AS3-Jon_CorrellThe Action Seminar speakers answer questions from the audience. In this excerpt, Jon Correll explains how to split test a page if you are getting traffic from different sources

Jon’s comments:

Is it a valid statistical test if I am doing an on-site split test and then I am mixing up my traffic, either by source or I am split testing my ads like over at adwords or I am introducing a new ad?

So, you are doing different types of traffic. And you are asking if it is… So, this is one of the things our tool does, and I believe EOW and Optimize now have this ability…to do segmenting. Segmenting is very powerful. You want to know the difference…you have to look at it from a traffic point of view, because we have seen clients with traffic that converts at 75% from one thing and 1% from another.

So, if you mix that together, you are going to get a lot of noise, and a lot of lies, right? The data is going to lie to you, because you are not looking at it correctly. So, you have to actually drive in and say…and most tools are pretty easy…you just say “what is the conversion rate on this source of traffic?” As long as you are tracking that, which you should be doing, and most tools do it fairly easily, you need to look at it statistically from that point of view. Otherwise, you are going to get lied to all the time. Does that make sense?

Yeah, although my question is…say the split test is running and I introduce a new ad. Like, say I am split testing over at Adwords. One of the ads wins. So, I introduce a new ad. Or, should I wait until the onsite split test is done.

How much money do you have? You have to weight these things. If you get technical about it you can say “what was the day of the week when it ran?” “When were the clicks coming in?” And this ad, and this. You start adding it in and pretty soon you will make it so compliecated that you will say “I give up. I am going to go be a life guard for the City of San Diego.”

It is a beautiful job, by the way, if you can get it, and they pay very well. I have seen the salaries. But, you are going to give up, because it is very complicated. So, I just say “simplify it. Don’t worry about the top end. Measure that separately. Worry about the bottom end. I do believe in splitting the traffic types. But, when you are talking about switching the ads. And, usually, with Adwords, you have several ads running, and it is automatically doing that, and, no, I don’t believe you have to do that. If there are radical differences between what you are messaging, and what you are driving to the page, yes, that is going to make a difference.

Right? But then you can also track the campaign ID. The ad ID. So, those are some things you can do. You just have to look at it and play it by ear, but I try to believe in simplicity, and I would ignore it as long as they are similar. And, then I would also say, in my little lab coat, “hey, that is a lie, Jon, you are lying to him, the stats are not true”.
But, if they are pretty close, go with it. Otherwise you will sit there in paralysis and analysis. “But I don’t know what to do…it’s too complicated”.

Don’t worry about it.

For more information about the Action Seminar, go to

If you like this lesson (and are not already a member of the Marketing Rebel Insider's Club), you will be glad to know that there is a vast library of small business marketing lessons and training programs just like this... a private Facebook-like chat stream where you can get answers to all of your questions about marketing tools and techniques, small business growth, and life as an entrepreneur...

...which is only available to members of the Marketing Rebel Insider's Club.

You can access all of this immediately for less than $1/day.

You will see for yourself why our members love this online club so much...

... and why they don't invest in marketing tools or strategies without first checking in with The Club.

Click Here to start your membership today.

We'll see you on the inside...

Stan Dahl & The Marketing Rebel Team

Powered by WishList Member - Membership Software