How Google's Panda Update Impacted Squidoo and Hubpages Traffic

As far back as 2007, Squidoo implemented anti-spam filters and policies to help it recover from a Google slap to traffic. Ongoing tweaks to spam filters and content policies, some as recently as December and January, may have helped Squidoo weather what was coming:

• On Jan. 21, Googleblog announced its 2011 campaign against low-quality content and a new single-page penalty designed to filter out keyword spam.
  
• On Jan. 31, Google challenger Blekko threw down the gauntlet by introducing an option allowing users to block sites it had designated as "content farms".
  
• On Feb. 14, Google responded with a Chrome extension allowing users to block sites from search results (already in the works, according to the Jan 21 post).
  
• On Feb. 24, Google unrolled the U.S. Panda Update with this announcement. Analysts initially borrowed Danny Sullivan's nickname for it, the "Farmer Update."
  
• On Mar. 2, Google employee "Wysz" confirmed on Google's Webmaster Forums that Panda includes site-wide penalties for "low quality content."
  
• On Mar. 3, WIRED Magazine interviewed Google engineers Amit Singhal and Matt Cutts about the "Farmer Update", which Singhai identified as the "Panda" update.
  
• On March 8, Squidoo unrolled a duplicate content policy, causing fear and confusion among members. Automated duplicate content filters sent warnings to members with lots of duplicate content, locking uncorrected lenses at month's end. (Also see this Squidoo policy update.)
  
• On Mar. 10, Google added personal blocklists to ordinary Google search for ALL Google users, not just Chrome users.
  
• On Mar. 20, Google quietly updated its quality guidelines to include hints about what Panda is penalizing/favoring.
  
• On Apr. 11, a second Google Panda Update (sometimes called "Panda UK") was extended to all English-language Google users. This update factored in the new "personal blocklist" data.
  
• On Apr. 12, 20 and 28, Hubpages announced several policy changes, causing fear and confusion among members.
  
• On May 6, Googleblog posted a guide to the Panda update, including a questionnaire which everyone should read.
  
• Also on May 6, Panda 2.1 was launched. Minor traffic perturbations were noted by many observers.

The SEO blogosphere buzzed with analyses of the Panda Updates: who got hit, who didn't, and what the Panda Update seemed to be penalizing. A lot of this discussion is speculative. We won't know the true impact of Panda for months. But here are the early returns.

Most news articles on the Google Panda Update cited figures from two sources: Sistrix and Searchmetrics. It is important to understand what these figures represent: keyword visibility. Keyword visibility is jargon for, "how many pages from a particular website appear in Google's results when you search for a particular phrase (a keyword)? And how many of those pages are near the top of the list?"

Searchmetrics and Sistrix each track a huge pre-selected set of search phrases (250,000 for Sistrix, 25 million by Searchmetrics) and monitor which sites rank in the first hundred Google results for those searches. Based on this data, Sistrix and Searchmetrics estimate which sites rose or fell in keyword visibility after each Google update.

Here's the results for Panda:

Sistrix report on Feb 26, 2011 (after first Panda Update):
Hubpages loss of 87% Keyword visibility
Squidoo not on list of 25 "biggest losers".

Sistrix report on Apr 13, 2011 (after UK Panda Update):
Hubpages loss of 72% visibility compared to before this update: a double whammy
Squidoo not on list of 30 "biggest losers".

Searchmetrics report on Mar 3, 2011 (After first Panda Update):
Hubpages loss of 80.1% visibility
Squidoo not mentioned. Says they analyzed 39 sites, but only posted results for 24 "biggest losers".

Searchmetrics Report on Apr 12, 2011 (After UK Panda Update):
Hubpages loss of 85.72% visibility
Squidoo loss of 35.89% visibility

There's a blind spot with these results. While both tests include a mix of popular, major search phrases and specific, narrow-niche, "long tail" search phrases (e.g. "ipearl macbook hard case"), they can only measure how often and how well sites ranked for the phrases they tested. This doesn't guarantee traffic. If a site gets a lot of traffic in unusual topics, or has a huge amount of "long tail" traffic -- obscure, one-time searches -- then actual number of visitors may not match the" visibility" of the phrases Sistrix and Searchmetrics tested.

