Mailinglist Archive: opensuse-project (240 mails)

< Previous Next >
[opensuse-project] Re: Bugzilla account creation.
  • From: Jim Henderson <hendersj@xxxxxxxxx>
  • Date: Thu, 9 Aug 2012 16:06:23 +0000 (UTC)
  • Message-id: <k00n5v$noc$3@dough.gmane.org>
On Thu, 09 Aug 2012 09:35:21 +0200, David Majda wrote:

The fact that every additional field in a form lowers the percentage of
people that complete it is a well established fact by multiple studies.
Just Google e.g. for "conversion rate number of form fields" and spend
some time going through the links to see some examples. Or read any good
usability book.

I consider the above to be "hard" enough. I don't see any reason why the
registration form should be an exception.

I would far prefer to see analytics that show the number of times someone
visits the registration page via an openSUSE site (ideally) and doesn't
complete the form. That would provide specific data as to the number of
times the form was reached and wasn't completed - for whatever the reason.

One might argue that given the nature of the registration form - that
Novell and SUSE uses for the registration are typically business-
orientated, which would mean you won't have hobbyists (generally) would
have high rates of completion - so the number of abandoned visits to the
page would still be a reasonably accurate measure for openSUSE, where
there's not a business driver on the end of the user filling the form out
(which is an instance where the option exists to say "screw this" and
either move to another distribution or just opt to not participate.).

That's not to say there isn't a problem, but as a friend of mine who
works for Google is fond of saying, "the plural of 'anecdote' is not
'data'."

The only reliable way to see how the complex and personal-data-acquiring
form affects the number of registered users I know of is an A/B test:
prepare a simplified version of the form and measure the difference in
registration rates.

As a predictive measure, yes. As a historical measure, web analytics can
provide some useful insights.

On the forums alone, we have currently 69,105 registered members -
which means we have that many people who have gone through the
registration process for the forums alone - and once that's done, they
have access to Bugzilla, the wiki, and everything else in the openSUSE
project that requires authentication. Even SUSE Studio and OBS.

Significantly less than 1% of those people have complained about the
process it took to get registered - and the openSUSE forums have always
used this registration process since their inception with the forum
merge project.

People who don't complete the registration process don't complain, they
leave.

Three classifications exist:

1. People complete and don't complain.
2. People complete and complain.
3. People leave and don't complain.

It's *usually* (but not always) a good indicator if you have a
significant number of people in category #2. If someone found it
cumbersome but felt that the value existed in filling the form out
anyways, they'll usually complain.

1% of the registrations that relate to the forums (which is a subset of
all registrations related to the openSUSE Project) would be about 600
complaints. If there have been about 700 complaints, I'm pretty blind
because I simply don't remember seeing them. I remember seeing maybe
half a dozen people ask "what's up with all this information", and most
of those people have been on the mailing lists that I read.

Also, the fact that "it was always like that and worked somehow" is not
a relevant argument here. Computers also worked well when they had 640k
RAM but they have GBs today. The question is whether this can work
better. Maybe these 69k users could have been 138k today if the form was
simplified long time ago.

Sure. And I'm not nor have never said that we *shouldn't* do this. I'm
saying in terms of setting priorities, identifying the severity of the
issue is significant. In project management, you don't spend a lot of
time or effort on something that's impacting you by 10%. You spend it on
things that are impacting you 80% or more. When you've dealt with the
most severe problems, then you have the luxury of looking at the smaller
issues.

Otherwise you're working on the wrong things.

Note that I'm not saying it's perfect by any stretch. I'm just saying
that relative to the other issues that have been under investigation, a
working (if even somewhat convoluted) registration process is
relatively "small potatoes".

To me, it seems weird to question the priority of this in 2012, when
every startup optimizes their front page and registration process like
crazy, people often refuse to buy in e-shops requiring registration, and
there is an entry on Hacker News about a workflow simplification leading
to increased conversion rate every month.

Again that's still anecdotal evidence, though. Just because everyone is
spending their time doing something or following the latest "hot trend"
doesn't mean we should.

To reiterate, though - I'm not saying "we shouldn't spend time on this",
I'm saying "let's find out if we really need to spend time on this".

For my part, I can see if I can actually obtain some data that says we've
got people who are not completing the registration process and how big of
an issue it really is. It may be a bigger issue than I think it is, in
which case I'd fully support efforts to reduce the number of people
abandoning the project before they even get started with it.

Jim
--
Jim Henderson
Please keep on-topic replies on the list so everyone benefits

--
To unsubscribe, e-mail: opensuse-project+unsubscribe@xxxxxxxxxxxx
To contact the owner, email: opensuse-project+owner@xxxxxxxxxxxx

< Previous Next >