Statistical Significance Tool For Offers

Fiver

Fiver

New member
Jan 30, 2009
2,011
12
0
I've seen the SS tools out there but they all compare 1 ad against another. For instance: Split Test Accelerator: Faster Ad Testing Now!

This tool compares two different ads (or I guess you can use offers like me) and gives the confidence level of the winning ad. But the outcome seems to be dependent upon the numbers of the 2nd ad.

What if I don't want to compare ads? What if I'm doing dayparting and just want to look at an offer run during a certain hour of the day and see if the C/R will hold up over time for that hour?

As an example:

Offer #1: 3 pm
Impressions: 25,000
Conversions: 7

How do I know if this is enough data to make a decision to keep running the offer that hour? Where is the tool for that?
 


Fiver said:
I've seen the SS tools out there but they all compare 1 ad against another. For instance: Split Test Accelerator: Faster Ad Testing Now!

This tool compares two different ads (or I guess you can use offers like me) and gives the confidence level of the winning ad. But the outcome seems to be dependent upon the numbers of the 2nd ad.

What if I don't want to compare ads? What if I'm doing dayparting and just want to look at an offer run during a certain hour of the day and see if the C/R will hold up over time for that hour?

As an example:

Offer #1: 3 pm
Impressions: 25,000
Conversions: 7

How do I know if this is enough data to make a decision to keep running the offer that hour? Where is the tool for that?
Click to expand...

Use the split tester and compare the 3pm result against your normal average 3pm result. You can't test confidence without having something to compare against.
 
mituozo said:
Use the split tester and compare the 3pm result against your normal average 3pm result. You can't test confidence without having something to compare against.
Click to expand...

The 3pm result is my normal result and all the data I have right now. So with that in mind, what should I put in the comparison box?
 
Fiver said:
The 3pm result is my normal result and all the data I have right now. So with that in mind, what should I put in the comparison box?
Click to expand...

Confidence levels are calculated in comparison to something. You can't just give a program two random numbers and expect it to give you a confidence level.

If that's all the data you have then set up a A/B split test and wait a couple days and use that.
 
You must log in or register to reply here.
Top Bottom