There's another way we can tell, thankfully: most sites (except, tellingly, Mahalo) let a company called Quantcast track their traffic directly.

Quantcast measures traffic by embedding a 1x1 pixel image on every page of each participating site, allowing it to track each visit. (Of course, this doesn't tell us which traffic comes from Google and which from other sources.) Here's Quantcast's traffic stats for Squidoo and Hubpages before, during and after the Panda updates:

Squidoo Six-Month Traffic Stats as of 4/29/2011:


Traffic drops each weekend: this helps you read long-term traffic charts. It's too early to be sure, but it almost looks as if Squidoo's traffic improved slightly with the Apr 11 update.

 
Hubpages Six-Month Traffic Stats as of 4/29/2011:


Hubpages traditionally had a little more traffic than Squidoo, but the Feb 25 Panda Update halved its traffic.

 
Squidoo vs. Hubpages Three-Month Traffic Stats as of 4/29/2011:


Zeroing in on the last three months, the impact on Hubpages vs. Squidoo is clear. The Feb 25 US update brought Hubpages down to Squidoo's US level, while the April 11 (UK / Aus / Can / NZ) dropped Hubpages slightly below Squidoo's global traffic level.

If you stop by a forum at any time, there's always some members lamenting, "Google just killed my traffic!" and a few chiming in with, "Me, too!" while the unaffected majority are silent. But when lots of people complain, something is happening.

On the Hubpages forums, the impact was loud and clear. I started a spreadsheet to chart the ratio of people complaining of traffic losses to those reporting no impact, and gave up after discovering approximately 40 unhappy souls:

Hubpages forum members reporting impact from Panda Update:


Ouch. Many complainants reported drops of 40% to 50% of traffic, which fits the Quantcast numbers (but isn't as dire as the Searchmetric/Sistrix numbers might lead one to expect.)

SquidU forums were fairly quiet in response to both Panda updates; most comments were by members impacted by traffic loss on Hubpages. There were few reports of traffic impact following Feb 24 and Apr 11, apart from one experienced member's traffic losses, and one who lost traffic, then got hers back.

On April 26-28, SquidU had a handful of reports of keyword visibility losses and traffic spikes for some individual members, but the timing suggests they're not due to Panda. Luckily, they lasted only a day or two.

(Incidentally, this article got its first Google traffic within 6 hours of publication. Of course, new content often gets a boost from Google and then drops in rankings.)

Watch out. Google is going social, refining search results based on user behavior and feedback. "Personal blocklists" are a ranking factor in the Apr. 11 Panda update. Focus group responses were used to refine the Panda algorithm (and by the way, Google may have given us the checklist used by that focus group, at least in part). In August 2010, Google was granted a new patent on using user "click information", which may or may not be part of Panda, too. And yes, Google refines its algorithm by looking at what users report as spam.

What's next? Google has pre-launched Google +1 as an experimental option for Google users. Google +1 will be the equivalent of Diggs and Facebook likes: user votes for a site will be listed right in search results. They will influence which links people click. Based on what Google has done with blocklist data, I think +1 data may eventually be incorporated (probably as only a minor factor) into Google's search algorithm.

We know -- and Google knows -- that SEO black hats use social media votes like Digg and Likes to ding their rivals and promote themselves. Google has surely considered how to head off people gaming the system, and just as surely, people will find ways.

Like it or not, Google is moving towards an algorithm that takes user behavior into account, with the goal of serving up results its users like. We know some ways Panda is doing it (blocklists), and I'm betting there are other user behaviors that Panda takes into account.

Michael Martinez has good reasons for arguing that looking at bounce rate data is mostly a waste of time. Still, I can't help thinking that even if Alexa is failing to capture a complete picture of bounce rates, it's probably over- or underestimating bounce rates on different sites in about the same way. I checked bounce rates on Alexa right after the first Panda update, and couldn't help noticing:



Bounce rates show how often people hit the "back" button after visiting a page. Are people backing away from Hubpages more often than Squidoo? If so, why?

