Save my client from building an Internet ghost town

September 23, 2008 5:26 PM

How frequently do users of community-based sites abandon their registration attempt? And how does that correspond to how painful we make the registration process?

I'm working on an e-commerce application for a cool little company. Their products serve a fairly passionate audience, so we thought it would be great to let users rate products and write reviews, a'la Amazon. My client is trying to force people to fill out a long registration page and validate the users email before they can participate. I'm worried that this will deter users from participating in the first place and suggested we either trim down the registration process or let them log in with their Facebook ID (via Facebook Connect). The client seems hesitant because then they won't "...own the user's data." Sigh. They guy responds well to arguments based on statistical data. I know from anecdotal experience that the more painless you can make registration, the greater your community will grow. Does anyone have some sort of web marketing statistic or hard data that would help bolster my case?

posted by centerweight to computers & internet (4 comments total)

You'll have a hard time getting other companies' conversion statistics -- you might as well ask for their revenue numbers. You need to do split testing, where half the people get the long registration page and half get the short, and do conversion tracking. Google Analytics will do the conversion tracking, including tracking the whole funnel so you know where the lost conversions gave up.

What better statistical data than that from his own split test on his own site? And then you've got the split testing methodology to use for the next marketing question.
posted by mendel at 6:18 PM on September 23, 2008

The book Don't make me think (2nd edition) has a chapter called Help! My boss wants me to _______!, with emails you can send your boss or customer to explain why those ideas are not viable. The first case is just like yours: The perils of asking for too much personal data.

The author highlights three points:

1. Long registration forms keeps you from getting real data, as users fill the form with rubbish (I think this might be the winning point: what's the point of owning false data?)

2. You get fewer completed forms

3. It makes you look bad. It's better to ask for the minimum information and then, when you've established a relationship with your customers, you can ask for more data.

Also, Designing for the social web has a chapter that talks about sign-up friction and has a lot of good examples.

Unfortunately the books have no statistics, but maybe if you showed them to your client he might be convinced. Both books are excellent, you should get them anyway.
posted by clearlydemon at 6:27 PM on September 23, 2008

If he is suggesting validating email, then he clearly is receptive to the idea (and aware of the difficulty) of collecting good data. That's good. But, tell him about anon email boxes (e.g. mailinator, temporaryinbox, etc.) and various registration workarounds (e.g. bugmenot). Explain that if the process is onerous, there will be less registrations and more bogus data. I don't think you need stats to make this argument.

The A/B testing is a nice idea, but really, that would involve a huge amount of work to get two entirely separate registration setups working, and you will only prove what you know: People don't like pain-in-the-ass registration forms. Not exactly a revelation.

As an alternate approach, try to explain to your users why you want each piece of data. For example: Instead of saying, "Give me your email, we promise not to spam you." Try: "If you would like to use our password reminder feature, please enter your email address." If you explain why people should give you info, then they are much more likely to provide it.
posted by kamelhoecker at 6:55 PM on September 23, 2008

These are all fantastic ideas. I think the combination of a/b testing and exposure to the various sites and authoritative books could do the job if some sort of industry rule of thumb isn't available. Thanks.
posted by centerweight at 7:19 PM on September 23, 2008