How many sales conversions before you stop Split Test?

by TheWebGuy

Posted: 9 years ago 3 replies

INTERNET MARKETING

Hey guys and gals,

When do you call it quits when you're doing a split test?

Do you call it once the winning version of a page hits 10, 15, ..50 conversions/sales? Or something else?

Basically, how do you know you have a winner?

#conversions #sales #split #stop

jamescanz 9 years ago

Originally Posted by TheWebGuy

Basically, how do you know you have a winner?

Greatly depends on the page.

Take a squeeze page for example.

I'll split test 2 pages and send 250-500 clicks to them...

Then determine the best there and use it.

.

.

Answering your question, if you REALLY want to keep cranking up conversions...

You're probably not going to stop testing anytime soon.
- Thanks
{{ DiscussionBoard.errors[10033860].message }}
David Keith 9 years ago

way to many factors to actually give you a number.

first, the closer the results are the more data that is needed to get statistically relevant results.

so if you have 2 pages and one converts at 5% and the other at 5.2%....its going to take a lot of data to remove the "luck" factor

if you get numbers like 5% and 15%, then it takes far less data to prove that the second page is much better.
- Thanks
Signature
Just a Little War Room Gift.
{{ DiscussionBoard.errors[10033867].message }}

Shaolinsteve

9 years ago

Originally Posted by TheWebGuy

As James said it really differs on the page. You'll want to sit back and look at the structure. Start with the squeeze page if you have one. A LOT of your testing will/should be done here. 200-500 clicks tends to give me enough data per page when split testing, to make minor adjustments and then run other tests from there.

Then... once you have your squeeze page nailed down and getting the consistent or higher conversions, then you'll want to be testing the rest of the pages throughout your funnel. You can do this all together, as long as you have the right tracking tools and enough data to see what's working and what's not.

I don't think you can ever stop split testing in this industry