|Can you infer useful data from Panda refresh NOT having an effect?|
| 4:30 am on Jan 25, 2013 (gmt 0)|
I have a site that has exhibited a few up and downs (mostly downs though) in the last couple years - started few months before and continued pretty much all through the Panda age (early 2011-current). However, one thing that stands out for me looking at the traffic stats - neither the ups nor the downs have ever coincided with any of Panda refreshes. The exact dates of Panda refreshes and several days before and after always look very unremarkable for this site.
The site has lost approx 70% of its pre-2011 traffic and yet Panda has never touched it (or so it seems unless it bites in-between refreshes, too). So, what can I infer from that data - the site's quality is good (if Panda was ONLY about quality, however it's defined) yet some other parameter is bringing it down? What would you infer from Panda refreshes having no direct effect on traffic to a site and yet the overall trajectory is downward?
| 6:12 pm on Jan 25, 2013 (gmt 0)|
Okay, so you can put Panda aside because it appears to be a non-factor.
What you have to do now is instead of saying that "the overall trajectory is downward" is get under the hood and get your hands a bit dirty by trying to identify the *exact* hows and whys of the traffic drop.
Was it certain key phrases? Certain site sections or pages? How has overall search volume changed? What changes have there been in the competitive landscape? Has the move to mobile affected traffic?
There are many other questions that might be asked. Start by getting answers to some and you might get pointed in the right direction.
| 6:37 pm on Jan 25, 2013 (gmt 0)|
@jimbeetle: Thank you for your input, Jim. Yes, of course I am looking for all possible factors but the idea of inferring some data from the Panda updates not having an effect would be to prioritize my actions. For example, if I knew that this means the content is considered a-OK, I would concentrate on getting good links instead of writing new content. Not too many links were gotten since 2011 and natural link accretion is low due to low overall traffic.
Does this mean that I can skip looking at the user experience factors and concentrate on adding features instead (for example, site search may need some work as it's getting antiquated)? And so on. I do realize that ultimately all of that needs to get done but if I can prioritize the tasks, I can do more good in less time, hence this question: I have this one data point, what can I do with it?
As far as the factors that I can see leading to the loss of most traffic: it's definitely the hollowing-out of the long tail. All the top KWs are still there, but tens of thousands of those that used to bring 10-20 visits a month or even less are no longer there. Likewise, all the top traffic pages are still top pages (and number of visits is roughly the same) but those pages that were visited just a few times a month now get no traffic at all. Given that there were thousands of those, the effect on overall traffic amount is considerable.
| 9:21 pm on Jan 25, 2013 (gmt 0)|
Well, you have more than one data point. such as these you sussed out:
|All the top KWs are still there, but tens of thousands of those that used to bring 10-20 visits a month or even less are no longer there. Likewise, all the top traffic pages are still top pages (and number of visits is roughly the same) but those pages that were visited just a few times a month now get no traffic at all. Given that there were thousands of those, the effect on overall traffic amount is considerable. |
Now that's interesting, an obviously healthy site that is keeping its head traffic but is losing long tail traffic.
I couldn't hazard a guess as to what on-site factors would cause such a phenomenon. Have you happened to see any other reports of this happening?
I will hazard one SWAG, albeit in very small type: Is it possible that G's search suggestions are starting to funnel queries away from the site's historical long tails? But that should still be something seen by other folks.
| 11:13 pm on Jan 25, 2013 (gmt 0)|
Are you still adding new content to the site on a regular basis? Over the years, I've had a couple sites that I ended up neglecting over time, and watched them slowly drop into the nether.
If nothing new ever happens on the domain, it seems, Google slowly forgets about you.
| 5:41 am on Jan 26, 2013 (gmt 0)|
Think quality not quantity.
| 7:11 am on Jan 26, 2013 (gmt 0)|
End of Oct: I got rid of roughly 2000 malformed urls in that Google now sees a true 404 and not a redirect to the proper url - this had absolutely zero effect on Google traffic after this latest Panda update.
GWT has reduced the number of "not chosen" urls to nearly nothing as a result, which was my goal, but it did nothing for traffic, yet.
| 5:20 pm on Jan 26, 2013 (gmt 0)|
I've seen this phenomenon before on a huge online education site. Their situation started in the late 2010 timeframe and slowly, progressively got worse over time. After digging deeply into their analytics, what I found was that their degree pages were previously ranking for tons of "loosely relevant" phrases which had been driving a lot of (non-converting) traffic to their pages.
For example, a Masters in Computer Science degree page might have been previously getting traffic for loosely related keywords like "computer programmer", "computer jobs", "data structures", etc. While these terms are loosely related, they rarely converted for this client. Much of the traffic lost in this situation was for phrases such as these.
Personally, I think this is often times the result of Google changing how they detect user intent. As this has changed drastically over the last couple of years, in a lot of cases they have stopped sending traffic to pages they probably never should have been sending them to in the first place.
