Its not really how long to split test for its how many unique visitors you should test. For example if you send 100 clicks to 1 offer and that offer gets 2 conversions you are looking at 2% conversion rate. Now you keep rotating offers and seeing your #'s.

To be honest, 100 isnt enough. When Im testing an offer I tend to run 500-1,000 clicks before I know if it converts or not.

Good luck!

im also recommend split test offer with clicks and no days or any kind of time parameter, results can show until later on if you didnt get clicks maybe the offer is good but your ad/banner is not?

hope it helps

Split test for "x" number of clicks - not time. Some agree and some disagree, but 100 clicks is ok between 2 test variables. If the results are fairly close, then you can test 100 more clicks more. Also, use the site :::SplitTester.com::: to make sure your results are statistically significant.

in my experience its not a matter of time or clicks

its a matter of the quality of the traffic
and if something converts
it most of the time works immediately or never

-not always but most of the time in my experience-

Frank

I wouldn't say there is a specified time, but I would test until you have enough data..
if both networks have received the same amounts of clicks and 1 network you have 5 conversions and the other you have 1, then to me that's a good enough test..

Use a landing page rotator. Then spend 1.5x the offers payout.

Don't worry about time.

@Greedy, right now im just direct linking.. gonna try that first

In research, I think that a sample size of about 1,000 should produce fairly statistically significant results, if my memory serves me correctly.

1000 would do just fine. Split test till you feel the conversion matches what you want.
Some say that the number does not matter as long as you reach your threshold of conversion

Thanks guys + gals!