Does Google count comments as a quality factor?

[My_Media]

Hi Guys,
I have great quality pages on my site but worried that my users low quality comments ruin the over all quality of the page.
I have many articles that have many hundreds of comments. Will the low quality comments lower the quality score?

Thanks,

[tedster]

I assume you're talking about the new quality algorithm in organic search - not "QS" in Adwords.

The exact answer is no one knows. But my guess would be that UGC of any kind, including comments, does play a part, at least. Search results often return a URL when the query terms are only in the comment area of a blog and not in the article, so comments matter for organic search. In my mind, that means they get run through the whole scoring process.

Now if I were Google, I would do my best to sort the comments out from the main article and score them differently. I think that's happening, because otherwise a lot of UGC would be a problem.

At the same time, many sites that recently tanked when the Panda Update was launched did get hurt. I'm advising two things to clients right now:

1. Moderate UGC like a hawk
2. Make sure your source code clearly delineates the Comments area, so search engines have a good chance to identify it as UGC and cut your site a break.

[crobb305]

Make sure your source code clearly delineates the Comments area, so search engines have a good chance to identify it as UGC and cut your site a break

I have a site that isn't technically a blog, but I do post user-submitted commentary. What protocol can I use to tell Google that the section is UGC? I have tried to clean up spelling, which is typically horrible when user submitted.

[tedster]

I would use something obvious and common across many automated systems - use the attribute class="comments" on the container element. We know Google keeps track of common class names - it was their research in that area that informed the new HTML5 element names.

Note that this is just my best guess - I haven't seen any standard or suggested standard in this area. If anyone has something, I'm all eyes and ears.

[TheMadScientist]

I would probably go one step farther than a recognized class name tedster, and switch to HTML5 a bit early and section the content too if I was concerned about it: [dev.w3.org...]

So, </section><!-- Close the Main Content Section -->
<section><div [<-- or whatever] class="comments">Comments Here</div></section>

[Lexur]

It seems you think Google is smart enough to recognize good writing or scientific from pub level comments.

From my point of view, user comments have two properties regarding Google:
- it is original content
- onpage content is updated

So yes, IMHO comments are important if there's no offensive words or spammy messages.

[dickbaker]

My product pages aren't tied into any database, so users can't just generate comments. I have it set up so they're emailed to me, and I edit them for punctuation, spelling and, to some extent, grammar. I don't edit the meaning of the comments.

I wonder if these comments count as anything?

[tedster]

It seems you think Google is smart enough to recognize good writing or scientific from pub level comments.

Google does measure reading level, at least. We know because it is an advance search option reference [webmasterworld.com]. That's not a major feat - MS Word has been doing that for years as well as checking grammar.

[Planet13]

I have it set up so they're emailed to me, and I edit them for punctuation...

I hope you don't mind me asking how well that works and if you have any tips on how to increase interaction by visitors. I have something similar (there is a link to post a comment that takes them to a single form page).

I don' really get any comments from it at all. Like one or two a year.

Any tips would be appreciated.

As for designating a comments section, I would look at various wordpress templates and probably blogger sites (blogger is the google owned blog, right?) and see what they do, and try to mimic that.

[denisl]

@tedster you made me take notice about identifying user generated comments or reviews.
Then I realised where I have reviews on a page, they are under an <H2> heading saying "Lates User Reviews". Surely G will realise that anything between that heading and the next is a review.

[tedster]

Maybe. Sorry for the wiffle-waffle, but that's really all we can do right now - guess. The issue I see is that H2 is not a strong enough signal because it is misused so much. I'd still suggest adding a container element wrapping the entire Comments area.

[crobb305]

I don't really get any comments from it at all. Like one or two a year.

I wonder if we're thinking in the wrong direction. Perhaps having NO comments (on sites where comments are sought) is a quality signal (to the algorithm --I'm not knocking anyone's work). I remember reading on Matt C's 2009 webspam goals that the top complaints were empty reviews and cost comparison sites. I take this as meaning any type of review/comments section that have no content. At a time when we're thinking of ways the algorithms can detect quality, it seems to me that user interaction (through pageviews, prints, shares, comments, etc) could all be indicators. Maybe we need more comments?

[TheMadScientist]

I thought the same things crobb305 ... Sometimes I think people look at what's there a bit too much and need to also focus on what's NOT there...

