|Some pages escaped Panda, when most of the site didn't|
in a different thread IBill said:
|Some sites hit by Panda still rank well for one page and nowhere for the next page that used to rank well, and it's those that pages holding all the clues. |
I want to agree with this logic, BUT there's something in my head that says Panda just somehow skipped over my pages that escaped the hit. If that little thought is right, then I'll be chasing red herrings when I take decisions based on analyzing the exception pages.
IBill's a whole lot smarter than I am about these things, and I want to follow his advice. Does anyone else wonder about G just skipping past certain pages, or am I crazy?
I saw this too. Some pages are unaffected. Like nothing has happened at all. But why? Nothing stands out in those pages. They are all normal, clean content... Guess I have to go through them word by word now!
why do you think some pages escaped? Did you notice any issues in them? What kind of issues?
Some of mine escaped too - they have a stronger backlinks profile ( more links from various blogs, forums, etc )
|I saw this too. Some pages are unaffected. Like nothing has happened at all. |
I observed this too (and I may have posted this similar observation on the same thread the OP referred to)...
I have a page on my Pandalized site, linked from the top nav, and monetized at the time of Panda with an affiliate link (just like other pages in the top nav that dropped positions), but this one monetized page escaped. I can't figure out why. Well I have some ideas (below).
My main 2-word phrase that my homepage ranked on for years in the top 15, fell 500 positions; but my unpandalized page ranks in the top 100 for that phrase and it is in the top 10 for longertails that utilize that phrase. This has baffled me. The ONLY factor I can find is that this unaffected page was linked more frequently IN CONTEXT throughout my site -- not just from the homepage or the top nav. In fact, while this page is linked from the top nav of all pages, it is linked only ONCE from the homepage; whereas my three hard-hit pages are linked 4 to 6 times on the homepage, with different anchor text and very rarely linked in context on deeper pages (I don't know why I did that, but it was definitely poor linkage on my part when I posted my articles).
I THINK there is definitely some phrase-based (almost OOP type) scoring in place. The words that were hardest hit for my homepage rankings are all contained within the anchor text of those hard-hit internal pages. I hope this makes sense... I will edit once I am outside this little reply box :)
|Some of mine escaped too - they have a stronger backlinks profile |
Oddly enough, my unaffected page has a small backlink profile, but it has a couple of E-How links (which are nofollow). Maybe those have helped? I still think it mostly comes down to excessive linking from the homepage with multiple anchor text coupled, and little to no linkage (contextually) throughout the site.
[edited by: crobb305 at 5:53 pm (utc) on Apr 18, 2011]
i have 2 theories for this (based on whatever I observed)
1) Lots of quality links do seem to save the page from loosing out in the SERPS. I think the quality algo works by discounting the links.When you have strong backlinks the quality discount may not be strong enough to have any impact on the page in SERPS.
2) Panda picks a set of pages and works on them.It might be that this page wasn't included.
Many of these lead me to believe that this is a very intensive process and google may not run this often.
[edited by: indyank at 5:53 pm (utc) on Apr 18, 2011]
But what makes you people think some pages escaped? do you see any issues with those pages?
I'm finding that the pages which escaped were in-progress rough copies that were not supposed to be posted yet. They have mixed topics and the wrong content. Coding was fine, links are ok. Its all content errors. One of the pages was at first demoted to #96, it came back today at #4.
This is not good news.
IBill was right, and I was wrong to think they were simply skipped over.
[edited by: dibbern2 at 5:56 pm (utc) on Apr 18, 2011]
dibbern2, Wasn't he trying to say that those pages which escaped will have clues on why they didn't get hit?
indy: yes, he was. I was wondering if it was something else, that G just skipped over those pages. He was right, I was wrong.
|But what makes you people think some pages escaped? do you see any issues with those pages? |
I am basing this on my ranking data for the past 7 years (mostly from memory which is fading as I get older lol), and also from my WMT data (average page positions) since Panda began. Even before February's data fell dropped out of WMT, I was actually comparing the average positions and monitoring them in 48-hr segments from then on.
For the unaffected page that I mentioned, WMT has consistently shown an average position of 1 to 5. This is also consistent with what I see when I search the terms it has always ranked for. In fact, I am seeing an increase in entry traffic through that page. The Pandalized pages fell 200 to 600 positions, then eventually stopped getting tracked (under 10 impressions = no position shown). I am aware of the problems with WMT, but it has revealed a lot of information about hard-hit and unaffected pages based on average positions (in a broad sense and when looking at longer data periods, e.g., 48h or longer).
|IBill was right, and I was wrong to think they were simply skipped over. |
I'll take a look at what he said to double check (because I'm not sure what you mean by skipped over), but my unaffected pages have been crawled/cached many times since Panda 1, and have escaped both Panda 1 and Panda 2. I still believe (speaking only for my site) that I overlinked from the homepage to my hard-hit pages, and linked more naturally to my unpandalized pages (only once from homepage and within context throughout my site).
