3 ms·
Are you saying this was how things were done before? Or this is how things are done now? P.S. Also, does your software's MVT optimize itself like genetify - ht
by rumpelstiltskin 16y ago
Are you saying this was how things were done before? Or this is how things are done now?
P.S. Also, does your software's MVT optimize itself like genetify - http://www.shoemoney.com/2010/03/03/split-testing-with-a-genetic-algorithm/ http://www.shoemoney.com/2010/03/03/split-testing-with-a-gen...
- paraschopra 16y agoThis is a new feature that we have released today! I'm excited about this because it solves pain-point of a lot of our users. No, VWO doesn't dynamically optimize because statistics become s really complicated once you try to model such optimization. As far as I know Genetify (or any other tool out there) doesn't have a statistically valid method of optimizing the test (while it is running). The math required to do such optimization while maintaining statistical validity is beyond me, at the moment. (Though I am actively researching on this front. Any pointers/links will be appreciated)
- rumpelstiltskin 16y agoSo no more iframing etc? Genetify is open source - http://wiki.github.com/gregdingle/genetify/ http://wiki.github.com/gregdingle/genetify/ - you can have your programmer go through it for the algorithm. That'd be a good starting point. Statistically valid optimization would be a HUGE plus. Enough for me to pay for VWO instead of using genetify.
- paraschopra 16y agoGenetify doesn't have a statistically valid algorithm so no point looking at its source. Moreover, there are some basic optimization features already built into VWO: http://visualwebsiteoptimizer.com/split-testing-blog/why-ab-testing-is-truly-risk-free-even-when-every-lost-sale-matters/ http://visualwebsiteoptimizer.com/split-testing-blog/why-ab-... Third party cookies (via IFRAME) is a choice, it isn't enabled by default for all tests.
- rumpelstiltskin 16y agoAhh I see. Cool.