What happened on September 17 on Google?

[fablau]

According to mozcast:

http://mozcast.com/

and to my own stats, Google had a pretty strong algorithm update on September 17. Personally I have experienced a drop of about 10% of traffic coming from Google on most of my main e-commerce site virtualsheetmusic.com.

Anyone know more about that update? Any ideas about what changed?

Thank you in advance for any thoughts!

Best,

Fab.

[RobertFisher]

Fabrizo,

I am intrigued whenever I see this question because it seems we notice only when we feel an effect. We handle multiple sites as an agency and I don't see any real "9/17" change across the board and no change that is noticeable for even a single site (I looked at five that I know are more likely to move).

With the mozcast, that is another area I find intriguing in that I respect those at Moz and their understanding of statistics and scientific method; I also scratch my head from time to time as to whether or not the given movement has any overall effect on "most" sites.

When you read the "About Mozcast," and they point out the numbers of algorithmic changes in a year, it is apparent that most won't have an appreciable effect on a given site. Unfortunately, for most of us change to any site we own or manage can have dire consequences so we have to always be vigilant.

I wish I could give you a better answer, good luck,

Robert

[Highland]

SER had a post a couple of days ago asking just that. I can't say my traffic suffered a lot (in fact, one site seems to have had a small bump in organic traffic) but given than it only got to 86 (100+ seems to indicate major shifts) I'd say it was likely a localized set that got hit. Probably a Panda shift (just a guess, tho).

[AdamWormann]

Earlier today, Google announced an algorithm change that should affect about 90% of search queries.  They said this has rolled out over the past month.  When more details come out and some people do some more testing, this may have something to do with it.

[Dr-Pete]

Yeah, Robert's right - with 500+ updates, the task of figuring out which spikes really mean something is very difficult. The pattern of 9/17 on MozCast looks more like a traditional update, with a relatively quiet period around a one-day spike, but I don't have a lot more detail on that particular day.

The update Adam mentioned ("Hummingbird") apparently happened "about a month ago", but seems to be tied to semantic search, Knowledge Graph, etc. Google's statements are pretty vague. It's more likely that is related to the 8/20-21 spike spotted by multiple tools and webmasters than the 9/17 spike.

Sorry I can't give you more information. I've seen very little chatter or reports about 9/17, other than what we saw in the tracking data.

[fablau]

Thank you guys for your replies and insights. It is my understanding that MozCast draws its graph based on the number of sites affected by a Google update... is that correct? If so, I deduce that people (or the algorithm) beyond MozCast knows which sites and/or how many sites have been affected by a particular update. If that's the case, and I don't see any different scenario, I assume that we can potentially understand if those affected sites have something in common (are mostly e-commerce websites? News? etc...). That would help us to understand the nature of any update, most of all the major ones since we would have more data to crunch.

Am I wrong with my brainstorming here? I am eager to know your thoughts an this.

Thank you again.

[Highland]

It's possible all these shifts are due to Google Hummingbird, which one person at Google called "the largest rewrite since 2001." This is the month-long rollout they've been talking about.

Still, Hummingbird is more about usability than SEO signals. The biggest shift is in "conversational search" (i.e. "How often has Rand Fishkin shaved his beard off?"). Google is now focused on returning more relevant results to those kinds fo searches. That would explain why temps never spiked. It doesn't seems to have affected generic search terms as much, if at all.

[Dr-Pete]

It's a bit tricky. MozCast (and other trackers like it, to the best of my knowledge) basically look at how rankings change over time. For MozCast, we track two fixed sets (1K and 10K) of keywords every 24 hours, and then measure how the URLs in the top 10 shift. This is tricky for many reasons:

(1) There are a ton of ways to measure this "flux", all of them valid in different ways.

(2) Baseline flux is very high. I estimate that as many as 80% of queries change daily, to some degree. Google is much more dynamic and real-time than most SEOs think.

(3) "Baseline" flux varies wildly across keywords, based on factors like QDF. I wrote a post about just how extreme this can be (http://moz.com/blog/a-week-in-the-life-of-3-keywords).

Ultimately, we try to gauge to an average, and then look for extreme variations, but the noise-to-signal is extremely high. The reality is that SERPs are change all of the time, not just based on the algorithm, but on changes to sites. Google also makes more than 500 changes per year, so even "algorithm update" is a tough term to define. We're looking for the big ones.

It's important to note, too, that all of the current flux tools are focused solely on organic results and movement of those results (as are most SEOs). We're not looking at how verticals come and go, Knowledge Graph entities, etc. We're actually working on some tools to track these entities more closely. "Hummingbird" is, IMO, going to power these entities and expand them, possibly for months to come.

[fablau]

Thank you guys for your replies and information.

Peter: I understand what you mean and I do understand  why it is not possible for MozCast to know which verticals are affected by a particular Google update. What you wrote makes perfectly sense.

Highland: you may be right about the Google Hummingbird, but I see that update was released about 1 month ago whereas I begun having a drop in traffic since September 17th (13 days ago). But we can't exclude it either since looks like my long tail keywords have been mostly affected (see below)...

It is worth noting the following events around the date the drop begun:

1. On September 15 our hosting provider had a major power outage which put our site offline for about 5 hours. I don't think Google cares too much for this since 5 hours it is not a huge down time and never happened for at least the past 3 years, but this happened just 2 days before the drop begun.

2. On September 17 (the day the traffic drop begun), we updated our website page rendering engine to improve our page speed of about 20% (this should be a good thing right?)

Also, I have analyzed the traffic coming to our website from Google and looks like that the most affected section has been our product pages which makes me think that long tail keywords have been mostly affected.

Any more clues?

Thank you again, I really appreciate your insights and thoughts on all this. And, please, if anyone has experienced a similar drop in traffic since September 17, please post it here!

[Whebb]

Had something similar happen to a client of ours. On Sept 17 they lost about 85-90% of their organic traffic among all search engines. I mentioned this in a post I added yesterday.

http://moz.com/community/q/loss-of-85-90-of-organic-traffic-within-the-last-2-weeks

Still trying to figure out exactly what happened, but am also curious to see if anyone else ran into similar issues.

[fablau]

Thank you jStrong, I have just posted something on that thread too. I am glad to know I am not alone! I hope we can figure out what happened and possibly tackle the problem.

Thanks!