Google’s Sandbox Still Exists: Exemplified by Grader.com

For many in the SEO field, Google’s “sandbox,” a filter the search quality team created to help fight spam, is a relic of days gone by. However, we’ve been spotting new cases over the last few years, and I finally found a great example to share publicly (and got permission from the site owner). Grader.com, and the four subdomains underneath it – Twitter.Grader.com, PressRelease.Grader.com, Website.Grader.com & Facebook.Grader.com – are all under pronounced effects that highlight this algorithmic element’s impact.

How do we know? There’s a few common signals:

• The domain is relatively new (usually less than 1 year old, but sometimes as old as 2)
• Pages on the site are unable to rank for even exact title tag matches
• Other sites/pages outrank it for search queries with clear navigational intent
• A temporal cycle in which the site initially ranks quite competitively for relevant queries, then suddenly drops off 30-500+ ranking positions
• Other Google SEO metrics appear that are typically positive signals (such as assigned PageRank via the toolbar & sitelinks for domain searches)
• The site ranks competitively in Yahoo! & MSN/Live
• There’s no indication of spam or manipulative tactics that could cause the rankings issues

Let’s check out some searches for these and see how they stack up:

• “Twitter Grader” – #55 at Google, #1 at Yahoo!, #2 at MSN/Live
• “Twitter Grader | Get Your Twitter Rankings” – #74 at Google, #1 at Yahoo!, #1 at MSN/Live
• “Twitter Elite | The Most Powerful Twitter Users” – #78 at Google, #1 at Yahoo!, #1 at MSN/Live
• “Websitegrader Website Marketing SEO Tool” – #Not in Top 200 at Google, #1 at Yahoo!, #1 at MSN/Live

Clearly, Google is penalizing or filtering this site in some fashion. They’ve got the pages in the index, but they can’t rank for exact title tag matches while the other engines are showing them consistently in top positions. The homepage is a PageRank 5 (according to the toolbar) and shows ~450 unique domains linking to it. It’s clearly a well regarded site that’s earned natural, editorial links for providing valuable content, but Google’s “sandbox’ algorithm is restricting the site to search obscurity. All the subdomains on grader.com perform similarly – website.grader.com, pressrelease.grader.com & facebook.grader.com, suggesting this penalty is tied to the root domain, not any single subdomain.

The big frustration for site owners is how to escape this purgatory. As with most penalties from search engines, there’s no clear path to resolving the issue – I’ve talked to Dharmesh Shah, who runs Hubspot (which owns grader.com), and there’s been no messaging in Google’s Webmaster Tools. Grader.com also experienced the common “sandbox cycle” of initially ranking, then losing their search traffic and seeing even exact match queries show the site in positions 50+.

I don’t have any foolproof methodology to work your way out of the box, but we have noticed (through many painful campaigns for clients and Q+A with PRO members) that some common themes emerge:

1. You never “pop out” alone – it seems that Google has certain internally triggered events where a bundle of sites suffering from this issue all “emerge” to their expected rankings on the same day. You’ll sometimes see forum threads and chatter about these “sandbox releases.”
2. Earning higher quantities of links, particularly from trusted sources, seems to be a common path to emergence, though grader.com certainly has its share of excellent links from places like C|Net, MSDN, Mashable, ReadWriteWeb & more.
3. It seems you can lengthen your stay in the box by exclusively attracting the more typical “low quality” links that signal manual link building campaigns – such as built-for-SEO directories, article submission sites, reciprocal links, dofollow blog comments, forum signature links, etc. (this is speculation on my part, and more correlation than causation, IMO).
4. Re-consideration requests through Google’s Webmaster Tools appear to sometimes (but rarely) help, and it’s always hard to say if the request itself had any action above and beyond what would have happened normally (as the communication system provides no feedback).

In the SEO world, the speculation is that Google created this filter to combat the rise of newer sites spamming the index, and while it’s certainly been effective for that (think back to how new, very low quality sites could rank from 2001-2004), there’s a case of throwing out the baby with the bathwater. The best we can do is be aware of the issue, mindful of how it might affect websites launches and be tuned in to its evolution for signs of how to avoid and fix. As always, I’d love your feedback about how the sandbox has affected your sites (or those of your clients) and any tactics you’ve used to successfully escape.

p.s. On a personal note, the “sandbox” was really what brought me into the SEO community more formally and in a participatory role. In 2004 & 2005, while fighting against it for two of our clients, I began blogging, creating tools and posting in forums hoping to figure out the problem. I fought for a long time to gain any credibility (as most established SEOs did not believe in its existence until later on), but owe a debt of thanks to Aaron D’Souza from Google who, during a meeting at my first SES San Jose conference, confirmed its existence and effect (though he was naturally coy about revealing any particular details). He also indicated that inside Google, it had a different name (though I still don’t know what that was/is). I still feel a great thankfulness whenever I think about Aaron and that event, as it completely changed my mentality & focus (from proving its existence to actually finding ways out).

p.p.s. Update in October 2009: Grader.com’s content was moved to WebsiteGrader.com and they suddenly rank normally for all their content. Though others may still disagree, I’m tempted to say that’s the final nail in the “does the sandbox exist” coffin. Whatever penalty/filter was on Grader.com doesn’t apply to the content or the links, just the domain. Moving content to an un-boxed site has solved this issue time and again (Grader.com is just the latest public example).