Instagram found a place in our hearts as an app for broadcasting moments. Take a photo (or later a video) and share it publicly, and specifically, to people who follow you. Now Instagram wants us to use it for private sharing. Take a photo or video and send it to one person or a small group. Those are entirely distinct species of communication.
Convincing a userbase to break their ingrained behavior pattern and use an app for something completely different is a tough sell. And it’s a lot tougher if that “something different” is actually “something you can do elsewhere”.
If I want to share a photo with a few friends, I can text it, email it, or Facebook message it. These each let me get friends’ reactions and have a conversation around the photo. In fact, they’re all more flexible than Instagram Direct in that I can reply with another photo — the absence of that feature is my biggest gripe about IDG. It also suffers from a creation interface that’s too slow for sharing to such a limited audience. Filtering and adding a witty caption bog down the flow, making Instagram Direct too time intensive to be a rapid-fire visual communication tool.
And of course, if I want to private message someone a photo or video, I can Snapchat them. Snapchat has carved out a purpose and following with ephemerality — something that’s actually different. I can’t send a photo that disappears with any other major messaging service, so I go to Snapchat when I have something silly or racy to share.
So really, the problem is that Instagram Direct is too different from Instagram, and not different enough from everything else.

Good piece from Josh Constine on a key problem facing Instagram Direct.

Allowing video to be uploaded was a natural extension for Instagram. Instead of broadcasting photos, you were broadcasting video. It felt comfortable right away.

Instagram Direct felt immediately strange. I'd never used Instagram as a one-to-one photo sharing tool, and the people I'd chosen to follow on the service were not ones I'd chosen with one-to-one sharing in mind. My Instagram graph is much smaller than my graph on other social graph services because I'd chosen who to follow based on who I wanted to see photo broadcasts from, and I think most people who follow me there were looking for my photo broadcasts as well. It will always be easier for me to share photos one-to-one through another app because my graphs are larger there and because, as Josh notes, the interaction flow is much faster.

I now use the following multitude of apps to message other people on an almost daily basis: email, Twitter, Twitter DM, Facebook, WhatsApp, Line, Snapchat, iMessage, SMS. One would think using so many different messaging apps would be annoying, that the shape of social graphs would see one of these services winning out through network effects.

But now that all of these messaging apps can easily piggyback off of my mobile contact book to easily find the people I already know on those services, the switching costs are very low. The interfaces are all easy to learn and largely equivalent (the other person's message in a chat bubble on one side of the screen, mine on the other side) so the learning curve is also negligible. Finally, since my phone sends me a notification anytime I receive a message through any of these services, I can launch any of the apps with one click and tap out a reply just as easily as I would on the next app.

For those reasons, It's not clear this has to be a winner-take-all space. That makes it challenging for investors in this area. If some other messaging app came along that was somewhat better and some of my friends flocked to it, I could switch in no time, and if for some reason one of these apps became unfashionable, I could delete it without too much regret.