This is probably not an ideal response, and it would definitely be better to just have inbuilt capabilities for things like Word Count, but I have got a workaround that’s at least mostly accurate (and could probably be tweaked to make it more accurate if needed).
Basically, I have a Blog CMS in Coda: a table which contains columns for title, body text (this column is hidden to prevent rows being gigantic), header image, summary, SEO tags etc. I then use a customised Layout in a separate Section as an editor (I actually have three Sections with similar layouts, one for Drafts, one for Editor Sign Off, and one for Published, each using filters).
That layout is designed to look as much like a classical CMS back-end as possible, but lacked a Word Count, so I added a new column to the main table and setup the following formula:
Length(RegexReplace([Article Text],"([^\s])",""))+1
“Article Text” is the name of my column that contains the main body of the blog, but otherwise it should be copy/pastable. All it does is reduce your blog’s text block to only whitespace (spaces, line breaks, tabs) and then count the characters. That gives you the number of “gaps” between words or paragraphs; the +1 is to account for the very last word, which will likely not have any space after it.
The results aren’t perfect, as some characters like en-dashes (i.e. “in the middle of text – having an aside – gives odd results”) which typically have spaces around them get counted as words but, as I said, if you need it to be perfect you could keep tweaking the RegExp to improve it 