It is so easy to enable Google, Facebook Connect, Twitter, and Hotmail account logins for your site that it makes absolutely no sense to roll your own. Instead of rolling your own features like login, signup, forgot-my-password, confirm-email etc., just implement the above for your sites from the get go. OpenID is good too but I've found almost nobody uses that to log into my sites. 75% are Gmail/Google, 20% Facebook, and rest are Twitter and MSN.
Hold on there. We added RPX (wraps Google, FB, Twitter, etc, etc) signup/login in addition to our username/password and the response has been...unexpected. We still have well over 50% of people sign up with username/password instead of RPX. I had expected a huge shift, but it didn't happen.
Sure, maybe some of those would have used RPX if that were the only thing we had, but apparently they still prefer a username.
Going 100% with the 3rd party logins is a tricky route. It's impossible to know how many potential signups you're losing.
I use RPX only on my site, and the first time someone signs in they can choose a username other than the one which was forwarded by the AuthN provider. So, using RPX does not preclude you from offering custom usernames to your users.
Not to mention that you're potentially putting a very crucial part of your company infrastructure in to the hands of another company whose interests are probably not aligned with yours.
What happens to your startup if (though unlikely) in six months time Twitter decides to start charging for authentication or different levels of service?
"If you want another company to use your services, then you need to make it easy."
Or, make a service that's so good, enough people demand integration.
This article makes some good points but I think things change once you start talking about startups that charge money ... and the Typekit example he gives is proof. It's the one paid service he lists–it doesn't use RPX, FB connect, etc.
10 comments
[ 1.8 ms ] story [ 25.2 ms ] threadSure, maybe some of those would have used RPX if that were the only thing we had, but apparently they still prefer a username.
Going 100% with the 3rd party logins is a tricky route. It's impossible to know how many potential signups you're losing.
What happens to your startup if (though unlikely) in six months time Twitter decides to start charging for authentication or different levels of service?
Something to consider.
If you'd like assistance with optimizing the user experience to maximize 3rd party login success rate, please contact johnb at janrain.com
Or, make a service that's so good, enough people demand integration.
This article makes some good points but I think things change once you start talking about startups that charge money ... and the Typekit example he gives is proof. It's the one paid service he lists–it doesn't use RPX, FB connect, etc.