this has rather large implications for any app developer thinking of entering (or already operating...) in this space.
the one area that I thought a standalone app provider could trump Google's A/B testing tools is to build an API so we don't have to use a clunky web interface... and now Google has done it themselves.
Disclaimer: I wrote an OSS alternative to Website Optimizer (GWO).
GWO is probably the worst Google-branded software I have ever had the displeasure of using. The end-user experience is terrible (page redirects to do variations). The developer experience is like pulling teeth -- I've written an A/B testing library before and I'm not even sure I understand them, but if I do understand them, it looks like it is probably going to take close to a day of work just to do something stupidly trivial like varying a call to action on your site.
Their quick start guide gets you about as far as authenticating to the service, which is wonderful, but you didn't crack open your IDE in the hopes of making some sweet authentication code, now did you?
Having the API doesn't reduce the risk in giving your conversion data to your #1 source of traffic and (possibly) your biggest single vendor.
And to put a cherry on top, you're limited to 10k accesses a day, which is suitable for a hobby site but which will be quickly exhausted in many actual businesses. (Some quick back of the envelope math suggests I'd be on the bubble.)
Huh! I did not know that. Re: the 10k access problem, that would completely negate the use of the API for enterprise sites, would it not? When I wrote the post, I said that enterprise would be the most likely users, too.
Not everything Google releases is a gem; disappointing to know that actual users of GWO aren't impressed.
and then close down the API, or require API keys but then stop issuing them, etc. etc. etc. Google makes great stuff, and some of it would be great for other apps to use, but Google is rarely developer friendly. Compare to Yahoo!'s well supported, well documented Dev Network.
the one area that I thought a standalone app provider could trump Google's A/B testing tools is to build an API so we don't have to use a clunky web interface... and now Google has done it themselves.