How attention flows on the web
A lot of SEO practitioners are touting the notion that social signifiers, i.e. ‘likes’, ‘shares’, etc. are also signals that are picked-up by search engine ranking algorithms. The argument is that these are more reliable indicators of popularity than links as it is likely that a real person is behind each signal. However, actual evidence that these ‘social signals’ cause fluctuations in rankings is non-existent.
I meant to highlight this post this last month as it is excellent - have a read to review the difference between causation and correlation as well as to assess the alternative explanations of what may be going on.
The way I look at this is that I understand the search engine algorithm to be a black box. I can see certain inputs I believe to affect the output but I have a hard time confirming the relative importance of these inputs. Over time, I believe that these signals have become more complicated, that the black box itself includes historical data that it has collected that it uses in its evaluation. I also believe that it is likely that the application of the algorithm is non-uniform - that is to say that certain factors are likely to affect the relative importance of other factors in the results on a case-by-case basis.
‘Social signals’ are by their nature ephemeral. If I was trying to return relevant results then of course I would consider these signals as potentially useful, say in combination with number-of-searches and other factors. However, the effect of these signals would be time-limited as the momentum behind them would be an indicator of whether people’s attention is staying with the topic or moving on to something new. For example, an affect of a decaying signal might be to ‘turn-down’ the similar results displayed as interest wanes. What concerns me is the implication that these signals replace links as a permanent signifier of relevance.