Testing Testing Testing - 1 or many changes

[p0rth0]

Hey Attila!

Thanks for being so awesome.. Really appreciate it
I hope I haven't missed this information on your blog...
Here's my question:
On a low budget (~$200/day) after my initial data accumulation for 5 angles, 25 banners, 1 offer and 2 other popunder offers I got a little confused. Would you remove the bad offers wait for x hours for results OR remove the bad offers, bad angles and possibly something else that sux altogether in 1 run and THEN wait until enough hours?
At this point I would remove all the sheit and leave only the most promising. But I am afraid this would skew the data..

Thank YOU!

 

[iamattila]

hopefully u created 1 campaign for each angle
if yes assuming u did this, that means it was $40 per campaign thus $200 in total for 24 hour period.

Id pick the 2 best angles, pause the other 3.

After that id look at placements. Id start blocking ones that are offer payout x1-2 (depending on your ability to handle risk) and not converting.

 

[p0rth0]

Thanks for the answer Attila!



Unfortunately I did not have that luxury of producing 4 campaigns as the source takes forever to approve campaigns (week). But yes 5 different angles, with 5 different funnels, therefor campaigns. I hope we are on the same page..

It is a display campaign on premium, but I have a RON display coming up, will use this knowledge with that. My focus was more on the strategy behind testing. With a sord-of a yes/no answer expectation.

So I'll clarify. Would you remove everything convincingly bad at once or one really bad apple at a time?

Thank-a-doodly-doo!

 

[iamattila]

One bad apple at the time to measure the result. If you touch too many variables then you don't have a controlled test.