Notification in terms of a transactional email wouldn't contain any sort of marketing messaging. You wouldn't be able to "sell" the app, so to speak. Just to let folks know there is an app. But then would that even be valuable? You need to be able to sell the app in order to get folks to download it.
And as much as your business may think your customers adore your brand, you will always need to sell and convince them do download and install an app.
But the email itself would be a promotion of the app, not merely a notification saying that there's a new app available, correct? It's the messaging that defines it as a marketing message.
Imagine the email is sent as a transactional email to all of those 50K customers. For those customers who haven't opted in, they don't have an expectation of receiving such an email. YOU may regard the email as transactional, but that's not how email recipients and subscribers may see it. And that puts you at risk of high unsubscribe rates and potential spam complaints. Not to mention potentially reversing some of the trust your customers have in your brand by sending unsolicited emails.
Of course, that's a worst-case scenario.
A transactional email, to me, is something that's triggered by something the user has done in-app, online, or in-person or as a result of user interaction.
The email you've outlined is about promoting an app to your customers—it's not triggered by an action the customer has taken. That makes it a marketing email.
Alternatively, you could include a small promotion for the app in existing transactional emails, like your receipt or purchase confirmation emails. And when I say small, I mean small promotional banner to give the app a bit of a lift in emails that are more regularly opened.
Wish I could say! I think Gmail relies more on Gmail users' own activity within the inbox. If it sees the majority of its users not interacting with your email, Gmail may move it to the spam folder. I'm not a deliverability expert (not by a long shot!) but that's usually my take on it when it comes to Gmail.
This has happened at Litmus a few years ago. To help us get back into Gmail's inbox, we segmented out our Gmail users and only emailed our most engaged subscribers (opened an email in the last 30 days) for 4 weeks. Once we saw an uptick in their engagement, increased the "last opened" to 90 days, for another 4 weeks. And so on. It was a long process, but it worked.
What percentage of your subscriber base are Gmail users? If you're noticing the drop in open rate was mostly Gmail users and they are a high proportion of your base, it could be that your email is landing in Gmail users' spam folder. That's what I instantly thought looking at everything else you've mentioned here.
At Litmus, we're in the process of getting this setup for our own emails. Working with our TechOps team to get it implemented. We'll share how we did it when it's ready.