A Simple Theory about Panda

[aristotle]

Here is a simple theory about Panda:

-- Panda evaluates the "quality" of a website mainly by analyzing user-behavior.

-- By user-behavior, I DO NOT MEAN BOUNCE RATE. Instead, I'm referring to more reliable indicators of quality, such as
----- User bookmarks a page as a favorite.
----- User saves a copy of the page on their hard drive.
----- User prints out a copy of the page.
----- User returns to the same page later.

-- Google mainly uses the Chrome browser to collect this data on user behavior. Tens of millions of people now use Chrome as their main browser. This is enough to allow Google to collect statistically meaningful data. And Chrome enables Google to collect data for ANY WEBSITE.

-- In order to evaluate a site statistically, Panda needs a minimum number of user-behavior data points. Thus, the data must be collected over a period of time. As new data is collected, the oldest data can be discarded, but enough must be kept to enable a meaningful evaluation.

-- At the present time, for some sites Panda could still be using data that was collected as far back as last year, because it is still needed for a statistically-meaningful evaluation. This could explain why people who made big changes to Panda-affected sites still haven't seen any major ranking improvements.

[live3life]

Bounce and time on site are things that can be monitored

[suggy]

Walkman - it is definitely Panda! Happened on UK launch April 15/16? Logfiles went from 120mb/ day to 45mb overnight.

Also, I didn't say anything about it taking a week. Like most webmasters affected (and especially those spending £1m building their dream home as this hit!), I have been working tirelessly on this since Panda hit the UK.

What I said is that I have been able to bring pages back to just shy of their former positions. So, instead of being number 1, now number 10. I anticipate (nee hope!) that once the site wide data is rerun (and takes into account my new slimmed down site), those pages will rise closer to previous heady heights.

Where others see a penalty, I see my failure to completely restore rankings (though I can hit page 1 on very, very competitive terms), as a symptom of the massive changes I made to the website, post Panda. It's going to take a few crawls for Google to survey what's left and someone to press the 'update panda score' button at Google!

[Whoa]

Suggy,

Thanks for this feedback. I've been thinking about peeling off two big sections of my main site and putting them each onto their own separate domain.

To identify sections to peel off, I rejiggered my XML site map so it conveyed more meaning. To do this, I added more hierarchy to it and changed the names to better identify the sections. Before, I had names like Site-Map-00010000-00030000.xml -- purposely vague to throw off would be scrapers. But it also made it hard for me to see any feedback from Google in webmaster tools.

To my surprise, the two sections that I now know I need to peel off only had 50% of submitted pages indexed. Every other section was high 90% ish. So to me, that immediately tells me that Google thinks the 50% sections are full of supplemental-quality pages: either they crawled the pages and view them as supplemental (not worthy of indexing) or they don't think the sections are worth crawling aggressively. Personally, I like these two sections but if Google is telling me that they dislike them, then what's a guy to do?

These two soon-to-be-peeled-off sections represent a very big percentage of total pages. My hope is that once the core site is lean and mean, things will rebound.

I'm sure others have suggested evaluating XML site map submit-to-indexed ratios to look for weak areas of a site, but it only recently occured to me that this was my best (and easiest) way to find out what Google didn't like. Everything else I thought of was data overload and too complext to evaluate -- I like the submitted-to-indexed ratio because it's so simple.

Now, on to the pain of peeling off those pages. The whole premise of judging parts of a site based on the quality of other distant parts of the sites is preprosterous (just serve good pages up google, with good results!). But I guess this is the new world order.

[indyank]

suggy, I am in the same shoes as you but for the following.

1) Not all pages recovered to the same or similar extent despite all of them were addressed.
2) Even pages that appear to have regained lost positions, though not completely, the traffic isn't anywhere near what is was before.It isn't because the pages are slightly down but it is because Google is controlling the traffic.

This might be the story for many who still believe that Google hasn't switched on the magic button that it turned off earlier. But it is the 4th month since we were hit and it is hard to believe that Google hasn't re-evaluated them yet.

[indyank]

I have to add that the duplicate content was because others have copied us and we had to re-write our pages to salvage some lost positions, though it isn't anywhere near what it was before.

It is precisely this which makes people cry foul against this update.

[walkman]

@indyank
when I talk about Panda recovery I talk about total G referrals, not a specific rank since they go back and forth. I deleted my 'thin' pages on 2/25/11 and added plenty of content by April 9th on all pages and been changing things almost every day. Whether Google is still holding traffic back or not or ranking haven't improved I don't exactly know. It's possible that a drop of 3-4 places accounts for page 2 or 3 and traffic goes away.

