Rankings for one of my important Keyword's -
21 days ago - 3rd place
14 days ago - 10th place
10 days ago - 15th place
7 days ago - 20th place
5 days ago - 17th place
3 days ago - 15th place
today - 12th place
so...things seem to be slightly improving here.
Just spent an hour looking at analytics graphs for my three sites, the two important ones hit by Panda and the unimportant one that got a 10% boost from Panda. First, the unimportant site definitely got another boost in the past couple days and is at an all time high for traffic, but that's only around 1,000 unique visitors a day. I know I have made no changes on that site since last year because I lost the FTP password at some point and haven't bothered generating a new one.
On the two sites that got killed, yesterday was a little better on the whole, but nothing outside of standard variations. The exceptions were a couple of the highest ranked pages which showed double digit percentage jumps. I didn't make any changes on those pages, but they are the ones I concentrated on for DMCA complaints. But it's just one day and they are nowhere near back to normal. If it holds up another week, I'll report.
On the JohnMu advice, I think JohnMu is reading backwards into what they believe their Panda algo does, improve quality. So in his logic world, it makes sense to tell webmasters that the way to regain traffic lost to Panda is to eliminate low quality content.
The problem is that Panda doesn't do what they think, and quality is an undefined term for them. I would define a quality web page as "publication quality", something you would see in a magazine or a book, something that gets stolen by university students (and quoted by professors). That's not what they are talking about.
One of my hard hit pages has a dozen citations in Google Scholar, it's the original and essentially the sole source for some niche research, regularly updated. A half dozen sites that quote it, and link it (organic, without NOFOLLOWs) rank above it now for most related searches. I think it's because Google is giving love to shorter pages that link out a lot - big benefit for scrapers.
I think there might be a big difference between "quality content" and "searcher preferred content".
The "easy answer" is probably more quickly found in the "searcher preferred content" however low the actual quality of that answer/content is compared to content that is actually high quality.
That might account for the prevalence of sites like that massive _how thing, or similar sites.
How to measure "preference" for this exercise? Maybe CTR from the serps?
I dunno, it's just a thought.
Anyone care to guess the number of searchers looking for the "easy answer" compared to those looking for the highest quality answer?
I think one mistake I made years ago was making my sites too easy to use. This method gives the user what they are looking for, very quickly, with very few clicks. This reduces duration of time on my site. If those metrics (time spent on site, hitting the back button to return to Google, etc.) are being used, then I need to make my site more confusing, sending the user deeper and deeper, in circles. I will have to give in to the A.D.D. epidemic by making the user forget what they were searching for. That shouldn't be too difficult. After all, it works on me. I forget what I am searching for all the time. I'll play your game Google :)
@crobb305 ... Inception for websites.... coool.... now what was I doing here?
|I think one mistake I made years ago was making my sites too easy to use. This method gives the user what they are looking for, very quickly, with very few clicks. This reduces duration of time on my site. |
Most of my sites are set up like this, and I have (so far) not suffered from it.
I'm taking a different approach to recovery.
Why chase or fight with the algo? Google is saying shallow content this shallow content that, but nobody really knows what's up, or is recovering lost traffic.
Let's say I sold coffee beans, and the price of coffee tanks 50%.
Rather than fight the market that I have little control of, why not just find a way to sell more coffee beans. Open additional stores, invent a coffee bean powered car, whatever.
Is the algo broken and are scrapers outranking your site? Scrape your own site. Take that Google!
I joke and jest, and yes, it's easier said than done to scale anything, but hopefully that made some sense.
So what hints do the raw serps give about Panda? I'm not talking about the sites listed in the serps, just the serps themselves.
How many people who removed or modified pages or modified robots directives have taken note of how these alterations affect their site(s) in other SEs?
Modifying to get google's approval, whether or not your site is a good one, could easily kill your site in another SE such as Bing.
Working on the lines that people unhappy with google (as a company or as search results) are likely to try other engines, it's possible that playing google's game could suicide your site.
By all means improve a web site if it's necessary to your customers' well-being but I would have thought it better to kill google's access to a few pages whilst leaving the site much the same for other SEs - assuming the site is ranking well in those SEs.
|Modifying to get google's approval, whether or not your site is a good one, could easily kill your site in another SE such as Bing. |
I agree. With the exception of two useless articles on my site, I have refused to noindex or delete anything -- after all, they are ranking in Bing/Yahoo. I have, however, removed Adsense on most of my pages, so Google loses revenue from the 1,000+ visitors I still get from Bing/Yahoo. They shot themselves in the foot with their silly ad-to-content ratio crap.
