Tuesday, August 16, 2016

Are spam sites using variate testing tool (mbox) more than other sites?

Mboxes are such an interesting topic, and the spam list I used last time was good, but I was not sure how representative. So my next two lists of spam domains were generated using Moz' opensiteexplorer. This tool has a feature where it categorizes domains linking to other sites by up to 18 parameters indicative of spam.

Spam domains linking to large sites using mbox

I used a few profiles (root domains like dell.com) from our industry first, and pulled all inbound linking subdomains with the highest spam scores. For most sites, I had to use scores 8 and greater (not 18 or 15 or 12), because the link profiles were actually pretty clean for all large sites checked.

On a total of 4544 linking 'spam' domains I found 196 mbox implementations on the homepage, 4.3 %. This is completely in line with the first set of spam domains last time (4.4%).

Spam domains linking to blogging platforms using mbox

Now the second set of spam domains are from the same tool, but this time I looked at the link profiles for the root domains blogger.com, blogspot.com, tumblr.com and wordpress.com, in the hope to catch some bad linkspam from some of the bloggers. But I was surprised how relatively clean these were again. Not as good as the large site link domains, though. Perhaps due to the large number of overall inbound links - the samples are much larger and I could focus on domains with a spam score of 13 and higher.

On a total of 2336 domains I found 99 mbox implementations on the homepage, 4.2 %. This is completely in line with the first set of spam domains last time (4.4%) and the other spam list.

(First I ran this with 5 sec timeout and discovered only 87 sites with mbox. 5 sec for all files would be nice, but perhaps not realistic, so with a timeout threshold of 30 sec more mboxes showed up.)

All in all this shows that mboxes (for variate testing) are used on all kinds of sites and are not an indicator for a spam-like site - but not the opposite either.
Bookmark and Share