[suggy]

indyank - I don't have an opinion on whether google's throttling traffic. I check the ranking on my target pages pretty often and from multiple browsers and in secret mode and they're consistent.

I do know that 5th place to 9th place is a graveyard. 1-2% max of possible referrals in these positions. The decay in potential traffic from 1st place is pretty steep, so just losing two places can hurt, big time and I expect poor referrals as a result.

The thing I noticed most recently was <title>. Short and precise became my trademark. Trouble is, it made them all very similar. Red widgets, blue widgets, green widgets. I don't think google likes that. Unique your titles up; think of them as Adwords entries.

How many pages you boys got?

Some research I read relating to a Google patent suggested 15-20,000 page clusters and above are correlated to spam. Remember Matt Cutts "this side of the line"?! If you have a section (like my site search folder that Google was indexing, doh!) or a whole site of this scale, you will more likely be on the Panda radar. This explains small sites floating up. They're too small to be spam!

[anteck]

Here's my honest opinion about a simple theory for Panda:

Google engineers looked at the issue which this originated from : The big content farms that people were complaining about.

They did an analysis of those sites and found some 'common factors'. Matt Cutts mentioned this process in a video, about using a 'document classifier' and low quality sites appearing on 'this side' and high quality on 'that side'

So they developed an algo which emulated that analysis, and then put it to work in the SERPS.

It worked. For the sites they wanted. It destroyed a bunch of great sites also. Why? because google only geared it to work on the sites they looked at. They didn't test it further.

The key point i'm making :

It was a rough slapped together move. Google don't care they took down heaps of great sites. The algo is crude, and only works on a number of factors which those original sites they wanted to take out possess. The logic is flawed. It's not a simple few factors.

It's akin to saying 'Criminals are unshaven, drive black old cars and wear baggy pants' thus:

If you drive an old car
or
If you wear baggy pants
or
If you are unshaven

You're penalised.

I know it's a simple analogy, but i hope you get my gist.

The worst part is, google have just gone silent. Serp results are rubbish. Try searching for anything slightly complex. Relevance is now gone. Page after page after page, just to find what i'm looking for. Oh! There it is, on page 8. Funny, that used to work... Hmmm...

[suggy]

anteck -- my thoughts exactly. It's a stastical profile of your site based on correlation. Matt Cutts mentioned a line and being on one side of it or the other. Sounds like multiple regression analysis to me.

[mcolom]

anteck, I'm in complete agreement with you. Also, the panda update is the perfect example of why G should be hit by an antitrust case, and splitted or removed from marketplace dominance like IBM and ATT were years ago. They have too much power. Their decisions affect all the internet. That's not a sustainable or desirable situation.

[freejung]

antec, what you say makes a lot of sense, but I'm not sure your theory contradicts aristotle's. Some of the common factors you mention could be user behavior metrics. Basically "people respond to criminals by crossing the street to avoid them, so if people cross the street to avoid you then you must be a criminal."

[walkman]

Some research I read relating to a Google patent suggested 15-20,000 page clusters and above are correlated to spam. Remember Matt Cutts "this side of the line"?! If you have a section (like my site search folder that Google was indexing, doh!) or a whole site of this scale, you will more likely be on the Panda radar. This explains small sites floating up. They're too small to be spam!

Far, far less than that even with searches indexed before Panda. Now I have around 1000 pages and I have been rewarded with decreased traffic each time. Content and 'thin /shallow' pages are absolutely not an issue now, just the non-penalty Panda penalty.

Basically "people respond to criminals by crossing the street to avoid them, so if people cross the street to avoid you then you must be a criminal."

I think we're questioning G's ability to do that in mass scale. While it's not jail we face, we face a loss a of our income and that's extremely serious.

[suggy]

Hi Walkman

15,000 - 20,000 was in the research, but maybe google chose a much lower breakpoint. Certainly, my other sites in the hundred-or-so page count have been completely unaffected by Google, but that maybe for other reasons. They're all Wordpress (whereas biggie hit by Panda is on own proprietary CMS), so maybe Google is giving mainstream blogging platforms (incl. Blogspot) a free ride? Again, content farms are unlikely to be powered by Wordpress or Blogspot?!

[walkman]

suggy,
I understand your point. I think that Google will run its algo to 'free' sites at a certain page count and down. They are scared to death that Mahalo etc are back at #1 if they run panda full speed again.

Regarding the rest: Not all niches had Panda IMO, as my other sites that could have been hit will tell you.
The site I was hit has no articles but suffered from a perfect storm of too many and too similar tags and possibly Google counted <!-- disabled--> Google ads too. All these are fixed and a lot of content has been added.

