| 6:52 am on Jul 31, 2008 (gmt 0)|
I'd sure like to hear more observations, Whitey.
One of the challenges here, I think, is that some people may not be watching traffic by keyword and especially not by time of day -- but only total traffic. If the Google yo-yo starts in on one query term but not others, all they may notice is total traffic dipping. Or it may even be improving, if their previous position was not first page and now they get position 4 for some of the time every day.
As you mentioned, there are also ranking yo-yos or cycles that go on without ever getting a first page position. I'm not sure if that's the exact same thing, or just something similar.
I mentioned that the rankings might change by a set time period or a set amount of traffic. I recently thought of a third possibility - a set number of impressions, whether their clicked on or not. A couple of my connections in the SEO world have mentioned that this wouldn't make sense. Well, that may be true but something certainly is happening. The -950 didn't "make sense" to us, but it happens.
Then there's the question of what starts the yo-yo. My current operating theory is that it can be triggered by a sudden but significant change (up or down) in the ranking before filters are applied. All of a sudden ranking goes way up or way down - well, that may be something Google wants to test rather than just assign the new position in a stable way. I know of one case where a strong site was actively working to optimize for a relavant phrase where they never ranked well before. A daily yo-yo effect kicked in between #4 and #14 and it has been in place for weeks!
But there's still so much guesswork involved, because I haven't collected a large data set so far -- and even when I have an example I usually don't know the site's history.
So yes, please, any related observations would be most welcome.
Unfortunately it's a tough measure.
| 7:10 am on Jul 31, 2008 (gmt 0)|
I'd give you the answer, but I'd get 20 responses saying I'm wrong :P
This has been happening for a while, so if you want to dig thru some of my old thread responses for a better answer...
Forget the reasons why for now. It doesn't matter.
Your primary goal isn't to reverse-engineer Goog, it's to rank for your chosen keywords.
Reverse-engineering - knowing the "whys" are only secondary steps to an answer WHEN needed.
That said, forget about what Goog is "testing". It makes the solution more confusing than need be.
Yes, it's a "filter" and Goog is "testing" something but it's entirely within your control.
(No, it's not impressions or analytics)
You're on the threshold of "some" filter. Usually authority/links, but not necessarily.
Get better & more authoritative links and your rankings will stabilize at the top.
if it isn't links, then figure out what threshold filter you're on the cusp of (usually on-page OOP), and add to or get rid of it
| 7:20 am on Jul 31, 2008 (gmt 0)|
|I'd give you the answer, but I'd have 20 responses saying I'm wrong :P |
Hey, that's OK - it happens to me all the time. I'm still willing to float ideas and get them knocked on, just for the chance to have some peer review and input from people who are watching other niches than those I am intimately familiar with.
|You're on the threshold of "some" filter. Usually authority/links, but not necessarily... if it isn't links, then figure out what threshold filter you're on the cusp of |
That makes sense, in the abstract, but finding what the threshold might be can be quite frustrating. I note again that the yo-yo can hit a url that was previously well-ranked, but it can also hit a url that never ranked well before. All of a sudden Google awards that url its first page 1 ranking for that query, at least part of the day.
| 7:26 am on Jul 31, 2008 (gmt 0)|
|I note again that the yo-yo can hit a url that was previously well-ranked, but it can also hit a url that never ranked well before |
"450 tweaks last year..."
Link-weighting gets changed, tested, and tweaked constantly, hence the jumping around at random times. If you're a careful datacenter watcher - hi reseller - you'll spot the trends and times of day they are testing with these urls. (I like to keep one of these pages yo-yo-ing to spot macro-changes to the algo)
In 95% of the cases I've had, authority links solved the problem. The other 5% were keyword density issues.
| 12:40 pm on Jul 31, 2008 (gmt 0)|
Tedster - Do you think this could have something to do with the thread I was reading the other about Google indexing some odd-billion URLs now several times a day?
I concur with whitenight about the 95% / 5% statement. That has always seemed to be the problem with most of the sites I track.