[Planet13]

I remember reading on Matt C's 2009 webspam goals that the top complaints were empty reviews and cost comparison sites. I take this as meaning any type of review/comments section that have no content.

You might be right, but at first glance, I am not sure i would interpret it that way.

My understanding was that sites that basically act as a doorway / affiliate site (such as shopping comparison sites) would take a hit.

I didn't see anything confirming that from Matt Cutts, but I did notice that LOTS of sites that make their revenue from Adsense were complaining that Adsense suspended their accounts because they provided little original value.

So it would be interesting to see if anyone had a page (or site) that previously ranked well, but that suffered ranking losses after adding comment sections that did not receive many posts.

[crobb305]

TMS, another sign which indicates to me that Google is valuing UGC more now than before is Yahoo Answers ranking #1 for many terms that have user-generated content matching the query (with 4 suggested urls below their listing in the Google SERPS). I also see an India classifieds site with user content, ranking in the top 3 for some terms in my industry. Here in the U.S. -- an India classifieds site? Really? lol

[dickbaker]

Planet13, I'm getting about one review a day. Considering that this one manufacturer has over 100 different models that I'm showing, it's going to be a while before there's a lot of reviews.

Per Ted's suggestion on having the word "comment" or "review" in the container, I've created a new CSS font style called "comments", and have put it in the <p> tags of the review area, as in <p class="comments">

Don't know if that will have any effect. Any ideas?

[crobb305]

dickbaker,

My comments are like yours: they are emailed to me via a comment form, and I post them on a feedback page. However, I have been bad about editing for spelling and grammar. Nevertheless, I followed the suggestion from Tedster and also looked at the format others are using in Wordpress:

I just wrapped my entire comments section in a div tag, without modifying CSS in anyway:

<div id="comments">all comments</div>

[tedster]

I certainly don't know if that step will help - I'm just guessing. But it seems to me that Google must be treating UGC differently in some ways. It can't hurt to offer a guiding hand.

[Sgt_Kickaxe]

Take a look at Google blog search results and notice the following...

- The top sites for many searches do not allow commenting at all or require a login.

[TheMadScientist]

TMS, another sign which indicates to me that Google is valuing UGC more now than before is Yahoo Answers ranking #1 ... I also see an India classifieds site with user content, ranking in the top 3 for some terms in my industry. Here in the U.S. -- an India classifieds site? Really? lol

Okay, I'll give you the total lol at the India site ranking, but let's stop putting all the 'blame' or 'burden' or 'obligation' on Google for a second, since their true mission to stay in business has to be 'be a service to users', so what do the rankings tell us about what Google's users like, and what Google's users may perceive as quality?

I know we often talk about what Google 'likes' or 'ranks', but what Google likes and tries to rank is what their users seem to respond in a positive manner to, so I think we can learn quite a bit about the end user with Google saying the implementation of Panda was a success and then looking at the rankings...

They're the ones with access to user feedback data and they're saying Panda did what they intended, so what's this tell us about the end users? If their feedback didn't say their users like it they would have reverted...

[crobb305]

but let's stop putting all the 'blame' or 'burden' or 'obligation' on Google for a second...They're the ones with access to user feedback data and they're saying Panda did what they intended, so what's this tell us about the end users?

First of all, the title of the thread is, "Does Google count comments as a quality factor?" not "Do users prefer user-generated content?" So my reply wasn't intended to address the feedback that users are sending to Google.

Second, my post didn't place any "blame" or "burden" or "obligation" on Google. In light of the discussion here and the title of the thread, I am looking for clues as to whether or not "Google counts comments as a quality factor," so my wording was, "Another SIGN which indicates to me that Google is VALUING UGC more now than before is..." which tried to offer my theory about Google's valuation of user-generated commentary.

Finally, Google only spoke of the positive feedback they received, they failed to mention any of the negative feedback. It's clearly one-sided propaganda. Now that is a criticism.

I like to read your comments here and yours are among the first that I look for, but your reply seemed like an attack. We're looking for signals that Google is using to rank sites, not feedback that users are sending to Google. Hopefully my mood is just sour and I maybe I took it the wrong way?

[edited by: crobb305 at 5:30 am (utc) on Apr 10, 2011]

[TheMadScientist]

Oh, sorry crobb305, I didn't mean them that way at all ... Really, I didn't mean for them to come across like they did at all ... I've actually had the issue with type rather than verbal communication more than once, and usually try to do my best to make sure I'm saying what I mean and even then I sometimes come across way more 'abrupt' than I intend when re-reading, but the way you read my comments was absolutely NOT the way I intended them, not at all.

