Comment by jessep | Hacker News Reader

Comment by jessep | Hacker News Reader

Parent Full thread

I'd like to add a little data to this discussion, from the perspective of someone with a minimal landing page that makes you sign up.

We (https://workflowy.com) built a try-first sign up flow, and a/b tested it against our current minimal signup page, which is a lot like those you describe.

Unfortunately, the try-first performed FAR worse. Not only did we get fewer signups (which we expected). The engagement rate for users who had tried first was significantly LOWER. Exact same product. Same first time experience. One group tried before signing up, the other didn't. They liked it less. Isn't that odd?

From my understanding, other services have experienced this same phenomenon. Why does this happen? It is hard to explain. It could be some aspect of our try-first design was flawed. But, my guess is that it's something psychological about signing up.

I had expected and hoped that the try-first flow would improve word of mouth. I think many do. But it didn't. So I'd encourage as many people to bring data to the table as possible. I've found through doing tons of a/b tests that metrics often don't obey my logic.

Fascinating. Thanks for sharing your data.

There is something about having to put in a little effort up front, and giving up a little, that commits a user to try harder to use a service.

I wonder if this phenomenon is a cousin of the well known characteristic that non-paying users tend to be more demanding and less appreciative of a given service than those who pay (premium users).

Could it be that those who took the time to sign up cared more about the service in the first place (hence, they were willing to sign up)? In other words, is a sign up process (even a minimal one) basically a weed-out for those who are more ambivalent about your offering?

That still doesn't explain why, on average, those who sign up in the no-trial scenerio are more active than those who sign up after trying the product. Of both groups that sign up, the ones without a chance to try the product were more active (including all the activity during the trial)

>"That still doesn't explain why, on average, those who sign up in the no-trial scenerio are more active than those who sign up after trying the product..."

Couldn't it though? If people were so convinced that they didn't need a trial, could it mean they felt themselves to be a better fit for the product. Kind of like an "I've been waiting for something just like this...don't need to try it...I know I want it".

Whereas a trial might reflect some ambivalence by definition: "Not so sure. But, maybe I'll give it a try."

Interesting. One opposite experience: anonymous Wikipedia editing.

Most registered Wikipedians today (59%) say they edited anonymously before registering. That doesn't mean they all necessarily did, but it suggests it's being used in this "try before you buy" way. One piece of data that backs this up for us...

If you're anonymous, there is a call to action on the edit window that informs you're not logged in, and invites you to sign up. That one small link on the edit interface is 9% of all signups on English Wikipedia, and these users have a much higher first time edit rate. On some other language Wikipedias, like Spanish and Japanese, it's even higher -- between 10 and 20% of all signups. I find this somewhat boggling, because in theory it's a horrible experience to interrupt people mid-edit.

I think your tests with Workflowly and the first commenters examples are still correct by and large. Getting users to commit after they try something without signup is tricky, and you lose the ability to easily call them back via email.

For me, the larger lesson is that you need to test this rigorously and be prepared for either approach to fail. It all depends on your product and what your users are like. Turns out many successful Wikipedians are the type who love to dive right in and change things. People who declare that either approach is always going to be superior are falling prey to a bad assumption about how all products and users are alike.

yeah, agreed. i'm sure it will work better in some cases, but it's really specific to each situation.

Thanks for sharing the information. I guess you are right, it is psychological. It is possible that two reasons contribute: time investment in the sign up that increases the value of the service and feeling of having personal space on the website.

One key difference with what you've done though, is that through a link on the bottom I can at least figure out what your product does and see a sample workflow. Some of these sign up landing pages have nothing in the way of explaining what something is.

But maybe what companies are finding is that people respond more strongly when details are withheld from them.

true. that link at the bottom, though, is recent. we used to have zero demonstrations outside the video.

The homepage for workflowy has 5 input fields (3 e-mail fields and 2 password fields), two buttons, and a forgot password link. With a little ingenuity, you can reduce this to two fields and one button (hint: "Sign Up / Sign In").

So a mistake should create a new account rather than throw an error?

Regarding the e-mail/password combination:

1. Valid account: sign in, done.

2. Invalid account: redirect to Sign Up page, currying the e-mail and hashed password.

Chances are the user's browser has already saved the account details and will auto-populate the credentials on subsequent visits, thereby reducing mistakes during sign in.

Integrate a service like http://www.mailgun.com/ to validate that the e-mail address is correct, and not a mistake, which should prevent 99.9% of accidental invalid account creations.

Another personal pet peeve with sign up forms is when asked for Country, Province, and Postal Code. Ask for the Postal Code first and then auto-fill the other two. You can even throw in MaxMind's GeoIP to make a best guess at City, Province, and Country. It's minimal effort for streamlined account creation.

Selection bias. When you make people sign up first, only people who are really interested are going to bother. Most people will just leave. When you let them try it first, all the people who wouldn't have tried it at all in the sign up first model are now trying it and going "meh".

Reply on news.ycombinator.com