[falsepositive]

Suggy,
Tech blogs and many other blogs in various popular niches were hit hard by Panda, including mine. I lost 75% of google traffic and all I have is a blog. After evaluating players in my site who won and lost, it really just boils down to authority. What Google looks upon as authority is now raking it in. What I mean is this: if you are an authority or a known brand in your space, it matters not what you do. You get away with everything that's seen as "bad" behavior (according to Panda). If you have a strong enough profile, you can really do a lot of stuff and that is the reason why we are seeing such diverse and inconsistent results.

A competitor of mine who is running all the way to the bank and monopolizing SERPs now has around 65% more links than I do, and I have some pretty strong links. Their site is ugly, messy, hard to read, but Google loves them. Design is bad, overwhelmed with ads, while smaller sites that are very clean but probably made a few mistakes here and there were punished greatly.

It is really all about stealing from the poor and middle class and giving to the rich (in authority). Google wants to consolidate search results to show the "safe" and popular sites, regardless of page quality, in some cases. So even if you are a good business and run a decent site, it's no longer good enough.

you have to look at your winning competitors and find out why this is happening.

[gadget26]

Falsepositive,

I agree completely. I'm pure retail in my niche, of which there are only about 4 main competitors. Three of us were Pandalized. The biggest and best known of the lot is now getting most of our clicks. What's different about them? Just bigger from what I can tell. More pages. More links in. More volume. Much bigger adwords spend. Their website is arguably no better or worse than the rest of ours. My company and website are ironically exactly the same age as theirs. We all conform to and violate similar Panda "rules".

Are others seeing this same thing?

One cool thing about the internet is that we can appear as big and corporate or small and folksy as we want. But Google knows too much. I don't see any way that we can fool them into thinking we're a big brand. (Without spending a LOT of money.)

So the rich get richer. And the great divide widens. If nobody (or very few) can cross the divide, what will the SERPS look like in 5 or 10 years?

[walkman]

gadget26,
I have noticed the same thing as have others: Google is picking a few 'winner' sites and slowly killing others. That's deliberate and there's nothing we can do. Listening to Google can make you even poorer as improvement of the site will not result in a comeback. At least we haven't seen one yet. Google gained the market share and now they are using it however they want.

The top sites in my niche have hundreds of thousands of pages, all but a tiny percentage is super thin but they are ranking. Others are going nuts and working night and day trying to decipher Google's lies about misspellings or mentioning the other point of views. Fact is, as things are now, 99.99% will not come back. It's not up to us, it's up to Google, unless you have a $1 million to spend on adwords etc to create a 'brand.'

[proboscis]

It's a stastical profile of your site based on correlation.

So that means a heavily scraped site will be thrown out with the scrapers?

[agent_x]

That is what I've been thinking. Google are using the "if it walks like a duck and quacks like a duck, and enough people are throwing bread at it, your site is a duck" method of analysis, and the problem is they can't tell the difference between a duck and a swan.

[proboscis]

...they can't tell the difference between a duck and a swan.

Couldn't that be exploited pretty easily? To take out a competitor I mean. If many people are trying this it could be really bad when they rerun panda...

[HuskyPup]

After evaluating players in my site who won and lost, it really just boils down to authority. What Google looks upon as authority is now raking it in.

You're obviously not in my sector since we are THE brand, we are the Coca Cola/Mercedes/Sony of our industry, we've been around since the 1840s.

Obviously it's different within all widget sectors however for people to keep posting that only the brands have won with Panda is ridiculous, I have seen many quality, authority sites simply disappear and replaced by garbage plus my own stuff scraped and ranking well.

IMHO anteck has come to a similar conclusion to my own therefore it is pointless doing anything to ones' site(s) since it's most likely/probably not going to make any difference simply because many have been caught-up in an almighty Google fubar.

For certain it is very frustrating but to chase after hypothetical suggestions is the worst possible thing to do. If the combined intelligence, experience and knowledge of WebmasterWorld members after 3 months cannot positively identify the exact issues, then the best advice is sit tight and see what happens, if it ever happens, when they supposedly re-run this flawed "update".

[walkman]

HuskyPup,
1- even the biggest G cheerleaders understand that the algo is not perfect. No algo is, but especially new algos.

2- brands are favored major time but see #1

3- if good sites go away in Pandaland, junk that was not a target of Panda (different niche?) makes it to the top until the next time.

4- Thanks to panda, branded sites can put even a tweet on their page and rank it as your 2000 word article is on page 30 becuase it's a sitewide penalty.

5- It will be improved but when...3 month$ and counting.