I hope things get better for you, and if it makes you feel any better, my day's been absolutely s***ty too ... LOL ... Sorry what I said was taken wrong.

I was trying to say more of 'let's stop looking at Google for a second' (me included) and focus on what their users are 'saying' about things (comments included), because I often look at Google's algo as 'the end of the discussion' too, but what it really comes down to is what Google's users value ... I think it might give us some new perspective?

[crobb305]

Oh, sorry crobb305, I didn't mean them that way at all ... Really, I didn't mean for them to come across like they did at all.

I have a feeling I am just in a crappy mood, after working for 12 hours on a site redesign today and then dealing with a laptop crash. I was being defensive. I think we're on the same page though, even if we said it differently. I feel like Google is valuing UGC more, and for the reasons you put forth -- that perhaps the user data they have support it. My gut tells me that with the internet becoming more and more social, increasing/expected user-generated content ties in with that trend.

Having said that, Sgt_Kickaxe provided a counter example from Google Blog Search that I haven't had time to look at yet (top ranking blogs that have little or no comments).

[tedster]

what it really comes down to is what Google's users value

Exactly. Here's what I've been able to surmise:

- Any algo factor starts out with a "best guess" from a human engineer
- Google's human evaluator squad may be called on next to check out preliminary results
- Then the new factor usually gets some low scale live testing, and the first level of live user data kicks in
- Now the new algo factor goes completely live and full scale user data is collected

There is most likely an automated loop following on this final step - a machine learning process that can tweak the algo factor's dials, to a degree. If automated tweaks of the results don't meet statistical expectations after some time, then we're back to step one, a human engineer's best guess at how to revise.

Yes, that's a lot of theory. But there is also a lot of study behind it.

[Lexur]

For those receiving comments by email I would suggest [scripts.oldguy.us ].

Regarding the quality of the comments under articles I still can't understand how could Google know if a text is wrote by an idiot or a rocket engineer. The Google's algo can't know if the solution for the hunger in the world is to eat less, produce more meals or to distribute better what we have now. Nor even if a text is wrote quickly from a iPhone in the train station or after a long reflexion in a silent library table. Of course this is only in english; can't see the Google search team understanding the spanish, german or french grammar rules.

I firmly believe that user comments without spammy words or links are a sign of quality. In my own case, I read a local newspaper and a lot of times I go directly from headline to user's comments.

I value a site for the amount and lenght of user comments. I.e.: webmasterworld.com

[pageoneresults]

Me? I'd just put the comments in an <iframe> and noindex, nofollow that feed into the <iframe>. There is just too much risk involved with UGC that is not moderated and/or not blocked from indexing. But, that's just me. Others want those comments to be part of the indexing equation. In a pre-moderated environment, that's okay.

[Planet13]

I feel like Google is valuing UGC more, and for the reasons you put forth --

Does google value the CONTENT? Or do they value the interaction?

Remember that site a few months back that faulted to the top of the SERPs because it had gotten a ton of bad reviews?

And I wonder if "unique" might, in some instances, trump "quality." I wonder how well a site would rank if it was a very controversial blog / forum / whatever that had incredibly idiotic posts based on made-up facts?

I often think about creating a site where the basic premise is some outrageous claim, like whale flatulence leads to global warming, and hence all whales need to be culled. I would create all sorts of made up stats and faked interviews with non-existent scientists and university professors.

My guess is that the "uniqueness" of the material might help it to rank quite well...

[Fribble]

I think this was part of it. I have several sites and the only one that got hit was the one with thousands of comments that were riddled with grammatical errors and spelling mistakes. U no da 1s dat im talkin bout rite?

We have corrected many and deleted many since then, but like everyone else we are still waiting for something to happen... We's seen a partial recovery (from about -40% traffic up to -30%) but that's it so far.

[walkman]

Some sites with comments were tagged as forums by Google, sending the down since they actually sold furniture and books and...

[crobb305]

Tedster,

I am going to test the <div id="comments">
Do you think each comment needs to be further sectioned so that the bots know that it isn't one big comment, versus 30 little ones?

I know it's just a suggestion and it's worth testing, otherwise I am going to block the whole page from Googlebot.