Thin content definition and its play in the Pandamonium

[whatson]

Ok, so we all know that duplicate content was a main crack down for Panda, despite it partly backfiring with scrapers out ranking original, etc.

But the other side we hear about is thin content. People saying that every page should have at least 400 or so words of unique content.
How true is this? Is it feasible? Is it fair? Is it relevant?

What about sites like: e.g.
A java game site - the only content is the java app to play a game.
Photo/image sites - the content is all in the images, perhaps their could be some accompanying text, but what if you have 1m+ images.
Flash site - some flash tool that people find useful, interactive map perhaps.
Business directory - like the yellow pages, they have web pages about businesses, but only contain the name, address and phone number.

And many more similar pages, that simply cannot or do not require much text. Are these pages/sites to suffer from Panda? Is there evidence of such sites being Pandalized?

[potentialgeek]

Peter at searchenginewatch, who did the Panda tests, posted a comment under his report:

"If you only have about 20 pages of unique text at 500 words each, those aren't likely to be hit."

His multiple tests, he had said, "had everything to do with thin content." (Of course everybody was asking for his definition of thin content. (I wish he'd post the information on the page lengths.)

There are other factors of course, but there are now two people who achieved Panda recovery who are suggesting a minimum page length of 500 words.

Funny how ezinearticles arbitrarily set their minimum at 400 words for their writers before testing. They made the announcement not long after Panda first struck and certainly before they had got recovery. It could also explain why their previous bar at 250 words tanked their site.

500 words isn't set in stone but webmasters who suspect their sites have thin content need some kind of general idea to know which pages to develop.

Panda 2.5 may have set the bar at 300 words or 400 words but hovering around any amount below 500 words could be skating on thin ice.

I doubt all pages at identical or very similar length would be safe. If every page on your site is 500 words, for example, it'll look like auto-generated articles.

[indyank]

I think looking at quality as "number of words" per page will be misleading. The various types of content used on those test sites and the way they were written and presented, would have serves as good examples.It would have been great if he had given those examples i.e. how the pages looked before and after recovery.

Yes, I do know the ones that aren't affected have lengthy stuff like for example most pages on squidoo, but I would be very surprised if Google had used the number of words as a direct or indirect factor.

[sundaridevi]

I still think that the type of page has an affect on what determines quality. I have a movie site that has moved up in every iteration of panda. Now it is number one for almost every search I care about (and is helping dilute the effect of the drop on my bigger site). Only 2 or 3 pages have more than 200 words. All my text is original, plus includes info not easily found elsewhere. The majority of competitive sites liberally copy/paste text from sites like wikipedia or imdb. But there is also a movie on all my pages....and oer 70% of visitors are returning visitors.

So taken by itself word count alone, in isolation, is a non-issue

[potentialgeek]

@freejung

I believe there has to be counter evidence for Google to give a page good ranking when it would otherwise be penalized/Pandalized. This would either be the quality of the page due to length, popularity, features, outbound links, inbound links, etc., or all of the above.

@sundaridevi

My experience, although limited, is that Panda likes videos - not only Youtube and Video sites but also sites with Youtube/videos embedded. One of my few pages on a Pandalized site that survived Panda in rankings has a YT video embedded.

@indyank

I think looking at quality as "number of words" per page will be misleading. The various types of content used on those test sites and the way they were written and presented, would have serves as good examples.It would have been great if he had given those examples i.e. how the pages looked before and after recovery. Yes, I do know the ones that aren't affected have lengthy stuff like for example most pages on squidoo, but I would be very surprised if Google had used the number of words as a direct or indirect factor."

A while back you made a post I printed out. You said, and I quote, "Matt and Amit are on record saying that Panda likes detailed pages and pages with different viewpoints." [webmasterworld.com...]

I believe it's virtually inherently impossible to publish relatively detailed pages or to express only one opinion of significance if the page is short. I know there are long pages and many of them on the web that are useless. But a large site full of thin pages with low word count per page is rarely a sign of quality. The site would probably have to pass other tests to be accepted and rank well (e.g., links from trusted/authority sites).

[indyank]

Yes I do agree that Google likes more details on a single page and squidoo is a good example for it.

But what I am saying is it is not just the number of words that determine how detailed a page is. You could explain a lot of stuff using different forms of content like photos and videos.

[edited by: indyank at 6:19 am (utc) on Nov 5, 2011]

[indyank]

sometimes you can display content using javascript too...Google is now recognizing content within javascript as well..

so it isn't necessarily the plain text and the number of words that should determine quality. It would be really bad if they aren't recognizing other forms of content.

[freejung]

But a large site full of thin pages with low word count per page is rarely a sign of quality.

You could explain a lot of stuff using different forms of content like photos and videos.

That's the thing -- in my case, my site is entirely about providing images for download. The images are not intended to "explain" anything. They themselves are the entire point of the site.

