If you're talking about landing pages. Never longer than a day. With Google that's really all the traffic sample you need. Now, which day might be more important than others. I would say split testing Wed - Fri would be smarter as Mon/Tue are the highest volume/highest converting days. Split testing on the weekend could yield drastically different results.

That being said, I'm sure some people like to test longer, but I don't see the point.

If you're talking about Ads, then it's across the board. You can tell VERY quickly how well an ad does or doesn't do, though within a few days a better ad (if you're sampling more than 2) might drop in CTR and then you've got nearly equal ads - that's a tougher decision.

As you've undoubtedly seen, there are thousands of different variables in a landing page - so keep testing, the more the better.

I don't have a certain time...

And I'm not sure if you will want to do this as it depends on volume...

But I usually wait for THIRTY (30) actions...

As in... if it's for an opt-in... 30 opt-ins...

Or a sale? 30 sales...

Statistically this makes very sure, how well your ad is doing...

(and we're not even factoring CTR in here, I'm talking about the end results)

Josiah

Awesome,

Thanks for your input.