Home > Paid Traffic Sources > Mobile

Confused about statistical significance and how long to run LPs for (6)

06-19-2015 06:53 PM #1 sawrubh (Member)
Confused about statistical significance and how long to run LPs for

So I'm running an offer in India in POP on ZeroPark with a payout of $0.48. I have gone Ruby Tunes' awesome tutorial and caurmen's Appetiser but I'm still confused when to cut my landers. Right now I'm split testing 3 landers.

How do I decide how much to keep my daily budget and campaign budget to?

06-19-2015 09:22 PM #2 Awesome Sauce (Member)

I think what you are looking for is this: http://stmforum.com/forum/showthread...Banners-Part-1

06-19-2015 09:46 PM #3 vortex (Senior Moderator)

Quote Originally Posted by sawrubh View Post
How do I decide how much to keep my daily budget and campaign budget to?
When starting a new campaign it may be best to keep the daily budget low just in case of any mistakes / accidents (e.g. tracking mistakes, incorrect campaign targetting, etc.)

The higher the daily budget, the faster your split-testing will go and the faster you can cut landers - but this would also depend on your bid. If your bid is low and your daily budget is high, you may not get enough traffic to reach that daily budget.

As for campaign budget - I usually just leave it at "unlimited" because I can always pause the camp when I need to anyway.

Hope that helps!


Amy

06-20-2015 03:27 AM #4 sawrubh (Member)

So according to the rules of cutting, we cut a LP once there is no conversion after 3x payout. How do we pace/spread this spend over time because it might be that the type of people who convert well on my offer are active during 8am-11am while I spent all my 3x budget during 1am-5am? I mean if we allow, the traffic source might spend all of this 3x in a very short time (if your bid position is 1) however we can also make sure we get very slow traffic? What speed is the right speed?

06-20-2015 03:46 AM #5 ocean25 (Member)

Check this out:

http://www.peakconversion.com/2012/0...al-calculator/

95% is statistical significance, you can safely cut the underperforming LP after reaching 95% statistical significance after comparing to your winner.

06-20-2015 04:33 AM #6 kepe95 (Moderator)

the type of people who convert well on my offer are active during 8am-11am while I spent all my 3x budget during 1am-5am? I mean if we allow, the traffic source might spend all of this 3x in a very short time (if your bid position is 1) however we can also make sure we get very slow traffic? What speed is the right speed?
What kind of offers are you running? So far I only noticed smaller differences between different times. If it's not too expensive I'd always go for a 24 hours timeframe. If you test something and it works "okay" you'd usually keep running anyways, and this would be over several days so you would notice any differences. Some traffic sources have an option to spread the traffic evenly throughout the day based on your maximum daily budget.

Cutting landers here is how I do it:
- Already know my landers are at least "okay" , so rules like 3 times payout 0 conversions - cut the lander , are not even an option.
- Determine if my lander performance might be dragged down by other parts of the funnel or un-finished optimization.
- Usually get as much data as fast as possible, monitor along the way. If one lander is particulary bad, you can usually see it after a total of around 100 conversions (I never cut earlier, have seen way too many opposite results when looking at the "after 40 total conversions" result , and then the "after 100 total conversions" result , or even the after 1000 total conversions result).
- If they both are head on head with only a small difference I just keep them running for days, or (golden nugget haha , sometimes cloning the campaign and low-balling the bid with a different but also well performing lander (and angle) can increase profit significantly without burning out your lander too much)

Home > Paid Traffic Sources > Mobile