Something I never thought of checking for my unimportant site, which is now up 30% since Panda (my eyes fell out when I finally compared to the past), is the Google only stats in Analytics. So clicked Google in the search engines section of Analytics and found that this site has a higher bounce rate from Google than my important sites, and a lower time on page. As in, MUCH lower. As in, less than a minute, with 20 seconds to spare.
So while we're worrying about time before people hit the back button, maybe we have it backwards. Maybe Panda wants people to hit the back button.
What a mess.
As to why the bounce rate and time on page is so low, the site is appreciably more technical than most consumers would be willing to read. It's actually a case where they might be better off with eHow for starters, and save my stuff for after the three easy solutions (like, "Is the thermostat turned off?", "Did you run out of oil?", and "Is there an open window?") don't get it done.
I put forth the following hypothesis to the masses:
An idea I had tonight was: perhaps measuring a site's multimedia ratio to text value rather than the ad to text suggested previously. High value sites tend to have image, video, and text... and I ponder if this may hold true here? Indeed my site that was killed had very little multimedia content.
I'm curious to hear people validate or debunk it... to see if there's any truth in it...
I am convinced that the Google algorithm as a whole, and Panda in particular, is not using the generic "bounce rate" as a ranking factor. As discussed here several times recently, that is a VERY noisy signal.
Google may watch what we've been calling the "click-back" rate - returning to the same SERP to make a different click. And even though Google (and other search engines) watch that factor, it cannot be a major factor, because it is still pretty noisy.
I keep reading about removing thin pages. Even though i never saw this writen officially by G myself, i take it this is mostly what we have to do. The problem is that 'thin pages' is not clearly defined. A page with lots of text and few images could be thin for someone looking for images. Someone doing research would find gallery pages thin...
I guess the first step would be to get a clear image about what is thin and what does 'quality' mean.
I belive removing pages, noindexing them and redirecting them based on a subjective, personal assumption can not have the expected result.
I think putting our observations together for defining 'thin' and 'quality' first, before taking any action, could get better results in the end.
Forget 'thin', think 'shallow'. In particular, think 'is this page largely unique and special?'
Don't let a target phrase be matched by many pages - organise your site so that any phrase, from short to long tail, is matched primarily by ideally just one indexed page. Noindex the others.
It's less about amount of content, more about amount of unique content; don't duplicate your content (as far as you can).
I think it was MadScientist who gave a good tip wrt duplicated content on your own site - if you must have it, separate it into its own page (and perhaps iframe it if it must appear on particular other pages); otherwise link to it.
So it's a matter of quality rather than quantity, and how well organised.
[edited by: bramley at 12:55 am (utc) on Apr 8, 2011]
I was wondering if you saw this on your more technical website, because i might have smth on that.
Does something new posted there rank poorly at first, but over time(by this meaning a few months), it starts getting better rankings?
To be clear, the website isn't more technical than my other websites, it's just that its subject is searched on by far more non-technical people who wouldn't know how (or want) to utilize the resources. Like anybody who owns a home vs HVAC professionals and serious home improvement types.
I haven't posted anything new there since last year, but it's typical for all three of my sites that new subject content takes months and years to catch on strong. That's just how organic linking works, takes time. It's true that all pages eventually find their level, whether it's a thousand visitors a day or twenty, but they all start in the same range.
I also believe that discussion about thin content, quality, etc, isn't the real issue. Google is using mysterious "as judged by humans who weren't searching for anything" signals, that's the one thing they said on the record about Panda. They didn't pick the signals they have promoted in accordance with a theory or search testing, they (Matt Cutts) merely said that it made sense in retrospect when they looked at the sites their human testers deemed "high quality".
The sites I see that have jumped above my two serious sites run the gamut from garbage scrapers to more or less equivalent quality. The quality ones were in the ballpark before Panda, the crappy ones are a post-Panda phenomena.
I believe they simply got it wrong and are drinking too much of their own Koolaid to admit it.
Very interesting theory on the Singluar VS Plural
Our site was hit pretty darn hard considering we are ecommerce.
Most of our subdirectories are in a plural form, which makes sense when viewing the site, but maybe not from the search point of view. If you're looking for a "blue widget" you arent necessarily going to search for it as a plural. Very interesting indeed.... I may do some experimenting on this.
|Very interesting theory on the Singluar VS Plural |
Bewenched, can you point me to the point in this thread where you saw a theory presented? I can't find it. I am seeing huge differences in rankings between singular/plural for some sites.
I have been analyzing some pages that rank on page 1 for a single word query (singular form), but rank on page 5 or higher for the plural form. I can't find any common characteristics that would point to a cause.