L8 - Saxman
| 1:27 pm on Jul 31, 2008 (gmt 0)|
|In 95% of the cases I've had, authority links solved the problem |
What types of cases were you observing that were in this cycle ?
| 1:51 pm on Jul 31, 2008 (gmt 0)|
You sometimes get higher than normal placement with this yo-yo effect? That doesn't sound like a penalty or even a probationary sort of measure. Maybe this is just an example of the expanding role of user data on the SERPs.
| 2:35 pm on Jul 31, 2008 (gmt 0)|
I've been tracking rankings on a single word keyword, ever since I found myself inadvertently ranked at #3 for it - I always figured it was WAY too difficult for my location-specific site, so I never went after it. Twice a day I enter the current ranking for it in an Excel spreadsheet - trying to keep to the general same time of day, but sometimes I'm busy or forget. I make sure I'm not logged in. I've been tracking it since May 20. So far I've been as high as #3 and as low as #319. I've been at #57 pretty steadily for the last four days. I dunno what I'm going to do with this information, but it's kind of interesting to watch - like a roller coaster.
My normal rankings are fine; have not really moved for several years, #1 or close to it for all locations in my niche. I also have a couple very high rankings for two word keywords that are not location specific, and they seem pretty steady. It's just the one word keyword that is bouncing around like a frog on a hot plate.
(Also, I'm not making any changes to the site at the moment, except for adding an occasional new event - for the most part its peak season is over for the year)
| 3:07 pm on Jul 31, 2008 (gmt 0)|
I've never been one for tracking key phrases in the SEs that much. I mostly use my logs, figuring that if I'm getting a bunch of traffic to a page then the ranking is okay. But since the yo-yo affect has been talked about so much lately I've been lightly tracking five phrases. And yes, they all yo-yo up, down and sometimes even appear to go sideways. But when I check my logs the traffic to these pages is just where I expect it to be.
There are a bunch of things G has done that can contribute to the yo-yo effect. The More transparency in customized search results [googleblog.blogspot.com] post from yesterday mentions things we knew were happening, but goes into a bit more detail. The "recent searches" factor is one I think might contribute to the yo-yo:
|we keep the most recent query on your browser for a limited time. After that, the information is removed from your browser and disappears immediately if you close your browser. |
So, we know this affects our searches (and we also know we search differently than other users), but we don't know exactly what the impact is. How does the algo treat repeated searches? Similar searches? What happens if you search for a place to eat tonight, then follow that with your money term?
I'm going to play around with this over the next couple of days and see if there's anything to be seen here.
| 3:57 pm on Jul 31, 2008 (gmt 0)|
|Then there's the question of what starts the yo-yo. My current operating theory is that it can be triggered by a sudden but significant change (up or down) in the ranking before filters are applied. All of a sudden ranking goes way up or way down - well, that may be something Google wants to test rather than just assign the new position in a stable way. |
I absolutely agree. I have a small site that is full of unique content with some links coming in. It's more comprehensive and better organized than most of other sites with the same information. About a couple of months ago it moved from the third page to the first page (I think this was just a matter of link and site maturity), but has since then been bouncing around between #4 (usually) and bottom of the second page.
The same happened with a couple of other small sites, but they ended up settling on the second page.
| 3:59 pm on Jul 31, 2008 (gmt 0)|
What jimbeetle reports is pretty common. It's part of the reason we've heard so little discussed about the Google Yo Yo, even though it's been happening for months. It's a rare webmaster that watches closely enough even to spot the phenomenon, except in certain special cases. And then again, only a few urls are having the experience in the first place - but it is seen on some very competitive phrases.
At any rate, our old mental model of Google ranking being a sort of contest with a semi-stable list of first place to 1,000th place winners is not holding up very well at all.
| 4:10 pm on Jul 31, 2008 (gmt 0)|
|our old mental model of Google ranking being a sort of contest with a semi-stable list of first place to 1,000th place winners is not holding up very well at all |
Yeah, with all the customization, personalization, localization and such we won't ever again see what a searcher sees.
The one thing I find a bit curious is how G is going to handle, for want of better words, the "Alta Vista effect." Remember back when AV was tired of us gaming its SERPs and started to jumble them up? That was one of the last nails as users couldn't again find what they found before. It was very frustrating for users and, as Google was just launching, they abandoned AV in droves.
| 4:36 pm on Jul 31, 2008 (gmt 0)|
One of my sites (which, by the way, is only three months old) has been bouncing from 3 to 10 to 3 to 10. Lately I've been steadily ranking #6 for a while, but today I'm back to #3...
