[volument]

This is good! I have searched for this actually.

However this seems to lack the most important thing (for me): the global usage percentage of a particular feature, which is the #1 deciding factor whether to use a HTML/CSS feature or not.

[longwave]

I think that would be difficult to calculate accurately. We can collect user-agent statistics from browsers but mail clients don't have an equivalent when reading mail.

[volument]

A rough estimate would satisfy me at least. Seems Litmus has farily broad statistics from 823 million opened emails: https://emailclientmarketshare.com/ (August 2019)

[yyyk]

It's probably strongly tilted towards email clients which don't block external images by default, and email clients used by people who are more likely to be targets of a marketing campaign (i.e. richer 1st world people).

[u801e]

A number of email clients will add a header line that identifies them to every email that's sent.

[jcranmer]

The problem is that you don't care about email clients as distributed by senders, you care about the distribution of recipients. Especially if you're a marketer trying to gauge what kind of HTML you can use in your email.

The only people who can really use the headers to identify which email clients are in use are people who maintain very large ISPs. Although, even then, keeping track of the results of the IMAP ID command are probably more useful.