Forum Moderators: mack
How many pages did you have per subdomain?
There are probably 5 to 25 pages per subdomain.
Well, perhaps high number of subdomains, and low number of pages per domain triggers spam filter (ought to be some exceptions for major sites). Hardly suprising given amount of spam involving keywords in subdomains.
Our subdomains are not based on spam keywords- we have a 'microsite' concept which which spilts content based on geo locations (states, provinces, countries).
Does anyone have evidence to suggest that we may be readmitted if the quantity of subdomains is scaled back?
we have a 'microsite' concept which which spilts content based on geo locations (states, provinces, countries).
No offence but it sounds like local spam to me: any site that has an unusually high number of subdomains is really a major spam suspect. Pretty much only valid exceptions that I've seen are sites that belong to big ISPs who create subdomains for customer's webpages. There are not that many of those, and content of pages on each of subdomains will certainly be very different.
IMO anyone who is using lots of subdomains is really asking for trouble.
I have noticed, however, that it seems it's our generated html pages which are not indexed, where our dynamic (CFM) pages remain in the index. Curious...
I insist, our pages are not spam.
You see, you might just be right, but from what you said on here it seems to me that your site has strong signs of spam:
1) multiple subdomains, probably found as keywords in text, even title of the page?
2) generated html pages: probably very simÃlar to each other, just what % of text is different on each of them?
Bottom line is this: subdomains historically being meant to be used to split data into major areas, like www.example.com for website, ftp.example.com for FTP, alpha.example.com for test area etc.
When search engine comes across your kind of subdomain usage then it is hardly suprising it is marked as spam. They may not have detected that you machine generated your pages, but if they did then it is reasonable to expect algorithm to flag you as spam. Its nothing personal, just an algorithm: from what you have described you are more similar to spam pages than not.
Contains ~30 subdomains, each named for a particular, topical, contributing author. We have between 100-500 completely unique articles written by each author. Each article is generated html. The sub domains are used solely to keep the authors happy (they have nothing to do with search keywords). This is a perfectly legit use of the 'microsite' concept- there, in theory, is nothing inherently 'wrong' with using subdomains.
The site also contains a large amount of topical news (about 50 articles per day). This news is not generated.
MSN is indexing all news content, but not the original, generated article content. The only page similarities are the general navigation structure.
It seems, Lord Majestic, this throws most of your speculation out the window. MSN seems to be discarding all pages which it determines to be generated (something which many serious content publishers are doing today). It's no wonder that MSN's results are so poor...
I'd say 30 subdomains is not high enough to be set as trigger threshold, but 100 is just a nice round number that is temping to be used as treshold for that.
It is likely that spam is judged by a complex formulae that looks at a number of elements, which is why it hard to say which one was the most responsible in any given case.
There certainly is no evidence of penalizing them at this point that I can see. Maybe less "valuing them like gold". At some point a lot of subsomain pages will drop like a rock because their only positive algo ingredient is the keywords in the subdomain (and usually many blog links too). When this happens it won't be a "penalty".
"If I were in charge I'd give them about equal weight to folder names in the URL."
That should go without saying, but clearly MSN has not been doing this. Keywords in subdomain have been far more useful than keyword in folder. I believe they have recently lowered the discrpency a bit, which is a positive step, but more to the point they need to turn way, way, way, way down the value they put on keywords in any part of the URL. Yes, this can show relevance, but it has absolutely zero value as a signal of quality.
