Went From 2k To 5k From This Simple Change

[ewenmack]

The story is here Split Test, Split Test, Belcher Button | Antony's Internet Marketing Blog

Best,
Ewen

[hpgoodboy]

Great story. Thanks for sharing!

[ewenmack]

You're most welcome.

Best,
Ewen

[Ross James]

Thanks ewen... It's funny... I came on here because of an epiphany I had from one of your other posts you made...and here this was right on top of the copywriting board... making my morning easy on me.

Now I have to dig it up the original sauce... It WAS right there on a silver platter, I just came to the realization, "wait ewen was posting on about it!"

[ewenmack]

Ross, I hope it isn't too scary for you that I can read your mind...[chuckle]

Best,
Ewen

[Ross James]

If you were reading my mind you'd just be great at structuring what ever it is that goes on in my mind...mind reading happens everyday! But I have to warn you, it's not recommended that you model my mind, you might find yourself wanting to stay up drinking copious amounts of coffee @ 3 AM - Just to get in another full day at work. It's not a bad thing. A lot of times I say to myself, wait, if I were in ewen's shoes I'd probably want to split test that. True story.

[markpocock]

Am going to add this Buy Button to our web sites this morning -

[MikeHumphreys]

Nice find Ewen but don't etch the results in stone.

The author didn't let the test run long enough to get statistical confidence. You need 100 actions minimum to usually achieve that.

After 600 or so total page views, the results could reverse... or both varients could just flat-line going forward.

Like any other test results, it's going to vary from niche to niche and product to product.

Hope that helps,

Mike

[mobread]

Hey cool advise! I seen this question asked on your blog, but couldn’t find an answer. What software were you using or have attached to your site to split test?

[ewenmack]

Mobread,

The link is to somebody else's blog, not mine.

I also don't know the software he used.

If you are looking for testing software...designed by a copywriter,
then check out Mike Humphreys at this link Easy Multi Tracking: Professional Copywriter Reveals His Proven Conversion Boosting Tool That You Can Use To Skyrocket Your Conversion Rates FAST!

You will find Mike on this forum a lot, if you haven't noticed.

Best,
Ewen

i

[Andrew Maclean]

Thanks for sharing this story .

[markpocock]

This sales letter has taken the Belcher Button one stage further
with the addition of anti virus software and the 60 day money back
guarantee.

Mark Anastasi's FB Auto Cash - Turn $5 into $100's in Minutes!

Thoughts?

cheers

Mark

[Murlu]

I personally don't like the belcher button but I can't turn my head to the effectiveness. I'll need to update my buy button on my new PLR site because I went with just a simple one - the credit card options are, in my mind, what really help - adds trust.

[hhunt]

Great story. I just had to check it out twice.

[Stephen Dean]

Originally Posted by MikeHumphreys

Nice find Ewen but don't etch the results in stone.

The author didn't let the test run long enough to get statistical confidence. You need 100 actions minimum to usually achieve that.

After 600 or so total page views, the results could reverse... or both varients could just flat-line going forward.

Like any other test results, it's going to vary from niche to niche and product to product.

Hope that helps,

Mike

What model are you using to find significance? SplitTester.com puts it at 95% confidence - meaning 95% confident the Belcher button converts better. Of course that doesn't mean it will perform 4 times better.

And of course, 95% confidence means there's a 1 in 20 chance the result will reverse. But still, that's a pretty good result.

Cheers!
Stephen

[MikeHumphreys]

Originally Posted by Stephen Dean

What model are you using to find significance? SplitTester.com puts it at 95% confidence - meaning 95% confident the Belcher button converts better. Of course that doesn't mean it will perform 4 times better.

And of course, 95% confidence means there's a 1 in 20 chance the result will reverse. But still, that's a pretty good result.

Cheers!
Stephen

Hey Stephen -- How are you?

I also use a 95% confidence level.

My point was the blog author is making the decision after 15 total conversions and 600 (or so) page views, not page clicks. Based on his screen shot, I'm not sure SplitTester is the right calculator to use since it's asking for Page Clicks, not views.

Both versions also didn't get an equal amount of views which is a no-no for straight split-testing.

Either his testing software didn't rotate the versions equally or he ran two separate tests and cut one of them off early b/c he thought he had the winner. If it didn't rotate them equally, then it showed the lower converting one which is contrary to what some of the "auto-optimize" testing scripts will do.

Maybe it was outside traffic to that one version which bumped up its specific page views.

The blog author doesn't explain the 70+ page view difference in his article so your guess is as good as mine.

In terms of the 100 action or 1000 uniques rule of thumb...

I've seen a variation start out hot on a split test (about 11.3%) and cool off over a 1000 unique visitors to a 1.2% conversion rate. It still wound up being the winner for that particular case but it was being much closer to the other variations. If I had cut the test early, I could have picked the wrong variation as the winner.

Okay... I just spotted another problem with his stats.

The Belcher button didn't just have a different order button shape but had the credit card logos added to it as well. The Kajabi style order button had no credit card logos. You could also argue the Belcher had a different design style to it as well.

So was it the shape of the Belcher button or the credit card logos that bumped response rates? Or maybe it was the design style?

Would the Kajabi button convert better if we had added the credit card logos beneath it?

Bottom line... He didn't test one variation (order button shape) of another element against another. He actually tested three variations on the same test (button shape, button style, and credit card logos) and lumped them all into one "answer".

Yes, Belcher button may convert better but I'd still caution people to TEST it on their own product & niche instead of just assuming it will convert better.

