If a candidate is popular, articles will be written about them, and their SEO will increase. (Yes, this may further increase their popularity, but they were already popular.) If anything, their original popularity is driving their Google rank *and* their likelihood of winning the election. I think the researcher takes two "effects," and says one is actually causing the other.
This study is really only interesting for its focus on "undecided voters," but in many electorates this is a really small sliver of the general electorate, so it's hard to say what exactly is swaying them. I mean, are people really Googling "Who should I vote for?", and then just reading the first few articles and deciding "Okay, this first one sounds good.,,"