I would say forget about Panda and any corelation you might attempt to draw from "not" having been hit by it. It could be a situation similar to the one I described above or it could be that whatever led to your issues is preventing Panda from even coming into play. I would be looking at a lot of other things like those that jimbeetle suggested and more.
| 6:31 pm on Jan 26, 2013 (gmt 0)|
@ZydoSEO: thanks for the great input. I am seeing similar thing happening here as well. I am not sure how I can check specifically for the effect of search suggestions jimbeetle talks about but I am seeing a (slight as it is) increase in conversions from the lowered amount of traffic.
It's hard to complain about better quality of traffic from Google given that it's always been pretty bad compared to Bing/Y! but I have to admit: if the site is being affected by changes in the detection of user intent, that makes the way forward even less clear to me. If it were about quality, I could have tried to add content, re-write some pages by adding more on the topic covered etc. But I am not sure what can be done about user intent?
The entire concept of user intent is a bit muddy to me, too. The only clear cut "intent identifier" that I can see would the to add the word "buy" to the original query. I've never ranked for anything of that nature anyhow - this is not an ecom site and nothing is sold here, so looks like it was properly intended in the first place. Have they started to parse intent so finely now that more than two intents - "research" and "buy" - exist?
About the "loosely relevant" matches - I can't quite put a finger on that, either. It's hard to know for sure because of the amount of KWs to sift through. But in the sample I took, I see it both ways - some strong matches (exact phrase - could there be a stronger match?) are gone and some loose ones remain. I don't know exactly what signals of relevancy Google uses and perhaps exact phrase match is too crude a relevancy signal, but in the top phrases that remain, at least one is a whopping 7-word long phrase, and a couple more are 6-word ones, neither with exact phrase matches on the page, which you would think be the first to go if either loose relevance (no match) or search suggestions (too long) would have strong effect.
ZydoSEO, what were your recommendations to the site you're talking about?
| 9:11 pm on Jan 26, 2013 (gmt 0)|
Um I was going to start a new thread on
as it is important to us and probably others.
|Google changing how they detect user intent |
We have isolated our 70-80% drop as it times with Panda, but that may be a coincidence. What we have isolated is searches for "[placename] widget information" does not bring up (even if you ignore auto-suggest dropdowns ) sites with information pages it brings up "[placename] widget information OFFICE" so we get all the telephone directories etc listed first, unless there is no OFFICE at that location in which case we are always on page 1.
Google has assumed , somewhere along the way, that the user needs the office contact details, when in fact, they don't.
So @1script may be suffering from this, it's worthwhile looking back over your logs and try to identify if this sort of change took place.
Sorry about the obscure search terms - WebmasterWorld forum rules !
edited to be polite and include OP name !
| 10:25 pm on Jan 26, 2013 (gmt 0)|
@johnhh : did I get it right that you're now seeing longer-tail KWs in your logs because Google is adding some kind of a qualifier ("office") to the usual queries assuming that's what people are looking for? It is an interesting observation, I never looked at the stats that way - take a good KW from the past and see if some words have been added to it. Perhaps the added words can give me some kind of a clue.
Or did you mean that Google adds the qualifier but your pages cannot rank with the qualifier and therefore you don't see anything in the logs because there are just no visits? This would be much more difficult to troubleshoot and local and personalized searches will get in the way.
Still, if you've identified a certain intent that Google is now steering the searchers toward (whether it's their real intent or not) - what would you do, add new pages addressing this particular angle or edit existing pages to add perhaps a paragraph or two? Editing existing pages is easier logistically but I'm always afraid to touch something that already ranks for *something*
| 12:57 pm on Jan 27, 2013 (gmt 0)|
@1script - a bit of both - we see longer tail queries as the qualifier is not appropriate, think "blue wigets around [zip code]"
correct ! unless the [placename] has no OFFICE, in which case we get on page 1
|Google adds the qualifier but your pages cannot rank with the qualifier and therefore you don't see anything in the logs because there are just no visits |
We added specific pages to address this problem, but as its only a small % of the site we are down on page 3 most of the time, where there is an OFFICE at the [placename]. One of my competitors caught this early and set up a new site with an exact match domain name, think [countryname]-widget-information-OFFICES.
We think this is why we can't get back our traffic for those search terms, and of course we can't do anything about it.. except chase more long tail
| 4:32 pm on Jan 28, 2013 (gmt 0)|
I'm seeing what johnhh is talking about, only sometimes it's backwards. And I've talked to other users who are seeing it, too.
The way I've seen it "backwards" is like this: I can rank for "widget thing" but not "widget thing tutorial." The more specific keyphrase may bring less, but more focused, traffic. Whatever Google is trying to do, I think we need to understand it better, and if even three or four of us are seeing it, maybe it does merit its own thread so we can really get into it.