Question

Alright now over the last weeks, I've seen a tremendous amount of concern when it comes to people debating how much the openid login system is hurting traffic, so I really would like to see some numbers backing this.

The common complaint is:

OpenID is too hard for normal users, thats the reason I am not getting people on my site, etc etc

So now the question really becomes, what is the "Exit Rate" for users going to your login/register page and getting frustrated and leaving? I am not saying this issue is totally unfounded, I just really want to back up and see if its really as big of an issue as everyone claims.

So if you look in your analytics over the course of the beta, what is the exit rate of users who LEAVE your site after seeing the login?

Here is ours at EpicAdvice.com:

As you can see, 9% of the visitors who clicked on the login page never logged in or registered, and left the site. The total rate site WIDE of people leaving on the login page (you can see this in the small text under exits) is only 0.36%.

What are your statistics? Is this really a huge issue for some sites? Lets put some numbers behind it.

Answer 1

You've chosen a flawed or incomplete metric I think. The real issue with tracking the effectiveness (or lack of) for OpenID is not whether or not people abandoned the process at the very first step (/users/login). It's whether or not they abandoned the process at any time. So, you should look at your exit rate just for /users/login/authenticate as well. Because you've got to have both steps to make the process work.

It all comes back to your audience. I'm currently trying out two different SE sites. One of them is a subset of the SO audience (a programming language that is very low volume on SO). My login exit rate for them is 1%. The closer your target audience is to the SO audience, the less likely OpenID is to be an issue.

My target audience for the other site is NOWHERE near programmers, WOW fans, Math geeks, etc. My target audience is theoretically 78% women over age 40 interested in a specific horseback riding discipline. In a "soft launch" to friends and friends of friends I had a 23.53% overall exit rate on Login and an 80% exit rate on /users/login/authenticate.

I also received email feedback such as:

I think I logged in, but I can't tell. You really need to fix that process, it's confusing.

and..

I love you babe, but the login is broken (that was from my wife)

Based on all of that feedback, I have ceased spending any time, effort and money on promotion of that site, pending a suitable OpenID replacement. If my login process is broken for that user-base, its a waste to blow my marketing efforts on promoting something I know will get sub-optimal results.

I've thought about everything from trying to write my own wrapper to Facebook Connect (or at least some Plain Ole Login) and integrating it via JQuery to (again using JQuery) trying to walk them through the process. If FogCreek aren't going to fix it, it might be worth doing. But, Joel told me at DevDays Boston that the OpenID issue was the top of the list (my assumption is that this is AFTER bugs).

Its a real problem that needs to be addressed. OpenID is known to work for technical audiences, it doesn't for non-technical ones.

Answer 2

From Merspi,

Filtered for pages containing "login"
% Exit: 5.66%
% of Site Total:  0.34%

Doesn't seem like a huge problem but our sample set (page views) is still relatively low.

Has anyone experimented with the idea of using a screen-cast explaining the SO concept briefly and maybe even a short video on how to use OpenID embedded into the FAQ / First-Timers page?

We had thoughts about doing this but haven't had time to create a screen-cast as yet.

Answer 3

facebook connect option would be great

Answer 4

Mine is .44%

But the number is useless. Since your current users/you logging in skews the results.

Answer 5

Here's the data form Math Overflow:

349 visits exited from 119 pages
Filtered for pages containing "login"
                  Exits    Pageviews    % Exit
                    349        2,275     15.34%
% of Site Total:   0.61%        0.52%    12.97%

I'm not really sure how to analyze this. This data corresponds to 24,503 absolute unique visitors, 57,232 visits, and 441,280 pageviews.

Answer 6

The formula is quite simple, all hotmail users won't bother registering. They have a significant market share in web mail (yahoo, aol, hotmail then gmail).