Sorry for that stupid thread title but I couldn’t think of something better.
What I mean is the following: Say you have a banner running on a site, when does the point come where you say "Ok, this is statistically significant enough data to see that this banner sucks so now I’m gonna change it up with a new one"?
What do you base your decision on? (F.e. if banner doesn’t reach x clicks after 10k impressions, change it.)
Something like that. Currently I have a banner running that got 3k impressions and 1 click. I know I know, that’s nothing. But is that already where I could say "Ok, 3k people saw the banner and only one of them clicked so I could now test a new banner"?
Your thoughts are appreciated.
depends where but on sites like PoF I cut stuff after as little as 1.5k impressions. it depends what ctr you are shooting for. If you’re shooting for .1’s then you need to let it go longer but if you’re only going to accept something like a .3 .4 or .5, well if it doesn’t get crazy clicks out of the gate it’s not going to later on.
You can also use SplitTester.com just use impressions and clicks instead of clicks and conversions.
I just look at from a numbers point of view and offer averages….like conversions rates and such…..
eg…my clicks are costing me $1.5 my offer payout is $3…that means I need to be at a 50% CVR….is that possible? prob not…..
you may not know the averages……and thats ok, ask your AM how its doing EPC wise….then thats a baseline for you to work off of……if he tells you the network EPC is 30c and your cpc is 40c….your above the average….so tread lightly…..
hope that makes sense…..
The Article Published IN 09-17-2011 11:29 AM