[Kenneth2]

6. Big G favors strong user engagement. (Without hitting the back button to G after clicking the searched result)

Example/my speculation : In the niche I'm in, Torrent and pirated sites seem to gain favor in panda because users usually get what they look for (at torrent site/pirated free site) instantly without clicking the back button to G for more search.

[suggy]

for people to keep posting that only the brands have won with Panda is ridiculous

Agreed!

Only superficial evidence to suggest brand dominance. It's probably coincidental. Perhaps many 'branded' sites simply share the same characteristics that unbranded don't?!

Unbranded (ie our sites) are often highly SEO'd (that's why you're hanging out here, right?). We figured out how to rank for keyword searches and then exploited that ruthlessly in rinse and repeat fashion.

Branded sites in my niche are mostly sleeping giants; the big name bricks and mortar retailers who rarely have aggressive, acquisitional SEO strategies. SEO is more of a check list of things to ensure their pages / sites are disadvantaged for many of them.

[walkman]

suggy,
brands are hardwired in Google's algo, since the Vince Update. It's not even a theory anymore. [google.com...]

As for Panda, did you read of the questions asked to start the algo up?
They rely heavily on trust, very heavily. Also other signal like social media, mentions or searches obvious favor brands by a wide margin.

[suggy]

Walkman, if by brands you mean you want to outrank Seiko for a search for Seiko watches, I can see your problem and you're daft to try that anyway.

If by brands, you mean that you cannot rank no.1 for "men's watches" because brands of men's watches are hardwired above you, then that my friend is utter tosh.

And, please -- forget social media. It's another factor that plenty of sites (ie the vast majority) that do very well, do very well without (as mentioned earlier in thread). When I look at what's now outranking me in my niche, it's not the high street multiple retailer's (and their gigantic brands) or the generalist online giants (like Amazon and Ebay), but the online only specialists who are often no bigger brand than my site and have equally inept social media strategies.

Oh, and when I say, niche I mean bloomin' highly-significant-portion-of-total-online-spend size of niche. A niche that is rife with competition and includes public quoted pure play and clicks to mortar players. I'm not trading in some obscure widgets, competing with Mom and Pop stores (in the main).

[superclown2]

For certain it is very frustrating but to chase after hypothetical suggestions is the worst possible thing to do. If the combined intelligence, experience and knowledge of WebmasterWorld members after 3 months cannot positively identify the exact issues, then the best advice is sit tight and see what happens, if it ever happens, when they supposedly re-run this flawed "update".

Huskypup, I agree with everything you said except for one point; this, in my opinion, is not an update.

It is a completely, radically, different ranking system altogether to the one that we have grown to know and love. Like every new system it has thrown up all sorts of problems that were never envisioned. I've not the slightest doubt that, right now, there are a lot of very clever people sweating hard at the Plex to get this thing working properly.

Page is an idealist who wants 'perfect' search results. He has a history of not letting anyone stand in the way of his mission to make all human knowledge available through his search engines and that includes webmasters and, yes, Google shareholders too. This means that Panda is not the final algo, and not even the beginning of it. Anyone who dreams of him abandoning this project to go back to the old system is grasping at straws, it just isn't going to happen.

We are going to have to get used to this sort of disruption whether we like it or not. Those who accept this and take on the challenge may reap huge benefits, those who sit it out and hope that things will improve will face an uncertain future. But that is the nature of all things.

[Whitey]

You're obviously not in my sector since we are THE brand, we are the Coca Cola/Mercedes/Sony of our industry, we've been around since the 1840s.

Absolutely .... I'm hearing back from a small number of major brands that have lost a lot of traffic. Whole associated SEO departments sacked in the wake of it. I guess it might be a lot more widespread amongst brands than we are given to think.

Needless to say those brands are spending their dollars elsewhere ... guess where.

Watch Google's results for Q2. This has to have been commercially driven using search quality as an excuse to raise revenue. And therefore why let sites return quickly that improved.

[sailorjwd]

I stumbled upon one problem with Panda today. When searching for the service I provide in major cities the results have been taken over by two or three websites (who don't have a presents in those places). But the real problem is the listing is reproduced 3 times in spots 1,2, & 3. Each of those websites is identical except for the Title and some footer keywords.

PS.. it is actually 3 different pages from the same website. You'd think a dup content penalty should be applied but I bet they are tricking Google somehow.

[walkman]

If by brands, you mean that you cannot rank no.1 for "men's watches" because brands of men's watches are hardwired above you, then that my friend is utter tosh.

Cannot is strong a word, good luck ranking, is more appropriate. And yes, I am talking about general terms like , say, 'men's jeans.'