[edited by: crobb305 at 6:32 pm (utc) on Apr 18, 2011]
Another reason COULD BE, at least in my case, that the keywords which those pages ranked for, were not English dictionary words ( nouns or else ) - they were product names
crobb305, one of sites got hit in Panda 1.0 and it was really difficult to find what pages were pushed back harder.it was because i was continuing to receive good traffic from other google TLDs and the hit was only in US.
But there was another non-US site that got hit in panda 2.0 and it was easier to find the pages that google hated the most.Since it was an international roll out things were straight forward in GWT.
I am sure that panda has worked in two phases.In the first phase, it definitely didn't pick all the sites. Those sites that escaped the first roll out were evaluated in the second one. I still believe that there are more site to be evaluated.
I also get a feeling that panda doesn't evaluate all the pages in the site.
[edited by: indyank at 6:10 pm (utc) on Apr 18, 2011]
|Another reason COULD BE, at least in my case, that the keywords which those pages ranked for, were not English dictionary words ( nouns or else ) - they were product names |
Well, funny you mention this. I have a funny story...A few weeks ago, I had a vivid dream about Panda (probably because it was so traumatic lol). But in my dream, I had a revealing thought: my hard-hit/Pandalized pages were linking out (via my affiliate links) too frequently using Trademark product names (along with Trademark symbols as required by the affiliate program). I'll point out that my site isn't "thin affiliate"...only 4 or 5 out of over 110 pages even had an affiliate link -- and only one or two links at that, with very descriptive service information).
The dream reminded me of my college days when on occasion I would dream about a math problem I was working on. Well, to make a long story longer, I woke up and looked at my hardest-hit pages, and sure enough, the affiliate links utilized the product name with a Trademark symbol. The unaffected page that I spoke about above didn't do this. I STILL believe the pandalized/unpandalized pages of my site boil down to my internal linking structure, but I did take my dream into consideration and changed the way I was linking to those services. So far, no changes, but then again no one has reported any improvement.
[edited by: crobb305 at 6:12 pm (utc) on Apr 18, 2011]
Over optimization (and overlinking) can definitely be a factor of the pandalized pages, but it does still not explain why the big drop in rankings of the homepage...
|the affiliate links utilized the product name with a Trademark symbol. |
I think they have trouble in measuring the quality of pages with symbols, data tables, names and so on.They are mistaking them to be some junk stuff like people use in spun articles.
They are definitely trying to evaluate the text and the language used but it seem to be triggering a lot of false positives.Textual analysis is definitely not that easy.
for example, when you have a big table of city names, the algo may consider the page to be of low quality (spun article) as there are no proper sentences in them.
I am seeing a lot of pages with tables of the above kind, getting hit on the seconds site.
[edited by: indyank at 6:25 pm (utc) on Apr 18, 2011]
|Over optimization (and overlinking) can definitely be a factor of the pandalized pages, but it does still not explain why the big drop in rankings of the homepage... |
Tedster proposed the idea of penalty bleeding backwards to the homepage. He would have to restate it (incase I am misquoting). In my case, if I am linking to three Pandalized pages 5+ times each, from the homepage, with different anchor text, then it wouldn't be hard to slam the homepage with an OOP-type penalty based on A) phrase-based scoring (phrases above and beyond what is "expected" for the page, including links); and B) a backwards bleeding of penalty (maybe as if linking to a bad neighborhood except that the neighborhood is several rooms of your own house).
|Tedster proposed the idea of penalty bleeding backwards |
I read Tedster's post and after quite a few minutes thinking - what's is he on about ? - I realised he could be correct - although , IMHO, it doesnt have to be the home page.
unPandalized page links to unPandalized page links to Pandalized page = opps your unPandalised pages are down-graded
<disclaimer>empathise COULD be correct </disclaimer>
I didn't pull that idea out of thin air - it came from studying ranking and traffic data for different URLs on an affected site.
One of my sites was hit quite badly, but a whole section of the site was not affected, believe it or not it was a gambling guide. The difference with the gambling guide is I have no ads on these pages (due to Adsense T&C), I also do not have any affiliate links on and monetize the gambling traffic in no way at all.
The rest of my site has reasonably heavy Adsense Ads. So this leads me to believe that ads play a major factor in this update.
|I didn't pull that idea out of thin air |
I didn't think you did - I was disclaiming my interpretation of your post ! Well thats what I meant anyway !