Question about Split Testing on Bemob

[gabe86]

So I have Traffic Junky as my Affiliate Network and using Bemob as my tracking platform

For the past Month I have been keep up with my conversion. I only got 13 conversion in total and written it down in my Excel

Traffic Junky campaign strategy :

I named a campaign name Original. Which as been getting me the conversions I have notice in Bemob platform that affiliate offer Landing page 1 has the most conversion. 2 banners ( as example banner name ABC ) have got me conversion twice and 2 others only got me one conversion one of each. All of the OS is Chrome. Position type : PH - PC NTVA. All Desktop and 1 Mobile.

My next step

I created another campaign in Traffic Junky called Winner Conversion which i have changed the setting and used all the information i have gathered from the original campaign just focus on them such as keep the 2 banners, focus on Chrome only. Change the keyword to one keyword etc.

After a week just running that alone and pausing the original campaign I have only got one conversion?

I thought the idea is to try to narrow it down on what is making me the conversion. And supposedly get conversion daily? Doesnt seem to be in my case

What I have done now is focus on getting another landing page and adding 2 more banner to my Winner Conversion Campaign

Am I doing this right ?
Is there a specific process on how to split test for adult affiliate in traffic junky? For example. Do I focus on banners more then I do with landing pages. Do I start with 5 banner and 2 landing page.

A/B Landing page = B landing page make the most conversion

A/B/C/D Banners = A/C make the most conversion

Create a new campaign using

B Landing page create another landing page to split test =B/C ... just want see what convert better

A/C Banner create another Banner = A/C/E/F .... just want see what convert better

if so when do I finally stop split testing.

Do Should I just stick with one campaign or do i mean new version and keep the original campaign while i run the newer ones ?

please keep your answers simple for me to understand. Which I'm still new to this industry. Thank You

 

[Graybeard]

campaign 1, LP1, page loads (CTR),sales (other CPA event) conversions (postbacks?) divide sales/page loads
campaign 2 ......
campaign 3 ......
you can set traffic variables (geo, time of day, device, browser version) differently for the A, B and C.

That's the least confusing way

Banners
clicks/impressions
100/30000 (or whatever)
if you are looking for banner affinity
the best is the one with the most sales but you really need over 500 clicks for a sample

 

[gabe86]

Something like this ?

campaign 1 LP1 and LP2, Desktop, Male, US, kw: All ( example) , OS : Window, Position: All, Source : All. language English... Banner A/B/C
campaign 2 LP2 and LP3, Desktop, Male, US, kw: All ( example) , OS : Apple, Position: All, Source : All. language English... Banner A/B/C
campaign 3 LP1 and LP3, Desktop, Male, US, kw: All ( example) , OS : Linux, Position: All, Source : All. language English... Banner A/B/C

After 500 conversion : Campaign 2 made the most sales. Keep L2 and change the header wording and keep banner A/C because they performed better. Dont narrow down just continue to split test the Landing pages and Banners

Is the key to make conversions every day or just make conversion even if it every other day ?

I'm just focusing in the US for now to start off with until I get the hang of this

So what your saying is I have to have 500 conversion to really start narrowing down on campaign and banners.

 

[gabe86]

Your right about he time i have notice my conversion come around 3pm to 11pm. i have got one 5am.

 

[Graybeard]

Yes basically. US traffic is lots of bots and finicky. I avoid after school hours you can narrow the traffic down to the states and use day parting to accomplish this. Best hours for adult are 6-8 and 20 - 2 in that time zone. pay attention to the time differentials --getting that right can be a real cluster fuck.

\   traffic          /
 \  impressions     /
  \   clicks       /
   \   sales      /

Landing page clicks or banner clicks first
500 sales conversions will take time.

 

[VeronicaWoodruff]

A/B testing, also known as split testing, is a marketing experiment wherein you split your audience to test a number of variations of a campaign and determine which performs better. In other words, you can show version A of a piece of marketing content to one half of your audience, and version B to another.

 

[gabe86]

A/B testing, also known as split testing, is a marketing experiment wherein you split your audience to test a number of variations of a campaign and determine which performs better. In other words, you can show version A of a piece of marketing content to one half of your audience, and version B to another.

Okay I understand that now when do I start to narrow down the setting inside the campaign that is preforming better ? Such as ISP, Keywords, Position, OS etc.

