Clear.CO or Cntortd.COM – Split Testing a Domain Name

I’m building a group travel platform to make the logistics of planning trips with friends as fun as the trip itself. I had a great name for it too. Since that was taken I started building under the name GrpTrp. It has symmetry, it’s relatively easy to describe (“grouptrip, no vowels”), and hey, I’m bootstrapping a proof of concept, I can’t be spending thousands on domains.

After listening to Jason Calacanis gush over .CO on This Week in Startups, I realized, you know, maybe I can have my cake and eat it too. The name I wanted with a big ole heaping of cash I can spend on more important things. Before I went through the effort of rebranding (while easy given I haven’t launched yet, still, it’s time I could be spending finishing out the beta) I thought I’d try to get a little data first. So here we go, the story of the vs the

Trust #

My first concern with the was that users wouldn’t trust the domain. After all, outside of tech circles people still think of renegade Russian hackers when they see anything other than .com, .gov or .edu. Or so I thought. To find out, I ran two advertising campaigns (on Facebook) with the hypothesis that the Click-Through-Rates would skew towards the domain users trusted more.

The first campaign pitted the exact same ad copy with the only difference being the domain that was displayed by the link. In my case, “” vs. “”.

Ad Impressions Clicks Conversion Rate CPC 196,684 27 0.014% $0.83 143,583 28 0.020% $0.80

While it looks like the ad performed a bit better than the ad did, we can’t say with statistical significance that will always be better than However, it’s pretty clear that at the very least, doesn’t perform worse than, so the idea that folks don’t trust .co is pretty well bunked.

Just in case you’re not convinced, I ran a second test. This time the heading of the ad was the domain name (“” and “”) to draw a little more attention to it. The body remained the same, and the URL at the bottom of the ad matched the URL in the title. Here were the results:

Ad Impressions Clicks Conversion Rate CPC 141,374 29 0.021% $0.78 114,388 30 0.026% $0.73

Again, we’re not statistically significant, but through two tests covering almost a half a million impressions, it’s clear that the .co holds it’s own against, and perhaps beats out, the .com.*

*Open Kimono Note: One factor that might have influenced this test is that the logo displayed for the ad was the logo for GrpTrp. This may have actually held the numbers for the .co down as there was dissonance between the logo and the product name. I would’ve gone for an ad with no picture, but I had ad-block on and didn’t double check it before running the test. Mea culpa.

So it seems clear that users will click at a similar rate when presented with a vs a, but once they do, can they tell their friends about it and spread the word?

Virality #

A common back-of-the-envelope suggestion is to see if your name is spreadable is to see if you could say the domain over the phone and have people understand exactly what you’re talking about. I would argue that that’s necessary, but not sufficient. Here’s how I came to that conclusion:

My original reason for choosing was that I could describe it as “it’s with no vowels.” I figured, if it worked for a company like SCVNGR then, hey, why not? Here’s why not. Search in google for scavenger. SCVNGR is nowhere to be found. If you go to page 2, you’ll see a sponsored ad asking if you meant SCVNGR, but guess who’s paying for that one? You think a one-time domain name expense is tough, try the recurring expense of running a sponsored ad every month until the end of time.

The point is, SCVNGR has made it work, but at what marketing cost? If you can’t afford the domain, you certainly can’t afford the marketing costs of overcoming an unsearchable domain. Make no mistake, SCVNGR is becoming a very successful company, but they’re doing it in spite of their name, not because of it.

Alright, back to the analysis…

To figure out which domain would be more spreadable, I broke out the mother of all ghettopreneurship tools – PickFu. For the uninitiated, PickFu uses the power of Amazon’s MechanicalTurk to get users to pick between two alternatives and provide reasoning for their choices. It’s quick, it’s dead simple to set up, and perhaps most importantly, it’s cheap ($17 for 200 responses).

Here’s the questions I asked:

Given the following descriptions and their associated addresses, which would you be more likely to remember and use?

The overwhelming answer was that people preferred the You can see the full responses here, but here’s a summary. The majority of the responses mentioned that the extra thinking involved in vowel-removal would make them less likely to remember the name. And guess where people go when they can’t remember the exact name? Google. See above.

Survey says wins 79% to 21%

Still, the responses that tended toward were largely due to people not trusting a domain. Over time I’d say that this sentiment will go away, but it’s there. One thing to consider, at least with .CO, is that .CO will be running Super Bowl ads this year a la GoDaddy, so that might make a big dent in fixing the legitimacy concerns with .co. time will tell.

Conclusion #

As always, there are holes in this “study”, but I think the data is strong enough for me to make the move from to I think we’ve reached a trend of diminishing power for the .com TLD, and .CO is exacerbating that trend.

As always, I’d love to hear your criticisms and suggestions for further study. Of course if you want to leave a supportive note, that’s cool too :-)


Now read this

Productivity Hack for Bootstrapping a Company on 2 Hours a Day

Here it is, the single greatest productivity hack I’ve learned while building GroupTrip. Perform a mental backup before you stop working. # Take 30 seconds in your editor of choice and write down what you would be doing for the next 15... Continue →