My Newsletter Finally Looks Like a Real Email

Last week I wrote about my newsletter arriving in Kit as one giant paragraph. That got fixed. Paragraphs stayed paragraphs, lists stayed lists, the link at the bottom actually worked.

I opened the next draft expecting to be done with it, and I wasn't. The text was right. It still didn't look like an email.

What was still missing

There was no picture. No visual break between the intro and the meat of it. The line that's supposed to send people to the episode was still just a sentence with a link buried in it, not something your eye catches. It read fine. It looked like a note, not like something a show would actually send.

The fix from last time made the plumbing correct. It didn't make the thing look designed.

What shipped

The draft that lands in Kit now looks like an actual newsletter. The episode's artwork sits at the top as a real image, and it's linked, so clicking the picture takes you straight to the episode. The line about listening is now a proper button instead of a sentence you have to notice. Section breaks that used to just be an orphaned line of text now render as real headings. And there's a clean line above the sign-off instead of it running straight into the last sentence of the email.

None of this is invented. I checked it the only way that means anything: I sent a real draft into my own connected Kit account and opened it there. The image showed up as the image. The heading showed up as a heading. The button worked. Kit even picked up the episode artwork on its own and used it as the broadcast's thumbnail, which I didn't ask for and was a nice surprise.

The thing I found while I was in there

While I was going block by block through what actually reaches the reader, I went and reread my own newsletter copy closely for the first time in a while. And there it was. An em dash. I don't write with em dashes. I have a rule against them, sitting right there in my settings.

Turns out the rule was sitting in the settings and nowhere else. It got typed into the instructions we hand the writing model, and then everyone just trusted the model to listen. Mostly it did. Sometimes it didn't, and nothing was there to catch it after the fact. Same story for the small list of words I don't want showing up in my own writing. The setting existed. The enforcement didn't.

That's now fixed the boring, reliable way. Every em dash in generated copy gets swapped for a comma before it's saved, no exceptions, no depending on the model's mood that day. And every piece of copy gets checked against my banned-words list before it goes out, flagged if something slips through, instead of hoping the instructions were enough.

What's next

I still want to go further on the writing checks, into the shorter, punchier formats like carousels and social clips, which don't run through this yet. And I want to actually see the flagged words somewhere in the app instead of just knowing, in the back of my mind, that the check ran.

If you're on Kit, the newsletter PodGlue writes for your episode now shows up looking like something you'd actually send, not something you'd have to dress up first. It still never sends on its own. That part is still yours.