Developers October 21, 2020

Should I create in-house chat for my startup?


I built a platform in the past with an in-house chat of my own implementation (django channels), unfortunately the platform didn't take off, so I never encountered any scaling issues.

What I found out though is that concurrent websocket connections scale differently from a regular http server and there are a lot of traps one could fall in a chat application. So I did a little digging and found out there are a lot of saas which handle chatting but the caveat is they are too expensive, like $250/month for startups.

So my question is, when you start a startup do you create your own implementation in-house or do you use a service like sendbird, stream etc.?

Thanks for reading!

  1. 3

    I’m assuming this is user to user chat and not customer support? Since I already use Firestore I’d probably use Firestore subscriptions rather than deal with websockets directly.

    1. 1

      Yeah, it's a flexible system to coordinate teams of users working on a project. I will definitely check firestore out.

  2. 2

    There are also self-hosted chat alternatives, from a quick Google search:

  3. 2

    I'd say ask yourself if this is something you want to spend "innovation points" on. It will almost definitely be more expensive and time consuming to build it yourself, but if this is something key that will differentiate your product then it may be important. You can't do this for every feature though, so if it feels like dev work that might be a fun challenge for you but result in a generic chat solution maybe skip building it yourself

  4. 2

    I'd say don't do it. Intercom (as an example) starts very inexpensive and scales with the number of users you have. I've been using it for years and it's still less than $500/mo—and I have much more than basic messaging turned on.

    Customer support is too valuable to get wrong. Pay someone else to handle the platform for you. If your own system goes down, you don't want your support system going down with it. Saving $40/mo is going to be pointless if you're losing >$40/mo because your support software is that much worse. Honestly never something I'd gamble with.

Recommended Posts