Ok, since Panda, there has been a lot of talk about the importance of branding. There is no clear definition of "brand" when it comes to websites. Therefore, Google must set a definition in their algorithms and rank pages accordingly.
Until now, I embraced (or I should say, tolerated) negative reviews (the very few I have received) because of the links they provided. However, I am rethinking this after my experience with Google's semantic detection this week
My site is suddenly receiving traffic and number 1 rankings for phrases that contain obscenities and slurs. Yes, people are actually appending slurs to their how-to queries. My site doesn't use this language, so it led me to look at my backlinks. Interestingly, I found ONE site that has been linking to me for two years (in a negative review of my site which I tolerated up to this point), but they have a recent blog post that contains the obscenities and slurs.
I have tested all sorts of phrase combinations using the words found on their page, and I rank #1 for all of them. My link is NOT used within context on that blog post...my links are actually on another page of their site (again from a negative review two years ago). So it seems the semantic detection can utilize words found ANYWHERE on the linking site, particularly if it is a recent article/post with homepage exposure.
So this leads to two questions in my mind:
1) Should I be concerned about the #1 rankings for these obscenities? I fear they could harm my business reputation, irrespective of the fact that it is the fault of Google's semantic detection. I personally think Google should filter slurs/obscenities in semantic detection if the words are not found on the destination page. They have filtered those words in auto-complete so they can apply them to the semantic detection/rankings.
2) If Google can apply those obscenities to my site via semantic detection, perhaps they can also detect the mood of the review that links to you. Clearly if a site has enough juice to push me to the #1 spot for phrases that contain obscenities, the detectable mood of their review could be doing harm.
Semantic detection appears to be the new link bowling.
Food for thought, and would love feedback.