The Google Sandbox Revisited
13 October 2006Greg from Bloodhound Blog has a post on his experience launching a new site promoting an individual home. He’s skeptical about whether or not the “Google Sandbox” exists - but in a more recent post suggests that his early results with that site may have disappeared. This will probably only be of interest to anyone who’s got a blog or a website, but here’s my view on what is going on:
The Sandbox is a muddled concept used to describe the phenomenon where a new web site seems to perform poorly in Google’s rankings for some period of time. Older web sites do not seem to be affected by it. The reason it’s muddled is that a lot of people have proposed a lot of theories for what’s going on, ranging from a flat time limit (your site must be x months old to be ranked well) to a “trust” theory that you have to be linked from important web sites if your site is new. The closest anyone from Google has come to describing / admitting it is here from Matt Cutts, a Google engineer:
Q: Does the sandbox exist?
A: Matt said here comes the audience part? How many feel there is a sandbox? How many feel there is no such thing as a sandbox? SEOs normally split down the line. There are some things in the algorithm that may be perceived as a sandbox that doesn’t apply to all industries. He knows it works to keep some spam out.
He’s also made other statements along the lines that the Sandbox is something that was unplanned - people started reporting it to Google after an algorithm change, and they liked what they saw, so they kept tinkering with how the search engine results are computed to encourage it. This leads to the more current view of the Sandbox: that there are a set of unrelated “filters” applied to websites based on various factors, which can negatively affect the way that they rank. Age may well be one of these, as well as the kind of links the site receives and the places the site links to.
I’m fairly convinced that new sites suffer from a penalty. In fact, I believe this site is sandboxed right now - even though it’s getting about 1,000 visitors a month from Google. How could that possibly be?
Part of it is my experience with other sites. The other part is the key phrase from Matt Cutts - that it doesn’t apply to all industries. Here’s an example: I promoted another site, targeting three search terms with the same method. It doesn’t work anymore with Google, but it did at the time. The site was brand new. On two of the terms, it ranked virtually immediately. One was a wholly noncommercial phrase, the other was “widget discussion.” I was in the top three for both. The other phrase? “Widget forums.” For that one, I wasn’t in the top 1,000 search results - despite the fact that the site was promoted in the exact same way for all three terms, and got many visitors from the other two. My guess is that it’s the word “forums” that triggered the filter - something about it is considered competitive or commercial, and the site wasn’t allowed to rank for it because it was new and “untrusted.”
Well, why do I think Free the Drones is sandboxed? Because I see the same phenomenon. I’ve been randomly searching for the titles of my posts in Google - the exact wording. For some of them, I’m the first result. Makes sense - the title is distinctive and no one else on the Internet has written those exact words. For example, this search from an old post on saving money by spending a little time on the phone. Type in the exact wording of the post title, and the post comes up at #3. Makes sense - the first two sites don’t have the exact words, but they’re close. And they have better PageRank and are probably more important. It sounds about right for where the post should rank. But for other posts, that’s not what happens. Try this search - the exact wording of a post from around the same time on evaluating your student loan packages. What happens? I looked through the top 200 results - and my post doesn’t come up as any of them. For a unique phrase which has been used only on this site (and one spam site referring to it). My guess is that if I browsed through the top 1000 results, I wouldn’t show up there, either.
Why the difference? I think it’s something about the phrases. The Sandbox is an anti-spam tool - so maybe student loans are subject to spam, while telephone calls aren’t. Maybe if I’d used a slightly different word or phrase in each one it would have flipped the results. But the fact is that on some searches, the site is where you’d expect it to be. On others, it doesn’t show up at all - and there’s no good explanation I can think of for why, other than a filter.
I’ll also add a prediction: I think that these filters applied by Google are lifted in 30 day increments. So if you do something to trip a filter, it gets removed after some period of time has passed. I’ve noticed on many sites that this clocks in at exactly the three month mark, or 90 days. My prediction is that by the end of November, this site will have a sudden jump in traffic as the Sandbox, or part of it, gets lifted. It was founded in late July - meaning 90 days after that is late October. But I’m also factoring in the time it took for Google to find the site and the fact that in my experience, it can take a few weeks for Google to reflect the freshest information. Maybe I’ll be wrong - but if it happens, for me it’s pretty good evidence of the Sandbox.
What about Greg’s site? I still think he’d be better off hosting it on his main web site. The reason is this: take a look at this subpage. There’s unique text there about the kitchen - but if you type it into Google, the site doesn’t show up at all. My bet is that if you add up all the pages he’ll end up creating on individual sites, there will be a lot of descriptive text about the houses that are unique to each one. Which means that if someone types in a search about houses in Avondale, AZ with cherrywood cabinets, the same text on a single, older, main site might show up as a result - whereas it won’t with the new one. I’ve got no clue how many people this would really be. I’m just guessing from my own sites, where I get searches from all sorts of random iterations of whatever I wrote. The Bloodhound Realty traffic logs might show something different from what I expect (i.e., that people don’t search much for specific features of the house, etc.). But I do know that those are pretty lucrative searches - people pay a lot to get what a well-done site can get for free, especially in real estate. If you’re starting a web site, your mileage may vary. You may not care about whether people come to your site searching for certain things, because it may not be of value to you. I don’t know exactly how the Sandbox works - no one does - but I’m pretty sure that new sites are nearly always affected to at least some degree.
Discuss this on the Free the Drones Forums.
4 Responses to “The Google Sandbox Revisited”
October 14th, 2006 at 12:00 pm
That is called the Sandbox Phenomenon. Google tries to index a site when it first appears and you can score relatively well with a decent page rank for within the first month while google figures out if it likes the direction of your site. So, you can land in the SERPs and see a good PR. Then you dissappear for the rest of your 3 or so month stay int he sandbox. You can find documentation of this phenomenon on SEOmoz.
check out my article on the sandbox:
http://www.rsspieces.com/2006/10/14/the-google-sandbox
October 14th, 2006 at 12:33 pm
[...] This might confound our friends at Free The Drones, who ferry us across the desert sands to The Google Sandbox Revisited. My take for now: A link to a custom weblog from BloodhoundBlog is probably more findable than building the website as a subfolder on our main website. [...]
November 9th, 2006 at 12:19 am
Hello to all. I am John Pillband, 22 y.o. This is my first day on the Internet. What should I do?
November 30th, 2006 at 3:25 pm
[...] I previously have made several posts here and here about the phenomenon called the “Google Sandbox” - something that many people who launch brand new websites see. In essence, Google blocks new sites from getting any visitors through its search engine until enough various elements of the site show that it can be trusted. A lot of people have a lot of various theories on it, but one of mine is that there is a flat, 90-day “holding cell” period for a new site, after which you can be released to start ranking on commerce related searches targeted by spammers. When I last posted on it in October, I made this prediction: [...]