Home > Paid Traffic Sources > Mobile

---

help with testing landing page strategy & multivariable testing? (12)

---

Hi Stackers,

Got a question about testing landing pages and testing multi variables at the same time.

Let's say I have a landing page that is performing the best out of my group. Let's give the CTR, CR & CV Stats like this:

CTR / CR / CV
25% / 5% / 1.5%

The format of the winning landing page will be like this:

[Headline]
[body]
[CTA]

So let's say I make 4 variations of the winning landing page.

[Headline 1] - 20% / 6% / 2%
[body -same]
[CTA - same]

[Headline 2] - 15% / 8% / 4%
[body -same]
[CTA - same]

[Headline 3] - 15% / 4% / 8%
[body -same]
[CTA - same]

[Headline 4] - 25% / 4% / 2%
[body -same]
[CTA - same]

Then I choose the best headline according to the data. Which in this case would be Headline 3 because it gave me the highest CV? Then test body text and CTA. Is this the right strategy?


But a little more advanced, how would I test multiple variables at the same time?

Would it be like this?

[Headline 1]
[body -1]
[CTA - same]

[Headline 1]
[body -2]
[CTA - same]

[Headline 1]
[body -3]
[CTA - same]

[Headline 1]
[body -4]
[CTA - same]

[Headline 1]
[body -1]
[CTA - 1]

[Headline 1]
[body -1]
[CTA - 2]

and so on?

---

bump

---

bump

---

I can't really help you with this one, have no experience with the topic.
The reason is that my margin of mistakes and other stuff going wrong / the time spent analyzing all the data is high enough already. This kind of testing is something I might get into once spending a lot more money, or having some employee well trained who analyses the data for me. Until then I just do plain simple A/B split-testing.

Would be interesting how more experienced guys are doing this!

---

Bump..bump bump bump. Bump *while singing sisqo's thong song*

---

bump

---

What was the profit on each of the 4 varations?

Was it the same traffic and bids on each of the 4?

---

I don't have a specific case in mind I'm curious about the system or strategy behind testing LP's and testing many things on a LP at the same time...

---

I test headlines first
Then I test a variation of the first paragraph
Then i test the image

if we're driving mega volume, lets say pop traffic, we might generate variation combinations of 2 or 3 things.. but that can be 100s of variations, but if u do a million+ visits per day u can get data ultra fast....

we usually hard code the stuff into our lander, and then call it using PHP
my coder at bannerslanders.com wrote so i just do this ?img=48&h=4&tx=8

that means show image 48, headline 4, tx 8

for example

---

If you are testing one independent variable, you should keep all the other independent variables constant. You need to test them to statistical significance so that your sample size is large enough. Your dependent variable should be whatever you are trying to maximize. What are you trying to maximize? CTR? CR? CVR? Profit? ROI?

Before you try to carry out multivariate testing, I think you should learn how to do very basic univariate testing first.

---

^^ this

If you want to do multi-variable testing on mobile traffic then you will need to run a TON of traffic. Apart from testing all the variables on your lander you will need to add to that all the mobile variables as well.

This is great if you want to fine tune a good campaign. If you want to do it off the bat with a fresh campaign then you will go in the hole quite quickly.

Start by testing 2-3 different angles. This is by far the most influential. Once you find an angle that's a clear winner then start refining your tests

---

Thanks for the help guys. So I will put multi-variable testing in the back of my head once I start running mega traffic

---

Home > Paid Traffic Sources > Mobile