I think it is well accepted that an online presence for early career researchers is a good thing, even though a minority actually have one. As an example, blogs on my website have been reposted a few times on various news sites, and they also appear high up the search engine list for keywords relating to my field of research. I also managed to get invited to speak on the back of my website, so in my experience a website really does improve your visibility.
The first challenge with blogs is deciding who they are aimed at. Remembering back to my early PhD, understanding papers was hard, and so I decided that I would focus on researchers in my field (magnetism) but try and avoid all unnecessary jargon, which should also make it accessible to undergraduates in the discipline. If you are blogging about your publications, then it means you can simplify the result for a broader audience, without compromising on the detail (those who are interested can look up the paper on your site - don't make them hunt for it or have it stuck behind a pay wall).
My main tips for an academic blog are:
Talk about all your publications - what is the main result in laymans terms and specify why it is important.
Don't worry too much about length limits. Academics are used to reading long documents and a blog a couple of pages long doesn't take long to skim. However, research is generally technical and so it is difficult to write a superficial post such as 'I had omelette for tea' or some such.
A blog is as much about you as the subject, so write in the first person and include your wider activities. For example, I always write up conference visits and talks I have attended/done.
Pictures and figures are a lot more effective than words, so use these extensively if you can.