About us
Udimi is a Top 1 marketplace to buy or sell email solo ads in the world. To guarantee the best traffic quality to buyers and fast payments to sellers, we stand between parties and control every step of the process.
Udimi is a registered trademark of Udimicom, Limited. Ifigeneias 14, 3036, Limassol, Cyprus
Secure payment methods
App rating 4.8 • 10,000+ users
Thanks. One minor correction I should make. Two choices of three parts is 2^3 = 8 combinations... not 16. You will however need to run your 4 experiments twice, once each at two different ad spend levels. This is how you obtain data to plot the function on a graph where the X-axis is the amount you spent and the Y-axis is the amount returned. You are looking for the highest slope to maximize your ROI. Good luck. J
Thanks. One minor correction I should make. Two choices of three parts is 2^3 = 8 combinations... not 16. You will however need to run your 4 experiments twice, once each at two different ad spend levels. This is how you obtain data to plot the function on a graph where the X-axis is the amount you spent and the Y-axis is the amount returned. You are looking for the highest slope to maximize your ROI. Good luck. J
Hmmm, over a month and no response. I infer one of two possible situations:
a) nobody knows the answer or,
b) nobody is willing to share the secret. LOL
To work it out for yourself I suggest starting with the concept of Ideal Function. Think of your ad spend process as a machine. Put dollars in, get dollars out. Of course you want some multiple of dollars out to make a profit.
And there are some internal parts of the machine that work better than others. Your job is to test the parts to see which ones work best and build your machine around those. Does that make sense? Pick the best parts, build the best machine.
When I say "parts" I mean things you have control over. For example the type and design of your ad, where you place it, who you place it with, WHEN you place it, how often you place it... anything you can think of that might reasonably affect your your results.
When I say"test" I mean you must be able to track your results from the different parts you are trying. You must test according to a plan. Below is a simple plan called an L4 orthogonal array.
See attach
P1 thru P3 represent three parts you may choose to test. Using this plan you are able to test 3 parts at 2 levels each. For example, maybe you want to test 2 different ads, placed with 2 different publishers, on 2 different days. Use the L4 to decide which combinations you need and do ONLY four experimental runs. The results give you enough information to analyze which ad, which publisher and which day produces the best results for the money you spend. You don't need to test one part at a type. You don't need to test all combinations (16 possibilities in this case). Four is all it takes. That's the power of the orthogonal array! The final step is to do a confirmation run to check your results.
Yeah, I left out the details of how to do the analysis. It's a bit too technical to describe here. You might be able to google it or hit me up by email. Hope it helped some.
J.
Hmmm, over a month and no response. I infer one of two possible situations:
a) nobody knows the answer or,
b) nobody is willing to share the secret. LOL
To work it out for yourself I suggest starting with the concept of Ideal Function. Think of your ad spend process as a machine. Put dollars in, get dollars out. Of course you want some multiple of dollars out to make a profit.
And there are some internal parts of the machine that work better than others. Your job is to test the parts to see which ones work best and build your machine around those. Does that make sense? Pick the best parts, build the best machine.
When I say "parts" I mean things you have control over. For example the type and design of your ad, where you place it, who you place it with, WHEN you place it, how often you place it... anything you can think of that might reasonably affect your your results.
When I say"test" I mean you must be able to track your results from the different parts you are trying. You must test according to a plan. Below is a simple plan called an L4 orthogonal array.
See attach
P1 thru P3 represent three parts you may choose to test. Using this plan you are able to test 3 parts at 2 levels each. For example, maybe you want to test 2 different ads, placed with 2 different publishers, on 2 different days. Use the L4 to decide which combinations you need and do ONLY four experimental runs. The results give you enough information to analyze which ad, which publisher and which day produces the best results for the money you spend. You don't need to test one part at a type. You don't need to test all combinations (16 possibilities in this case). Four is all it takes. That's the power of the orthogonal array! The final step is to do a confirmation run to check your results.
Yeah, I left out the details of how to do the analysis. It's a bit too technical to describe here. You might be able to google it or hit me up by email. Hope it helped some.
J.
Hi Guys
I'm new to this game and want to learn about what works best for squeeze page design.
While I don't know much about that yet, I am pretty good at applying scientifically planned experiments. I am about to do a run to test things like pre-headline, headline, sub-headline, colors, image, call to action, form position, and form layout.
The thing I'm concerned about is the fact that this type of testing assumes the factors are independent variables. In other words, no interactions. Nothing like the case where a blue page is good, a female face is good, but a blue page with a female face is bad.
Are there any such situations this group might know about?
Hi Guys
I'm new to this game and want to learn about what works best for squeeze page design.
While I don't know much about that yet, I am pretty good at applying scientifically planned experiments. I am about to do a run to test things like pre-headline, headline, sub-headline, colors, image, call to action, form position, and form layout.
The thing I'm concerned about is the fact that this type of testing assumes the factors are independent variables. In other words, no interactions. Nothing like the case where a blue page is good, a female face is good, but a blue page with a female face is bad.
Are there any such situations this group might know about?