vitavee

I've just started a new campaign to promote a baseball digital product.

I'm going to explain what I did and show you my pages and ads. If you have any question, please feel free to ask, I'll be happy to answer them.

But first, I need to mention that these are all basic stuff - so if you don't learn anything new don't bite me That being said, I believe that the foundations are critical and this is what many affiliates, mainly beginners, are missing.

Before exploring the details, here's a general overview of my initial funnel:



Let me explain what is happening here... let's analyze each part of this funnel one by one:



The red circle at the top left represents the incoming traffic. The first thing I do is sending the visitors to 1 of 3 different optin pages (A/B/C split test). When I advertise on FB, I *ALWAYS* capture the email addresses before anything else - when I have those leads I can promote to them anything under the sun without having to check if it's against FB's TOS or not.

Here's the 1st optin page:



The 2nd one - the background is a video showcasing the product I'm giving away:



And the 3rd one:



After someone optins, there's a test to see if the traffic is coming from FB. If yes, a conversion pixel is triggered.



In both cases, the lead is then redirected to a rotator that will send him to 1 of 2 different thank you pages (A/B split test):



Those 2 thank you pages have links to affiliate offers.

Let's look at the 1st thank you page - as you can see on the screenshot below, 3 actions are possible on this page:



The 1st action is an affiliate link leading to the sales page of the product I'm giving away. People are impatient - some of the ones entering a giveaway think they will not win and decide to buy the product without waiting.

The 2nd action is an affiliate link to a digital product called "Baseball Swing Mastery". It's a video course sold on Clickbank.

The 3rd action is also an affiliate link to another digital product called "PRO Hitting Mastery". It's also a video course sold on Clickbank.

Even though that thank you page has 3 different affiliate links, only 2 will be displayed at once. Action 1 + Action 2 or Action 1 + Action 3. The goal being to find which digital product converts better.

Before I show you how the thank you page looks like, I need to mention that there's also a contest presented to the leads at this point. They are given a unique referral link and earn points when they share the giveaway with their friends. When they subscribe they earn 1 point. When they have one of their friends subscribe through their referral link, they earn 10 points. And when one of their friends refer someone else, they earn 5 points.

It's all done to motivate them to share with as many people as possible.

So here's how the page looks like - in the version below, Action 1 and Action 3 are shown (sorry I had to scale it down as it wasn't fittin vertically):



And 50% of the time, the banner at the bottom was the one for Action 2 instead:



The 2nd thank you page was just a variation of the 1st one, where I just swapped the top banner with the bottom one to see which one would bring the most revenue:



Very quickly after starting sending traffic, I could see that the optin 1 was performing poorly and had only 3% chances of being the best:





So I eliminated it from the split test to know faster which page between optin 2 and optin 3 would perform better:



Shortly after I could see that optin 2 had 99% chances of being better:





So I cut optin 3 from the split test to send all the traffic to optin 2 (the one with the video background).



Continue reading below (I've reached the max number of images in this post)...

FunnelFlux - Stop Tracking Clicks, Start Tracking Funnels

vitavee

At this point, I should have added a new optin page to try to beat optin 2 but I didn't. I was busy in another part of the funnel. I noticed that people reaching the thank you pages, barely clicked on the affiliate links.



So I changed the bottom part of the funnel to this:



Instead of sending them to the leaderboard pages with the top and bottom banner, I split tested 3 new pages instead.

This one corresponds to the "Hurricane 4 Confirm Email" node on the left part of the funnel - the top bar is animated:



It has 2 possible actions. The banner to the product I'm giving away, and the video screenshot that either redirects to "Baseball Swing Mastery" or to "PRO Hitting Mastery"

Here's the 2nd version of the confirm page, corresponding to the node "Hurricane 4 Confirm Email 2"



And here's the 3rd version, corresponding to the node on the right "Hurricane 4 Confirm Email 3"



As you can see below, those pages performed MUCH better than the previous ones and the click through rate was great:



I still kept the leaderboard pages. But instead of sending the leads to those pages after they optin, I sent them there in the welcome email they receive. That way, I could still explain to them that they can earn more points by sharing with their friends.

Here's the welcome email, if you wonder how it looks like (real urls hidden):

----------------------------------------------------

Subject -> Hurricane 4 Giveaway: Increase Your Chances of Winning!

Thanks for your interest! Your entry is now confirmed.

BUT are you THE one who is going to get the Hurricane Category 4 Solo Swing Trainer?

You currently have 1 point.

You can dramatically increase your chances of winning by following the 2 steps below:


STEP 1: Send an email to your friends and give them your personal referral link:

xxx.xxx/hurricane4?kid={social_id}

You earn 10 new points for each friend you refer.

And you earn 5 additional points each time a friend you have referred refers someone else!


STEP 2: Share the giveaway with your friends on your favorite social networks by using the buttons below:

{% share_buttons %}

These links were created just for you. Using them will credit you for anyone you send us.


You can check the leaderboard and your current score at any time by clicking **this link**.

Have an awesome day!


Brad Hudson
xxx.xxx

PS: The winner will be selected based on a weighted draw. Everyone has a chance to win,
but the more points you earn, the more chances you have.

