Home > Questions and Answers > General Questions

Split testing means competing against yourself? (8)


04-03-2014 03:11 AM #1 zd__rd (Member)
Split testing means competing against yourself?

I'm realizing that as I am split testing, I am actually competing against myself. Is this inevitable when split testing on the same traffic source? By changing only 1 variable such as banner image, from campaign 1 to campaign 2, I am now competing against myself. Do any of you guys avoid this buy testing one, then taking it down, then testing another? That would just take ages it seems...

Example:

Campaign 1: banner 1, US, TJ, NTV A
Campaign 2: banner 2, US, TJ, NTV A

variable changed was banner image, all other variables remain constant, but now I am my own competition, making the CPM bid even higher for myself, therefor skewing the results of the split test!


04-03-2014 11:16 AM #2 craigm (Veteran Member)

why are the 2 banners in separate campaigns?


04-03-2014 11:42 AM #3 zeno (Administrator)

There is going to be a chance of self-competition on every traffic source when running multiple campaigns. Sometimes even ads within the same campaign will 'compete'.

However, competing by being shown at the same time isn't necessarily a bad thing.

In that one moment when a user looks at adverts and thinks about clicking, you now have multiple chances of convincing them to click.

On the other hand, on most high volume traffic sources your chances of self-competing are low unless you are dominating all the traffic.


04-03-2014 01:16 PM #4 kamaleon (Member)

Why don't you just add banners to a single campaign as in the getting started guide that caurmen has written?

http://stmforum.com/forum/showthread...TION-AND-INDEX


04-04-2014 12:28 AM #5 zd__rd (Member)

@craigm and @kamaleon, I am using separate campaigns for each banner because I am using Exoclick, and I have read multiple times on this forum and experienced as well, that it is not in your favor to have multiple banners in the same campaign specifically on Exoclick. Another reason is because I do not know how their rotation algorithm works, I would want to know precisely how it's choosing which banner to rotate.. how often each banner will be rotated? is it evenly rotated? is it based on user reaction or click-to-impression ratio? etc, etc.

By using separate campaigns, I have eliminated this unknown variable from the equation.

For example, say I have 1 campaign with 10 images, I will get a TOTAL impression count. So how would I know the impression count of each banner? I might make a poor decision based on not knowing the impression-to-CVR rate per banner. Traffic source impressions are obviously not a statistic I can see in my tracking software, so AFAIK, there is no other way I can think of doing this. Please do share any insights/knowledge you may have! I am well aware that I may be doing this completely wrong as I am very new to this, it just seems logical to me.

I would love to hear the perspective of a well-rounded Exoclick advertiser who knows more about this!


04-04-2014 12:35 AM #6 zd__rd (Member)

@zeno,

In that one moment when a user looks at adverts and thinks about clicking, you now have multiple chances of convincing them to click.
I don't really see it that way. Don't you think this could skew the data? I may be micro-optimizing at this point by thinking that, but I do have banners very different from the typical ripped stuff you see everywhere, So I feel that having 2 of my own unique banners stacked on top of each other could bring a different effect to the user VS having 1 banner they are blind to, next to my banner which is much more unique. More simply, I want my banner to stand out to the competition, and all the usual crap people are blind to. If I fill the page with banners that are all the same style, I feel that they will more likely be 'blind' to them because there is nothing to compare it to in that given moment!

Good point about the low likelihood of self-competing, because you are right, I am not dominating the traffic source at this point.


04-04-2014 03:15 AM #7 zeno (Administrator)

We can conjecture how it might go either way... i.e. a positive impact or negative. However, as I said it's probably a negligible worry until running high volume relative to the placement(s).


04-04-2014 10:28 AM #8 andyvon (AMC Alumnus)

The Exo rotation algorithm you are talking about only has effects when you run CPC because they optimize impressions in order to get the highest eCPM.

If you run CPM, you should get a pretty even traffic distribution, so no reason to micro manage banners in separate campaigns.


Home > Questions and Answers > General Questions