string is language dependend. This seems a signal for google to completly disregard the country the search comes from.


ecom, germany

[samwest]

@MIW - I wouldn't say traffic is exactly "steady like a rock" it's actually going down daily like a large tank with a very small leak. I remember just a year or so ago when search queries were in the 1000 range, now they are hovering just above the 100 range. It's mainly a loss of long tail terms. This definitely seems to be hitting ecoms the hardest. Just downgraded my hosting services last night from dedicated to shared for the first time in a decade. It's like trying to fit 10 pounds of "stuff" into a 1 pound bag. The hosting company wasn't happy, Oh well, trickle down economics.

[Panthro]

Tons of conversions today for a particular client site. May just be that time of year.

[tibiritabara]

Hi there,

I'm detecting google dance in Google.es Several landing pages that are failing in SERP are "doorway pages". <snip>

I think can be interest if you see same kind of landing pages affected in SERP. It would be a new Panda update that is including doorway pages anti-spam.

cheers

[edited by: Robert_Charlton at 8:14 am (utc) on May 19, 2015]
[edit reason] Removed link to blog post, per forum Charter. [/edit]

[Nutterum]

Google.de is broken for sure. Since first of May I have had increase in Sessions, lower bounce rate and higher average time, yet 1 conversion. Not to mention the keywords positioning is going nuts, jumping from first to 5th page and back every day or so. The UK has been hit but luckily the conversions are OK there and the website just got way more popular in India. Google can deny there is an update all they want but see huge differences in sessions and visitor behaviour in many countries, due to the nature of the service I do work on.

Also @dannyboy - excellent checklist, a one I will certainly put to good use!

[samwest]

I'm curious if Google is taking into account this new "page sitter" behavior. It's not actually people attentively sitting on the same page, it's them opening a page, then leaving the browser or app open as they move to a different task, like answering a call, a FB message from their BFF or satisfying a sudden urge. (I hope they don't take their phone to the crapper) Theoretically you'd think this would decrease your bounce rate, but maybe they've already factored this in and consider a long page sit a bounce. Who knows.

[Spiekerooger]

@Nutterum:

Google does not deny that there was an update - they even confirm that there was one to their core algo (see post above about SMX Sydney) they just decline to talk about it ;)

@Samwest: about page sitters vs. bounce rate: in google analytics this could be easily done by the page visibility api in javascript. But I'm not sure at all about bounce rate etc. as a ranking signal. If would have to decide about algorithmic twists for result relevance, I'd rather lean towards search history/customer journey. E.g.: 1. search for widgets, than blue widgets than brand blue widgets than brand/navigational search to brand site.

For example shortly before panda rollouts sites that could be panda candidates see rankings and traffic rising for a short term spike - believed by many to be a testing scenario. Now google could easily check if this produces more navigational searches for the brand / domain : if not, panda is justified and down you go.

[RedBar]

Google.de is broken for sure.

You ought to try .co.uk for some of my searches, I'm convinced I've been moved across the Atlantic to the USA!