 

[gabe86]

I forgot to mention about keywords. When I find out what keywords are working, do I start to narrow it down. I realize some keywords are repeating. Do i start to only use the kws that i see repeat itself more than once? Is the rule of thumb only select 4 kws and only use exact?

i attached a screenshot of the keyword from bemomb. TF told me the reason I have so many keywords for each winning conversion is that the person went thru all those keywords before clicking on your banner?

Whats the point of type "60fpt" ( for example ) the keyword setting for the campaign? if it not going to target just that specific keyword

 

[gabe86]

What does 'conversion' mean in this case:

1. a click on a creative, or;
2. a sales conversion event on another external offer?

Conversion as in I got a commission ...

 

[gabe86]

Sorry not sure I understand your question on the first response. Are you asking me what the setting is it understand new traffic source?

If I change this what will this do ?

 

[Graybeard]

FPS is frames per second
Those are the keywords for each page?

So I have Traffic Junky as my Affiliate Network and using Bemob as my tracking platform

Traffic Junky is an in house ad network
Those keywords ? not sure. Looks more like what was on one page by relating the text of the referring page?
Converted for what and external offer?


What was your traffic loss %???

 

[gabe86]

Do you mean this ?

 

[Graybeard]

I only got 13 conversion in total and written it down in my Excel

What does 'conversion' mean in this case:

1. a click on a creative, or;
2. a sales conversion event on another external offer?

 

[gabe86]

I just got another commission just now... the only problem I noticed in bemob did notshow any keywords that I added in my campaign here a screenshot what I have.

But if you look at bemob none of those keywords are in the tracking platform ---->

What should I do:
A. Just run my campaign with all keywords without being to specific
B. Write down on excel which keywords that are show up more then once
C. I'm doing it all wrong ha-ha

 

[gabe86]

yes external offer

so are you saying after few hundred conversions just keep track of my keywords for the ones that are repeating itself more and more? just set it to "exact"

 

[gabe86]

then*
yes sample size is needed

i just added some those keywords because i figure it common sense since it was related to my offer. i was just really surprise to hear the TJ support to tell me that it took all those kws to search up my banner. Not sure why they have that settings that way. it confuse someone. it would make more send to leave the kw option blank. then run it for a month and see what exact keywords got me the commission and then just save it then add it to my kw list inside my campaign.

 

[gabe86]

in been a 2 months now when I"m been trying to split test different landing page looks like landing page 1 and 3 seem to be working better. i been getting conversions once a day or every other day. I thought adult AM i would expecting to make 3 or 4 + conversions a day? i have increase my bid by .10 cent. I'm still testing the split test phase ( changing the title, sub title and button color ). What am i doing wrong?

im afraid to increase my bid to .50 just because i might be looking profit and not gaining conversions

 

[gabe86]

Not

Percentages not numbers.

How long is the string on my desk?

Sure I understand what you mean...sorry can you elaborate

 

[Graybeard]

Just wondering I had an issue where Adultforce was only counting 66%~ of the traffic I KNEW I was sending. I log ALL clicks out.
I just wondered if that tracker derives some value for a traffic loss % inbound too (buying) over-counts?

What are you converting --some external offer? Or, an internal one?

As far as keywords: until you have a few hundred conversions to segment keywords I would just run it as is.
If you segment now I think you would limit your net/funnel too much. Too small of a sample.

 

[gabe86]

4000 clicks 20 sales
(20/4000)*100
.5%

are you saying I change it to .5 on my bid? base how much clicks and sales I made using simple math is that formula every time ? I thought the ideas was to make ( example $20 daily or more ). This almost seem like a slow process

looking at my math I made 10 sale and 482 clicks which I would have to change my bid to .02.

When do I start to do the math every month or ever time I run out of funds?

 

[gabe86]

0.5% is a percentage calculation.
(10/482)*100 = 2.07%
If your win rate in the placement's visibility is less than 15% - 25% up the bid and see if it scales OK and you have a positive ROI on the ad spend.
It's more complex if the payout is an upsell or a revenue share percentage --then you need to evaluate a ROMI (return on marketing investment) 'revenue share/customer acquisition cost'

okay I see what you saying. i remember seeing a video that did talk about if someone is making 150 each month but hasnt notice any increase. they should increase their bids from as example .09 to .15

I figure there was a math formula for AM but I didn't know it was (sale/click*100 =) thanks for telling me

I did talk to TJ but they always. They always tell me to start at 1.00. I don't believe that is right for a beginner like myself. I feel like I will waste more money.

I just want to make sure I'm not over spending but at a least making money for profit every day.

 

[Graybeard]

then*
yes sample size is needed

 

[Graybeard]

Percentages not numbers.

How long is the string on my desk?