My term is a not-so-often-searched one (~1k searches / month). Are people experiencing this effect also on 'bigger' terms?
| 4:49 pm on Jul 31, 2008 (gmt 0)|
Experiencing this since a site redesign. After a few weeks of YoYos happening several times in a single day, there is actually now a steady decline as far as averages go. My past experiences with Yo Yo have been the same. Yo Yo for a period of time, followed by a distinct drop overall.
I don't think this is an indication that if someone Yo Yos they're going downhill, but merely that the problems that originally triggered my specific Yo Yo case eventually bloomed into something uglier the longer they went unaddressed.
| 4:56 pm on Jul 31, 2008 (gmt 0)|
|Are people experiencing this effect also on 'bigger' terms? |
One example I know of is a query with 57 million results and 128 Adwords advertisers. The affected site is a "big brand" household name in the US - so it's not just the little guys who are seeing this.
| 5:15 pm on Jul 31, 2008 (gmt 0)|
|What types of cases were you observing that were in this cycle ? |
could you clarify what this means? "cases"?
| 5:25 pm on Jul 31, 2008 (gmt 0)|
In some of the bits I'm reading parts of this sound like the old question, "Why are my pages moving in and out of the supplemental index?"
| 5:26 pm on Jul 31, 2008 (gmt 0)|
I'm a little guy (so to speak), but I'm following a big guy term with 57 million results also (how weird is that) and anywhere from 2 to 12 ads depending on when you hit them (it is a seasonal thing)
| 6:41 pm on Jul 31, 2008 (gmt 0)|
|Why are my pages moving in and out of the supplemental index?" |
For those who like to debate the "reasons why", I look to this, as well as, implementation of Universal Search for reasons why it happens.
(Also study Position #6 threads for clues)
It's sometimes just one, sometimes both, sometimes neither.
Like I said before, too confusing for all but the most technical SEOs and I don't like over-analyzing/discussing situations with relatively easy solutions.
| 10:18 pm on Jul 31, 2008 (gmt 0)|
|could you clarify what this means? "cases"? |
I guess you meant "situations" . But what I'm asking is , what were those situations - if you could open up on this it would likely assist folks understanding and relate it to their own observations in nailing the various nuances.
I'm seeing an example of the Yo Yo effect an entire site, capping it's rankings to minus "penalty" levels - but fluctuating from #4 to #7 when it should rank #1 [ and did ] .
So at least for this example, I see the Yo Yo effect applied in a penalty situation. However, it's as though Google hasn't made up it's mind of what level of penalty is to be applied.
[edited by: Whitey at 10:29 pm (utc) on July 31, 2008]
| 10:41 pm on Jul 31, 2008 (gmt 0)|
you mean for FREE?! lol
ok, without giving away too many secrets or the lessons learned from studying the SERPS diligently (instead of hearing me or anyone else give advice that may or may not apply to their site)....
A current page i'm particularly proud of, was resting on a site that dominated for every other page ...except this one url.
It would "yo-yo" between #10 and 100+ or 400+.
So I knew it "should" be an "authority" page.
Long story short, it had plenty of links but not from a particularly diverse or authoritative sites. Threw some truly authoritative links at it and it stopped yo-yo-ing from nowhere (400+) to #8 and now rests steadily at #3 behind 2 Fortune 500 juggernauts.
(2-word "main" money keyword + the hundreds of sub-keywords most people would pay some "guru" 5 figures to rank for)
Like I said before, a variety of pages on various sites were solved with similar techniques.
The "fine-tuning" involves the individual page and individual webmaster.
| 11:38 pm on Jul 31, 2008 (gmt 0)|
OK - so let me clarify what i think is being seen out there . We have the Yo Yo effect on :
- individual page urls within a site , effecting movement within a small SERP range
- site wide minus penalty applied sites , moving within page ranges
- rotating results in say the top three or four results
any other's ?
| 12:36 am on Aug 1, 2008 (gmt 0)|
K, silly question, but can the different data centers have a different "build" of google's algorithm?
I can't help but think about how changes are pushed out to various DC's and if so, maybe just adding some more authoritative IBLs will help the different builds settle on a more appropriate ranking...