My 3 cents,

Mike

P.S. Just to be clear, I'm not picking on the Belcher button. I've used it on certain salesletters I've written. I'm picking on the inaccuracies of the test example that was shared.

[Ken_Caudill]

Originally Posted by Stephen Dean

What model are you using to find significance? SplitTester.com puts it at 95% confidence - meaning 95% confident the Belcher button converts better. Of course that doesn't mean it will perform 4 times better.

And of course, 95% confidence means there's a 1 in 20 chance the result will reverse. But still, that's a pretty good result.

Cheers!
Stephen

Confidence itself is an arbitrary factor.

[Stephen Dean]

Originally Posted by MikeHumphreys

Hey Stephen -- How are you?

I also use a 95% confidence level.

My point was the blog author is making the decision after 15 total conversions and 600 (or so) page views, not page clicks. Based on his screen shot, I'm not sure SplitTester is the right calculator to use since it's asking for Page Clicks, not views.

Both versions also didn't get an equal amount of views which is a no-no for straight split-testing.

Either his testing software didn't rotate the versions equally or he ran two separate tests and cut one of them off early b/c he thought he had the winner. If it didn't rotate them equally, then it showed the lower converting one which is contrary to what some of the "auto-optimize" testing scripts will do.

Maybe it was outside traffic to that one version which bumped up its specific page views.

The blog author doesn't explain the 70+ page view difference in his article so your guess is as good as mine.

In terms of the 100 action or 1000 uniques rule of thumb...

I've seen a variation start out hot on a split test (about 11.3%) and cool off over a 1000 unique visitors to a 1.2% conversion rate. It still wound up being the winner for that particular case but it was being much closer to the other variations. If I had cut the test early, I could have picked the wrong variation as the winner.

Hi Mike It sounds like you have a solid testing method. And on a lot of points, you're right. It's very possible that the results will even out a bit or even reverse.

SplitTester.com seems to have the same result as the Chi Square calculators, however. (Maybe it even is a chi square calculator.) A Chi Square calculation puts the test at 96% significance.

Just like the chi square test would have likely correctly predicted the winner of your example test, even though the actual conversion rate of the variable hadn't been settled yet.

But there are different methods for finding significance. (That may be why Ken calls it arbitrary.) How do you determine 95% significance?

But if we're going to use a yardstick of sorts, chi square is one. And I think there's enough data to make a decision.

If I was running split tests, I'd start a new test.

Cheers,
Stephen

[luisimon]

what's kind of the software mentioned there? confused

[ptpasta]

Nice advise in all the thread. I'm taking notes :P

[MikeHumphreys]

Originally Posted by Stephen Dean

Hi Mike It sounds like you have a solid testing method. And on a lot of points, you're right. It's very possible that the results will even out a bit or even reverse.

SplitTester.com seems to have the same result as the Chi Square calculators, however. (Maybe it even is a chi square calculator.) A Chi Square calculation puts the test at 96% significance.

Just like the chi square test would have likely correctly predicted the winner of your example test, even though the actual conversion rate of the variable hadn't been settled yet.

But there are different methods for finding significance. (That may be why Ken calls it arbitrary.) How do you determine 95% significance?

But if we're going to use a yardstick of sorts, chi square is one. And I think there's enough data to make a decision.

If I was running split tests, I'd start a new test.

Cheers,
Stephen

Hey Stephen,

I'll keep my reply short so I don't confuse the 3 remaining people following our conversation.

My East Multi Tracking testing software uses the Chi Square method to determine significance.

And I agree... I would have run a new split test where I'd break the variations down even further to find the true conversion boosters.

Take care,

Mike

[yankforlife41]

Belcher Button has always increased conversions for me when I've worked it in.

[AdamWagner]

Interesting study! I like the original button because I think people resonate with having the card options underneath that way they know their card is accepted.

[MegaWealthX]

Nice story! Thanks for sharing this with everyone.

[sprintingwriter1]

Isn't it amazing that when you change one technique, you can double your income?

[Patrick Warren]

Cool info thanks for sharing

[Fuad Tolieb]

Thank you so much for this, Im going to launch my product soon and Im going to split test the 2 as well. Again, thanks!

[Drewry_Media]

Split test is what they always say, and yes, it is surely the truth. It goes as far as anything when it comes to page optimization. If you place certain things in certain places on your site, it becomes simpler for the site user to access it, thus, you collecting a bigger retainer, for making it simple for them.

[ewenmack]

It's been very cool reading all your posts of thanks
and expanding further on the subject of testing and tracking.

You guys rock!

Ewen

[wakeup75]

Ewen, trying to get in touch with you don't have enough post to PM could you PM an email address or another way to contact you.

[ewenmack]

Originally Posted by wakeup75

Ewen, trying to get in touch with you don't have enough post to PM could you PM an email address or another way to contact you.

Contacted.

Best,
Ewen

[scott william]

Great story thanks for sharing and thanks for all

[Len Bailey]

Originally Posted by Murlu

I personally don't like the belcher button but I can't turn my head to the effectiveness. I'll need to update my buy button on my new PLR site because I went with just a simple one - the credit card options are, in my mind, what really help - adds trust.

I know what you mean. The first time I saw one I thought, "Damn, that's ugly." But like Rule #1 says... TEST EVERYTHING. Sometimes, the results surprise you.

[skyla]

Just read this great lesson on testing, thanks for the info. It was a really interesting read.