From what I see, pages that have the plural form in their title or body (eg: widgets) rank well for widget.
But I see a drop for pages with "widget" even when I search for "widget". The drop is even more if I search for "widgets".
Bewenched, I am seeing something that contradicts what you say.
ken_b..great point...because if some one is looking for high level contents/information/discussion/solution,but having no idea,how to find high level contents..they might be get stuck with low quality,shallow content quickly..just a thought..from humans psychological behavior..
G classifier can identify in near future that level of contents,and give a preference to those valued websites...
|If you're looking for a "blue widget" you arent necessarily going to search for it as a plural. |
There have been times when I ranked #1 for both singular and plural forms. I always had much, much more traffic from the plural. I think a widget dealer might go on TV to say "we have many widgets to choose from" not "we have a widget." The plural seems much more natural and common to me, and my stats from the days when I ranked the same for both seem to support that.
I am seeing more and more SERPS displaying page titles that are not from the <title> tag. It seems Google is using header tags in some cases. I have a site that this is happening on and I have just observed it happening with two competitor sites. It is only happening on select queries. For other queries Google is using the <title> tag like normal.
Has anyone else seen this lately?
Yes - see Google Changing Page Titles & Descriptions [webmasterworld.com] from February 18.
They've been doing it for several months and the practice began to accelerate in mid-February. Panda wasn't until a week later and I doubt that there's a relationship.
I don't understand the rush to add all those changes one after the other without making sure they're working well in the first place.
|If you have pages called blue widgets, red widgets, black widgets; none will rank well just for widgets; for this one needs a widgets page that links out to blue widgets etc. |
I've observed something similar, but I don't correlate it just with the Panda update. I saw perhaps one big change at the time of the "Scraper Update" [webmasterworld.com...] that I attributed in part to this kind of data organization, but this kind of organization as a ranking factor has been evolving for a while now. I think of it as one factor, not necessarily the only factor, for ranking on single word plurals. I've also seen pages ranking on single word plural queries which have other strong reasons for ranking.
With regard to the linking described, I believe that the important factor is where a page is located in a variety of taxonomies/folksonomies. I feel, this same principle is at work within the structures of sites that did well in Panda. It may have gotten turned up in Panda, or it may be reinforced by other factors, but it was happening before.
An added factor that I believe I'm seeing in the serps now is the effect of ongoing evaluations Google has been making using its search refinements, extra pages returned, difference in geo-related data sets, etc. These refinements may in fact be lodged on a separate user satisfaction "layer" in the data structure, in a way akin to Universal results. I believe that the evaluations have moved from single-words to longer phrases.
here's a theory i have
When a new subject surfaces, people start reporting on it. Because there's not much info at first, pages with less (i'm not calling it thin) content get the top spots in the SERPs. searches for this subject would be at this point purely informative(a freind told them about the subject and they just want to see how it looks - an image perhaps...), so it makes sense not to crowd the first results with tons of info. This could be a reason why we see so many shallow pages in the top spots.
As the subject gets old, there's new info to be gained, reviews, comparisons and more in depth looks at it. it makes sense to have more content on that particular subject as the time passes. searchers could also want to get more info on it too, say for research before purchasing something. so, as the time passes, google puts more content rich pages in the top spots.
|If you're looking for a "blue widget" you arent necessarily going to search for it as a plural |
I don't agree with you here too.Many of the searches are guided by google instant and auto suggest. Even if you don't intend searching for the plural, Google might force you to use it using those tools. These days, a huge majority of searchers are influenced by these tools.
I too used to rank no.1 for both singular and plural forms of a key phrase(5 words) I track. Example "Buy blue widget in webmasterworld"
My page used the singular form of the phrase in the title. But, post panda, it has gone down to #4 for the singular form and I couldn't even locate it on the first two pages for the plural form.I wasn't interested in looking beyond.
The current #1 page uses the plural form in the title and it ranks #1 for both the singular and the plural forms. It uses "Buy blue widgets in webmasterworld".
google instant is currently suggesting the plural form.
It seems like making changes in the blind might be an exercise in futility.
|It seems like making changes in the blind might be an exercise in futility. |
Depends if the changes you are making are due to an identified weakness, or just because "you have to do something"
|I am convinced that the Google algorithm as a whole, and Panda in particular, is not using the generic "bounce rate" as a ranking factor |
I concur. Our site with the highest bounce rate is doing just fine (did not get hit). Our site with the lowest bounce rate got hit the worst.
I think it is a factor, but a small one.
| This 195 message thread spans 7 pages: < < 195 ( 1  3 4 5 6 7 ) > > |