Almost like there is a sensitive filter that gets passed over with the addition of some authoritative IBL's
Would like to reseller and g1smd pop in on this as well...
| 1:01 am on Aug 1, 2008 (gmt 0)|
I am bouncing all over for about 2 weeks now. For 3 competitive terms I monitor our home page is anywhere from Page3 to #3. Sometimes I am #3, checking a few hours later - at third page. I am using a few "check your google ranking" websites and it's the same dance (just to make sure it's not customization).
I hope it means I am on the list for "promotion" :)
| 1:11 am on Aug 1, 2008 (gmt 0)|
I was called away when this thread was only half a page long; finally caught up with it.
I haven't exactly seen the effects discussed above. What I have seen is for a site:domain.com search, that different Google IPs return a very different set of results... so I want to know from those that see the cycling if this cycling is seen at a constant Google IP, or whether it is a result of google.com connecting you to a different datacentre at a different time of day.
I'll also note that I do see an effect where if I make a search I get one set of results, and if I immediately do that search again I get a very different set of results. If I search again, I never see the initial SERP again, unless I don't make the search again until an hour or more later. If I see it again later, then again it will only be in the first search of that session for that keyword.
| 1:21 am on Aug 1, 2008 (gmt 0)|
We have seen the same yo-yo for a highly competitive term. Usually from position 17-19 to 4 and back down. This too has been going on for weeks, starting slowly even a few months ago, but definitely having ramped up since additional efforts were taken to make the home page of the domain more focused towards the target phrase.
We are also watching a synonym (also) two-word phrase with the same root keyword and a semantically equivalent modifier. For this one, we see very little yo-yo, but a more deliberate test lasting longer periods of time, it appears.
The good news I guess if you look at Ted's theory is that we could be close to gaining permanent ground here.
One thing we noticed was that when we shifted into the four spot and then left, a news box universal result usually showed in that spot directly after. Did not notice if it was there before but it was not there when we had the 4 spot. In fact there is a news box there now. Wonder if Google is trying to gauge whether people will click on news if they display it for this term. If so, how do we get to be the randomly pulled form the second page result? Perhaps since the domain is being given more authority based on the latest crawl, and that makes it a good dual purpose test.
It would be great if next time it is at 4, we announce it and get everyone to click on it to see if that “makes it worthy” of staying. Anyone buy that one of the metrics to gauge the worthiness of a result is by CTR?
As a gauge for competition: 32 million and approx 180 ppc bids (really)(and speaking of sponsored listings how long have they been displaying those in 3 rows of four AdWords box-style listings? There are 12 per page...pretty cool)
We are looking very closely at this phenomenon and look forward to continuing to share so we can maybe figure this one out together. Sounds in fact like a great question for “Meet the Engineers” at the G dance…
| 1:24 am on Aug 1, 2008 (gmt 0)|
The cycling effects I see happen on the same IP address.
That doesn't rule out the possibility that a given IP address gets routed to a different data center or server cluster - in fact, that's what I suspect happens. Calling an IP Address a "data center" is probably rather inaccurate, technically.
| 1:47 am on Aug 1, 2008 (gmt 0)|
Good thoughts on the IP/database question. Usually our strategist for this account checks across a number of datacenters and finds consistent or at least majority results showing this pattern. Maybe he will join us and confirm, but I am pretty sure he checks across Google IPs.
I tend to see the same results when looking from our T1 and on wireless. I also observe similar trends from my home wireless. This would probably be consistent with getting fed the same datacenter the majority of the time since everything is within 20 miles in Ohio.
| 1:55 am on Aug 1, 2008 (gmt 0)|
I think it's a HUGE mistake to look at this issue as a MY SITE being tested/tweaked instead of A GENERAL RANGE/PROFILE OF PAGES that the algo is affecting.
I can't emphasize this enough.
Which is why I was so adamant(and correct) about the #6 penalty and ESPECIALLY for those pages that are affected by the Universal Search testing (ie jumping around the top 12 or so)
Google engineers don't even realize what specific pages it's affecting (again, it isn't a "YOUR SITE" issue), it's simply that you're on the edge of whatever threshold/filter that's being tested on and off that causes the yo-yo-ing
Just get the links to rank #3 (or whatever) and above and be done with it.
[edited by: whitenight at 2:07 am (utc) on Aug. 1, 2008]
| This 207 message thread spans 7 pages: 207 (  2 3 4 5 6 7 ) > > |