I have a hunch. If I'm right, it may explain why Squidoo escaped Panda.

Squidoo's got an in-house "lensrank algorithm" that ranks all articles (called lenses) according to a combination of factors: traffic, clickouts (clicks on links), ratings by other Squidoo members, sales, and participation in interactive features like polls and quizzes. Notice how many of those are related to user behavior, NOT traffic. Squidoo's tiered payouts are tied to lensrank. In other words, Squidoo allows you to write on almost anything, but it only PAYS articles that engage users. Contrast that with true content farms, which incentivize drawing traffic, or Hubpages, where pay depends solely on ad clicks and affiliate sales. Google says it's looking for high quality sites, but one of the ways it measures quality is to measure user behavior -- the very factor on which Squidoo payouts are based.

Hubpages does try to encourage "quality" pages, via "Hubscore" and the current writing contest being advertised all over the site. However, sadly, good writing as measured by a panel of judges (even, or especially, university-trained English professors) will not necessarily match up with what web surfers find readable, useful, entertaining and/or clickable.

What to do about it: Whether you're on Hubpages, Squidoo or Planet X, you can't just write Pulitzer prizewinning prose. You've got to give people something useful. Preferably, give them something to click, interact with, or share with others ("you've gotta read this article!")

Unless you're a vanity press, respond to your readers' wants by responding to what search phrases brought them to your article. I often review the keywords (search queries) in my traffic stats and ask myself: did they find what whey were looking for?

1. Yes? Great! Onto the next query.
   
2. Somewhat? Maybe I could address the query better. Improve the page.
   
3. No, but I really can't add that to this page. Great! I'll write a new article.
   
4. No, and that had nothing to do with my topic. Uh oh, search engine is confused. Identify and rephrase whatever language the search engine misunderstood. People looking for XY will probably not be satisfied by an article on XZ.

User satisfaction reduces bounce rate, which Panda may or may not notice; it will also reduce your chances of landing on a personal blocklist, which Panda DOES notice.

All Google's announcements about Panda emphasize that it's trying to cull sites with lots of "shallow," spammy or "thin" content from search results. SIgnificantly, this penalty applies to a whole site, based on how many of its pages Google considers low quality.

[Update May 6: Google has now given us clues about what it considers high quality].

In my first attempts to answer "Why did Panda hit Hubpages worse than Squidoo?" I speculated on my Squidoo blog that Squidoo's Squiddont Policy and filters banning spammy topics may have helped: these filters were, after all, designed to get Squidoo back in Google's good graces after the Squidoo slap of '07 -- and and they worked.

But doesn't Hubpages crack down on spam, too? Yes, but perhaps it wasn't quite as hyper-vigilant as Squidoo, since it hadn't been burned yet. Jennifer Ledbetter of One Week Marketing did a mini study of spam on several article-submission sites suggesting that Hubpages had a little more spam than Squidoo, and Panda decided to send it a little less traffic than Squidoo.

What to do about it:

• Don't write on spam-prone topics (obviously) like gambling, ab workouts or those topics listed on Squiddont. Hubpages hasn't banned them outright, but as of Apr 28 will be holding hubs on spam-prone topics to a "higher standard."
  
  
• If you write on a user-submitted content site, search on Google for site:hubpages.com [your keyword or topic] and/or site:squidoo.com [your keyword or topic] without the brackets. If there are dozens of spammy, low-quality articles on your topic, it might be one that Google considers spam. Before, you might have thought, "Oh, good, then I can outrank them!" But Panda may not be able to tell the difference between your well-researched article on herbal supplements and yet another acai berry ad.
  
  
• If you write on a user-submitted content site, report spam to the host site when you see it. (Use the "Report Abuse" button at the bottom of Squidoo pages, "Flag" on Hubpages.)
  
  
• Remember that you don't get to decide what's spam; Google and the web do. If they think it's spam, you lose.'
  
  
• Read and heed Google's guidelines on high quality content vs. spam.

Several of the Panda-related Google announcements and interviews have mentioned that Panda penalizes sites, not just individual webpages, with lots of content copied from elsewhere.