PPS: Do you want to learn or teach your son to hit like a PRO? **Check out this video now**.

----------------------------------------------------

To be continued...

FunnelFlux - Stop Tracking Clicks, Start Tracking Funnels

savidge4

First off GREAT POST! I love the fact you started at the entry of the funnel and then worked your way back. This is something I preach any chance I get. Being someone that does a ton of this stuff, there are many things you have done right, but a few spots I think maybe could be done better to create overall more effective testing.

I personally would have chosen a "Path" and run 100 impressions across it. So I would have picked a Opt-in a thanks and then a follow up series. Doing this creates a TRUE "Control" This becomes the measurement that the rest of your testing must meet and hopefully surpass.

This does a few things for you. #1 it gives you a set of numbers as a base. If you have 3 pages to be tested, without a control, you are comparing one to the other and the next, and you don't always you don't always have a clear line of what statistical significance is.

The next thing is running these test to a set value. More often than not you would run a test based on time for the most part. A week or 2 or 3 OR you run them to a set value of impressions. Because of the low number of impressions in your test.. I would have run them out to a set number. we will say 100 as an example here.

If we look at your Optin numbers and do some quick math we can determine that your "worst performer" was 7 impressions shorter, AND 7 Conversions short of being a contender. But if we are running this to 100, that makes it 10 impressions short and 7 impressions from being a contender.

At that point you could have had 2 optins that meet the math requirements of statistical significance. That then means optin 1 and 2 are equal in conversion power.. and testing is the process of finding "Better" so there would have to be further testing and changes made.

If we go back and look at the removal of optin #3... This is not something I would have done. I see this all the time, that a lager later in testing ends up being the winner. it simply happens A LOT.

the other thing I am noticing reading your post. you are simply concerned with the click percentages... what about the sales numbers on the back end? Ultimately this number is far more important. When you go that step further and you start optimizing for Revenue, they dynamic of your testing can change a bit. Click thru winners may not always be revenue winners.

Below is a great article by Neil Patel that discusses many of the points I have hit apon and why you would or would not do them.

What Spending $252,000 On Conversion Rate Optimization Taught Me

Again, @vitavee GREAT POST

vitavee

@savidge4 thanks a lot for your insightful reply. Also for Neil's article, I enjoyed reading it. Testing for macro conversions instead of micro conversions definitely makes more sense.

Concerning the statistical significance, for my tests I've been using a calculator based on the Bayesian Statistics algorithm plus the Monte Carlo method.

(source and calculator)

When the Bayesian calculator shows me that a trial has less than 5% chances of being the best, I cut it. When it shows me that a trial has more than 95% chances of being the best, I declare it as the winner. I get results with a pretty good degree of confidence much faster and way cheaper than if I had to run traffic until I reach statistical significance.

The mathematics behind it is quite complex, and to be honest I didn't try to understand it, but if you're interested, here's an article explaining it:

Bayesian A/B Tests - RichRelevance Engineering Blog : RichRelevance Engineering Blog

It is also interesting to note that pharmaceutical companies use Baye's Theorem to be able to get meaningful data from small datasets.

FunnelFlux - Stop Tracking Clicks, Start Tracking Funnels

savidge4

I get what you are saying... Cost of running a test and the time needed to run a test ARE of great concern.

When using the Bayesian Stat algo you have to understand in this case there is no NULL Hypothesis involved. In other words... if you have a page that converts 50 times out of 100 and the other 48 times out of 98 It will say that the first option is the winner.

If for example you are using the two-sided Chi-Square statistic equation with the above numbers... your test is registering a .11% confidence level. Simply not something that I would want to count on.

If you read: 12 A/B Split Testing Mistakes Businesses Make All The Time you start to get a better idea of what can happen when you cut corners on testing. The article makes a very valid argument for doing testing in week long periods. It also makes a valid point for Statistical significance.

As much as you are seeing Positive results from your testing, there are many indicators that may suggest they could be short lived. I would rather pay for accurate significant testing, than pay the price for approximate probability testing in the long run.

Again, I love your process, I commend you greatly for the time and effort in posting your process. Being someone that basically does this stuff for a living.. I want to see guys like you getting the results, testing can achieve. But if you are happy with your process and your results.. that's great!

vitavee

That article from ConversionXL is really good. The strongest point I got from it that I didn't think about is the conversion rate differs depending on the day of the week. For that simple reason a test anything shorter than a week can't be valid. Very good point.

I'll definitely have to review my process. Thanks for this!

Are you doing this for yourself or for clients?

FunnelFlux - Stop Tracking Clicks, Start Tracking Funnels

savidge4

I do this for both.

Here is a thread where I helped a friend of mine living in AU do some testing. http://www.warriorforum.com/conversi...l-savidge.html Actually I just designed one of the landers for him, and watched as the testing played out. The testing was done over a 1 month period... and when you look at the data on this test you will know why I mention that. ( really low traffic )

Look specifically at the last 2 posts in the thread. They are 18 days apart. Look at which page was in last ( in the second to last post ), and which ended up being the winner ( if you really could call it that - based on the numbers ). this could have been one of those throwing the baby out with the bath water moments...