During a panel discussion at Inman’s Real Estate Connect 2012, Placester CTO Fred Townes talks about how much data you need to gather to make a decision about your website design, and how to get the most results out of A/B testing.
Gahlord Dewald: When you were talking about landing page optimization, like A/B optimization—has anybody here done an A/B test on their website, like they asked some people, “Hey, I just want your name and email,” and then other people you ask for their name, email, phone number, and all the other stuff that the agents insist that you ask for? OK, great.
And you had mentioned that you can’t start a test on Monday and have an answer by Tuesday, so there’s always that question of, “How long do I let a test run for?” Or, “I don’t get enough traffic to make decisions based on data to improve my B quality.” Do you have any thoughts on those kinds of issues? How much data do you need to make a reasonable decision, or how to get over that if you don’t?
Frederick Townes: Well, a couple things there. In terms of the amount of traffic that you have and feeling that it’s inadequate to make a scientific decision around what’s effective and what’s not, you can always spend a couple hundred dollars, maybe a few thousand dollars, depending on how niche you are and how expensive that traffic might be to realize in terms of sending it to your website.
To send that traffic to your websites, you have a great sample set, but for lots of us who have already spent the time optimizing our websites, have spent the time driving traffic through whatever means that you do it, even if you’ve got a few hundred visitors that are hitting your site week after week, that’s still enough, provided that it’s consistent, whatever it is, to make some inferences based on some testing, and it doesn’t matter what tools you pick as long as you know what you’re testing for.
So, test for one thing at a time, like maybe it’s a particular page or it’s a new layout of a page that you’re testing. Do people actually fill out this form? OK, that’s your first test. Great, they fill out the form. OK, how many of those people are actually doing it? Can I make some changes to see that …