The Perl Survey, email addresses, and conversion rates

I haven’t mentioned it here before, but I’m running a survey of the Perl community over at perlsurvey.org. If you’re a Perl programmer, or even just an occasional user — modifying the odd CGI script, or writing a quick script to automate some sysadmin tasks, for instance — then please take 5 minutes to fill it out.

One thing I had to consider when setting up the survey was how to avoid duplicate submissions from a single source, or at least make it tricky enough that it couldn’t be got around by a simple ballot-stuffing script. I don’t *think* anyone would be malicious enough to bother with that, but there is a history of ballot-stuffing for fun in the Perl community, and I didn’t want to have to police that.

So, the solution I went with was to require an email confirmation, and to throw the email address away at the end of the survey. Many people have complained about this, but I’m not sure whether there’s another good option that wouldn’t get just as many complaints. I suppose next year I can try something else and see.

One interesting thing that came out of the discussion about taking email addresses was the question of how many people are turned away by it. I did a little bit of shallow digging and came up with the following stats:

There’s a little bit of apples-and-oranges going on in there because the first and last figures are unique visitors, while the two in the middle are page-views, but I suspect the overall gist is right.

In any case, it’s interesting to look at the conversion rates from people visiting the site to actually completing the survey. I suspect that 40% is a great figure, but unfortunately I have nothing to compare it to. The only figures I found about Internet conversion rates was for e-commerce, where they suggest that 2-5% is the normal range. I suppose if I were in market research, I’d know how to find these figures.

Regardless, I find it interesting the sort of metadata (in the sense of meta-data, rather than, you know, *metadata*) we can get from this survey before it’s finished. Hopefully next year I’ll be able to use some of this information to make the survey even better.