Quote:
Originally Posted by Shoplifter
How's it going for you guys today?
I think I am having one of those "statistical anomaly" days.
|
The rollercoaster continues BUT, this being Wednesday, I don't freak out as much. Actually, Tuesdays are notoriously the worst sales day of the week (which is why many retail stores have sales on Tuesdays), and Wednesday is almost as bad. Percentage-wise I'm holding steady tho so this tells me it's a day of the week thing. Now, come Friday....LOL
Quote:
Originally Posted by VGeorgie
CCBill's form hits aren't always (ever?) real time.
If you do any A-B comparison, the only meaningful way is to write a server script that exactly divides the clicks to the signup form between the two, alternating one then the other. That's the only way you'll smooth out the variations in time, weekday, week of the month, payday, month, and all the other variables that can affect purchases.
|
You are absouletely correct in this observation. A/B testing is best done with a server-side script that does a simple 'split' test: one click goes to A, one click goes to B. This is the type of A/B testing I mean when I refer to my setup.