Google doesn't care whether you have the right to repost content (e.g. Wikipedia content, which is Creative Commons), or whether it's your own content posted on several sites. It's copied content, and if you post it over and over on many article submission sites, it's spam.

As Squidoo's Original vs. Duplicate Content FAQ notes, you're probably safe promoting an article by sharing an excerpt with a link back to the original article: this is called "push" promotion. Just be careful; don't repost large chunks or full-length articles. Hubpages disagrees: in response to Panda, it is banning ANY duplicate content whatsoever.

There are also three subtle forms of copied content which Panda is probably penalizing:

• Scraped content, content taken from other websites and sometimes mishmashed together to disguise it. Don't do it.
  
  
• Syndicated content. Many sites make videos, photos, headlines or other kinds of content available through "feeds" or embedding. You can enhance your article by supplementing it with this content. But lazy article writers are using this shortcut as their only content. Both Squidoo and Hubpages had tools that help pull in this kind of content, and both sites have had to remove most of these tools due to abuses (e.g. the Wikipedia module on Squidoo, or the news capsule just dropped from Hubpages).
  
  
• Similarly, many sites pull in API data, raw data that the receiving site displays however it wishes. Examples of API data include Amazon product prices and descriptions, movie theaters' showtimes, or temperatures from Weather.com. Most of this content is noindexed on Squidoo; in January it made the default Amazon module noindexed as well. That means the content isn't crawled by search engines. (The Amazon Spotlight module is still crawled, since it's paired with a mini-review you write yourself). Hubpages has now implemented a 50 words per Amazon/eBay product rule to offset this copied content.

The duplicate/copied content penalty is, I think, a huge part of the Panda update; it's certainly something that's been mentioned by Google on several occasions. No wonder both Squidoo and Hubpages are being paranoid about it.

What to do about it:

Read and follow duplicate content policies on any site where you publish. Here's Squidoo's; Hubpages has taken the simple approach and banned all duplicate content.

Don't repost articles. Do not buy or use distributed content, aka PLR. Don't even repost your own full-length articles on multiple sites.

Don't piece together an article by summarizing, spinning, scraping or paraphrasing information you found elsewhere on the web. Algorithms are learning to detect even rephrased content. Instead, add something of value that can't be found on other sites. If you want to rank well for quality content, you have to provide some.

Help your site stay in (or return to) Google's good graces by reporting copied content. On Squidoo, use the "Report Abuse" button at the bottom of each page. On Hubpages, it's the "Flag" button.

Protect yourself against plagiarism. If an article experiences a sudden traffic drop, or for any important artlcle, Google a sentence or two from it enclosed in quotes to see if it's been reposted anywhere else. If you've been plagiarized, get the copied content taken down.

Make sure you add original, useful content if you are featuring content supplied or syndicated from other sites, e.g. Amazon Associates. I'm not much of a marketer, but my product review articles sharing my own photos and experiences about products that solved a problem for me continue to get visits and sales post-Panda.

Google spokespundit Matt Cutts frequently uses excessive keyword repetition in his discussions of spam, and the Panda announcement mentioned repeated spammy keywords. We don't know how much is too much, but that sounds like Panda is getting more picky about keyword stuffing than Google was already.

Even before Panda, I once managed to get a page dropped right out of Google's index by including the phrase "funny sign" as part of the alt-name for every image on the page.

What to do about it:

• Take Google seriously when it says you should "use keywords appropriately and in context." If it doesn't flow smoothly for your readers, rewrite.
  
  
• You only need keywords in a few places on your page. Really. Title and/or URL, a header or two, an image name, and where it falls naturally within your discussion -- that's more than enough.
  
  
• For those who are comfortable with basic SEO, try using related terms and synonyms. Again, use language naturally; write for readers first, search engines second.

The SEO industry obsesses about "Pagerank," a Google concept based largely on how many pages link to a page versus how many links point away from it. Pagerank is just one of the 200+ factors that the Google algorithm considers, yet it's triggered a huge industry of people spamming the web with links pointing to their site. It also has created paranoia about outbound links, links pointing away from a page.