If your site is all about the text content, it doesn't matter much in practical terms whether it's about length or a more sophisticated measure of quality. You should flesh out your content with more detailed, complete information, which will probably involve increasing the length -- that way you benefit either way.

But for an image site, it matters a lot which theory is correct. I can spend time trying to write more text to describe my images, or I can spend time updating and improving the images themselves. It's a tradeoff. I don't want to waste a lot of time trying to write long descriptions -- I honestly don't think my visitors give a flying fig leaf about the text. They're there for the pictures.

If the measure of quailty is sophisticated and based on usage metrics, I should focus on improving the pictures. If it's simply based on text length or even based on more sophisticated textual analysis, I should focus on improving the text.

Thus far most of my efforts have gone into improving user experience -- interface improvements, better design with images featured above ads, improvements to the download process, additional features such as custom image sizes, speed optimization, that sort of thing. I've done most of what I can do in that regard, and it hasn't helped. So now I'm turning more towards improving the actual content, and I have to decide: do I focus on the images or the text?

The guideline of optimizing for users rather than search engines suggests to focus on the images. Your theory, potentialgeek, suggests to focus on the text. My inclination is to think that Panda is more sophisticated than simply looking at how much text you have, that it's smart enough to understand that for an image site the quality of the images is much more important than the quality of the text. So I'm going to focus on the images and hope I'm not wasting (more) time.

[sundaridevi]

I've also heard that videos perform better in Panda's recent updates. And for sure having a good mix of media on the page is better. But all things being equal, the video or photos site with the best descriptive information should rank better.

On the google webmaster forum I saw a post directing somebody to a page that describes how to use a form of tagging that is something like snippets. I thought I bookmarked it, but can't find it now. Anyway the general idea would be that for something like a photo, if you call out stuff like: author, date, location, etc. The page is richer, more informative => higher quality

The other problem you can usually find in sites like this is that the navigation really sucks. So if i want a photo of something. Let's say a black dog, can you easily tell me how to get there from the home page, in words? And once there, how easily can i get to a page of a white house? Can i just click a category menu then find it in one or two more clicks, or do i have to go back home? Or is it only via search (really bad IMO). This stuff is easy to evaluate algorithmically and I'm sure people who are marrying Panda and thin content are not even looking at the site architecture aspect of it. Well meshed site architecture has always been an important secret for doing well in google, but seldom discussed in SEO discussions.

One thing that my movie site does much better than the ones it easily rose above, is simple easy flat navigation

[freejung]

schema.org - and good points, though I'm not sure I have much room for improvement in those respects. I use schema, and have put a lot of thought into my nav.

[sundaridevi]

Yeah - schema.org - that's it!

and ahem

... a site map should be considered fundamental ...

to good nav and to get under google's good graces, since they even offer their own tools to creating one.

It reminds me of a university course I once took. When the professor handed back the final he said that if the syllabus included reading that was written by the professor you can be sure that stuff from it is going to be on the final exam. He was really peeved that so many people hadn't read his book and got the question wrong.

In this case, I don't think you need to use google's site map you could also make your own. But I think you are infinitely better off with than without.

[freejung]

Good point -- more generally, good information architecture is absolutely crucial with or without Panda,and something we don't often discuss in the context of Panda. However, content farms could have great architecture and still be useless. Again, it seems likely that user experience is the primary factor, of which nav may be a significant but secondary component.

[sundaridevi]

Digging in and digging back out is bad information architecture and bad user experience

[freejung]

Hmmm... six months ago I would have said that an HTML site map is superfluous if you have a decent taxonomy. Now I'm much more open to suggestions. OK, I'll put one up. I'll be truly amazed if it helps with Panda, but maybe some visitors will get some utility out of it.

[potentialgeek]

I returned to the old thread "Our Panda Solutions" [webmasterworld.com...] and something canuckseo wrote caught my attention:

Rewriting pages. This is where my speculation begins because after weeks of testing I found a "formula" which appears specific to sites like ours - I rewrote the pages to be between 300 and 400 words (up from 100-200 words). Every site that I've done so far has recovered to about 60% of pre-panda 2.2 traffic . . .

Because [Panda] is site-specific the "rules" are different for each site, which is why I think 300-400 words is appropriate for us, but may not work for everyone. Shorter definitely does not work as it used to for us (because sites I haven't rewritten have not recovered as quickly as sites I did). And longer doesn't work as well - pages seem to drop further with longer pages.

-

My experience is similar. Just because the new "short" rule is true, doesn't mean super long pages do better. In fact, if the goal is to find unique content, it can be more difficult to write long pages without repetitive content - or content that looks as if it's trying to catch every keyword.

A lot of webmasters who aren't professional writers can too easily "fill" a page with similar content. We all need an aggressive editor - or pretend to be one.