constant or variable?

[watermelonman]

hey Charles,
thanks for answering my last question!!

Here's another one that has been puzzling me for a while now:
when we say traffic, we are actually referring to visitors and they are ppl, right? ppl come and go and you'll get new visitors everyday. let's say you've tested your landing pages etc and are ready to scale up.

do you consider your "traffic" a constant or variable? how can you be sure that once you invest more money to scale it up, the result would be linear? I mean, if you've spent $5k on split testing and found the winners and now you are ready to scale up. how sure are you that the result would be linear(if you invest $10k now, you'll make $5k profit and if you invest $20k now you'll make $10k)?

what puzzles me is that i personally think "visitors" is a variable, not constant. you get a different number of visitors everyday, and these visitors are different(some are first time visitors and some are repeating visitors). Is the testing done on a small sample really reliable?

Is it possible that your initial testing result looks very good but once you scale it up, it totally bombed?

thanks again!!

 

[charles_ngo]

Dude...you are overanalyzing.

Yes traffic = visitors = people

The traffic is variable and can fluctuate. A website might get more traffic on the weekends, or around holidays it might decrease. However the fluctuations are small and I find that overall the traffic is pretty stable.

What causes the spend to change day to day might be the number of people that click on your ads. But for the most part if I'm spending $1k a day, it should be spending $1k a day next week provided my banners perform.

It is possible for campaigns to bomb when you scale up because it depends HOW you scale up. Are you simply increasing bids? Then the CPC will increase, and the jump in prices might cause your profit levels to go below negative.

As far as small sample, yes it is if the data is statistically significant enough (splittester.com).