Avoiding linking out because your Pagerank is going to "bleed away" is silly. The problem is not outbound links, but links pointing to spam and irrelevant, useless content. Pointing to useful, relevant content can actually boost a page's search relevance: "In the same way that Google trusts sites less when they link to spammy sites or bad neighborhoods, parts of our system encourage links to good sites." -- Matt Cutts

Unfortunately, Hubpages seems to throw the baby out with the bathwater: until you get your "Hubber score" up by participating a lot in the community, outbound links are nofollowed. I've also found them very strict on locking educational articles as "overly promotional" if they include even one link to your own topic-relevant blog or site.

What to do about it:

• The measure of whether or not to include a link should depend on whether it benefits your readers, not arbitrary quotas. Or, sometimes, a link may serve as attribution, the online equivalent of a footnote.

Of course, if you're on Hubpages or Squidoo, you have to abide by their outbound link policies set up to thwart those who don't follow this practice.

A huge firestorm erupted after the Panda Update, because many sites with traffic losses had Adsense. I've seen alarmists claming that Adsense and affiliate links kill Google rankings.

If that's so, Squidoo should have sunk like a rock. I was, frankly, appalled when Squidoo added a third line of text Adsense up at the top early in 2011: they've practically pushed all our content below the fold. But Google didn't blink...at least, not yet. Here's another person for whom "above the fold" advertising was NOT penalized by Panda.

Too much advertising will result in more users hitting the "Block site" option on Google, which Panda will notice. I wouldn't be surprised if Panda measures the amount and/or type of advertising on a page in some other fashion, too. [UPDATE May 6: Yes, advertising is one of the factors Panda considers]. When people complain about web spam, they're often complaining about advertising and, yes, affiliate marketing. But Squidoo's experience shows that the Panda Update did not penalize advertising as broadly or universally as some people are claiming... at least, not yet.

Post-Panda, Hubpages has banned affiliate links to many programs it fears Google may consider "spammy". Squidoo already had an "overly promotional affiliate links" policy put in place after the Google slap of '07. In December, Squidoo limited the number of Amazon and eBay modules per article. Time will tell how these approaches play out with Google, especially as it introduces further algorithm tweaks.

What to do about it:

• Familiarize yourself with the affiliate marketing rules for Squidoo and Hubpages (see links above) if you write for those sites. I think it might be a good idea to avoid the programs that Hubpages has banned.
  
  
• Use rel=nofollow on affiliate links, if you're able to edit the code yourself. (For example, edit the code given you by Amazon or Allposters). Here's how, and here's why.
  
  
• Don't do affiliate marketing for junky spam products you wouldn't recommend or buy for a loved one. Find products you can truly recommend.

Hubpages.com's Front Door:

Squidoo.com's Front Door:

When I started on Squidoo, it had a boring white background with black text, no skin. No cartoon fonts. No monsters. No graphical ads. Sharp, neat, dignified. A lot like Hubpages. I want to marry Hubpages' interface designer, who employs clean, organized visual design.

But Google doesn't care about visuals. Google cares about content. In particular, Google cares about content aimed at your users. And also, site navigation. What things are you choosing to highlight on your site's front door? How can visitors to your site find content? Admittedly, most won't be coming in through the front door; they arrive through search. But search engine spiders come in through the front door and crawl their way down from it.

What do visitors to the front door of Squidoo find? A Topics directory, with the very best five articles in each topic right on the front page and a "More" link leading to more of the best articles in each topic, which is divided into subtopics.

What do visitors to the front door of Hubpages find? A few miscellaneous showcase articles (personally, I'd rather look at Queen Elizabeth than a Squidoo monster), a slogan telling visitors they can make money by writing on Hubpages, a panel featuring Hubpages members whom visitors don't care about, and a lot of news posts from Hubpages trying to argue that it's not a content farm. But WHERE IS THE CONTENT?! What if I'm Googlebot looking for the most important content on Hubpages? Could it be those featured articles? Or maybe those community members? Or, I know, the news blog posts about Hubpages!

Oh, ah. There is a topics directory on Hubpage's front page. Unfortunately, it's down below the fold on a standard laptop screen (indicated with a purple line). Yes, Google cares about "above the fold" real estate.

