I've been thinking recently more over some ranking signals that I released (perhaps foolish) in the past. Some of these have disavowed the engines before the attention or discussion may earn while others don't get. My list includes:
Mentions a domain / brand name - especially into which the engine has classified as "News." I suspect we would find that a reasonable correlation and probably plenty of examples of domains that begin, as soon as you earn this mentions ranking.Nofollow links from trusted sources - could who see by running a little analysis of across domains on the net, engines, simply too good sites/domains and with what degree of consistency links. From there it is just a simple step "count" these links nofollow attribute as follows or treat you similar to that mentioned above. This metric Gets a lot of attention and our correlation data, at least, suggests that a large number of links/link root domains with no follows to better correlate rankings.LinkedIn + Twitter profile links - since these sites (and probably others like you) will not make used primarily by real people, most of which may have seen a spammy site from potential employers Networkers, these links are used for search engine probably Golden.Traffic patterns through aggregated Google Analytics data - if a list of domains that receive sent the search quality team get / transport and the relative quantity levels, I suspect you could put this sort as a methodology to use spam from real pages (spam tends to not send traffic, yet it received from a variety of good sites). It would also be an incredibly hard metric to game - how pull down much referral traffic from many unique high value sites (direct - most would show filtered get) without really be interesting and visiting value?Mobile visits, check-in the and interaction-although difficult to identify or track down compared to some other metrics, I think that a local or a relevant site gets only clicks and interactivity of mobile browsers, if it is highly relevant and useful. This could be a more solid way to filter spam and to obtain data types for local / maps of rankings (provided that the engines had can tell access to data on the scale... Android Windows Mobile?):-) Links and references in Gmail - again, it is unlikely that Google's actually reading our email, but certainly could the search quality team get a list of the number and diversity of Web site references used e-Mail (much the same way Gmail provides "personalized" show based on the content of the e-Mails) content that reaps comments/UGC - if real people are actively involved on a site around unique content, I'd wager to guess that the content is probably the type engines rank would want. Things like comment of RSS-feeds, trackbacks and content uniqueness maintains analysis could be used all around to help sort.Rich media present on the ground and around the Web - not make a lot of unique graphics, pictures and photos spammers. Also, you not original video film dont post podcasts, Flash build items, upload, Excel spreadsheets, graphically heavy PDFs or similar. Run by real people and companies do real websites and businesses. Because the engines already indexing and segmentation ability, there's nothing, to you stop from examining the data as a signal quality.I'm not saying that Google Bing use them anyway, but I would guess that all of you have practical applications to improve search quality and relevance. And, by correlations and analyzing these pairs of numbers run us (where possible), we can may learn more about that a Web site, the power of "natural look", what and rank-worthy to engines, in particular, because so much of my email and our F + seems to be concerned A to late false alarms.
I'm curious about all other factors, you think, fit this pattern/system?
No comments:
Post a Comment