1 It is easy to install and start.
Of course it is your friend if you are a techie but installing elasticsearch for simple dataviz purpose is very easy.
On debian, ubuntu and maybe Centos you have preinstalled package. You can even dezip the source archive and launch !
2 You can store any csv like or json data.
You can store any json like data natively in elasticsearch. If your source data is csv coding an importer can be done in five minutes in your favourite language.
If your data source is in json its better ! Store it directly. Just one warning: you would maybe have to tweak your elasticsearch model mapping. The default one will work only if your input data is super clean.
3 You can query it easily.
You get results by json. Perfect for d3 or other js libs. Just make ajax like queries and parse the reply. It is easy to code a proxy with your favourite language super fast too. This way you can handle cross domain ajax query. If you are a superstar you can map a subfolder in your domain with ha-proxy.
4 You can make lot of analytics.
Analytics and agregrations is one of the main features of elasticsearch. This is a good news for data visualization ! Plenty of possiblities for histograms, date based heatmaps and more !
5 Query complexity give you a lot of freedom for data representing.
You can update your data blazing fast. An interactive data visualization will be super responsive if you query es to update your data.
You can also filter data in real time, by full text search or facet query.
This is uncomparable to csv !