Notice how most of the material on the front page of Hubpages is oriented towards writers of content, not consumers of content.

After some exploring, I found the "Topics" link at top pointing to a bona-fide topics directory. Maybe that's sufficient for site navigation as far as search engines are concerned. However, the meat-and-potatoes of the front door of Hubpages isn't designed for finding content; it's a self-congratulatory set of "all about us" info.

I notice, too, that if you use Hubpages search box, results are listed in the following order: Adsense advertising, Hubpage members, forum discussions, and finally, articles (as if they're the least important content). Whereas Squidoo just gives you a list of related articles.

So that's the top level of site navigation. Now let's click on one of the "Featured Topics" links hidden down below the fold on Hubpages and a Topic header on Squidoo. Say, Autos/Cars, since both sites have that category:

Hubpages Topic Page:

Squidoo Topic Page:

Has Hubpages gotten hacked by some writing contest site? No. This is on purpose. Most pages in Hubpage's topic directory have a few random articles relevant to the topic in the sidebar, but the bulk of the page consists of: a list of subtopics pages like this one (navigation without destinations), a banner for a writing contest, a panel of judges, contest rules and a linked list of daily winners going back to March 31st. Presumably, before that, these topics pages featured articles on that topic, or other content relevant to the topic. But right now, just when Hubpages is suffering from a Google algorithm gut-punch, they've replaced site structure and topic-relevant navigation with something of interest to members, not visitors.

Whereas Squidoo lists the best-rated articles in each topic, a navigation menu for subtopics, a "more" link pointing to the top 100 lenses in each topic, and discreet arrows letting users (but not search engines) browse through and see all the articles in that topic. In other words, the topics pages make site navigation clear, and one can step off the topics tree at any point straight into relevant content. Prominent members in some topics get some love as well, but they're below the fold. Recent articles in the the topic also get a temporary place at left, giving them a chance to be crawled and/or found and rated.

I think Hubpages' directory is normally closer to Squidoo's as far as showing relevant articles (and it does have the rotating "featured hubs" in the sidebar), but even when they remove all that contest-related info, they've still got an awful lot of links to stuff chiefly of interest to members, not visitors.

How about actual Hubs/Lenses (site jargon for "articles")?

Hubpages article on Nile Crocodile by laurentmikhael:

Squidoo article on California Sea Hare by moi:

Here, Hubpages actually comes out a little ahead by doing something interesting: smaller font to get more content "above the fold." Squidoo scared me earlier this year by adding a third line of Google Adsense above the fold, pushing our content down; I often yammer about how much "above the fold" screen real estate Squidoo hogs with its ginormous header.

But navigation-wise, they're similar: "related articles" in the sidebar, and below the fold, a "related tags" box linking to tag pages that hopefully link back to related articles. Squidoo also adds an optional table of contents just below the fold (click the link to see it) which organizes the page's content for visitors and search engines. Google includes these TOC links in some search results, and I suspect the Google algorithm uses Table of Contents links (if it recognizes them) as a very minor ranking signal for the most important keywords on a page.

One other HUGE thing. I was going to use a Hub with sea hares as a comparison for my sea hare page, but the "above the fold" content of the article is just a block of photos the width of the article body. Don't do this, Hubbers. Make sure you've got search engine crawlable text above the fold, an introduction to your page, before you launch into pretty photos, videos, or other non-crawlable content.

Squidoo members need to make sure they've given every lens a relevant topic, subtopic, and (if possible) sub-subtopic, to get links to their articles from the topics directory linked to from Squidoo's front door.

Squidoo members should also learn about tags and use tags wisely to build cross-links between similar, excellent lenses (yours, or someone else's).

Hubpages users need to build more links pointing to their Hubs from relevant off-site pages (e.g. your own blog posts), and link between hubs on similar topics. Choose relevant tags used by other good, relevant hubs so that they'll appear in the "explore related" box.

I may be barking up the wrong tree here, but I think Hubpages would do well to consider making navigation for users more front-and-center from the front door all the way down.