I think you should test to find the converting offers first, not all of them convert. Without conversions, there is no point doing split-test. Once you have a seemly best converting offer, you use it to split-test landing pages, once you have your best landing page, you go back to test all the converting offers to have a winning combination.
I'll imagine if you test everything as your layout in the diagram like that, it will cost much $$$. Did you go to AFFcelerator?
Hello timothy.
I guess we need to define what split testing means first.
You can test things at different levels, but the true meaning of a SPLIT test is to test several variations of something that is related in some way.
- You can split test banners that are in the same niche/vertical, you can split test offers that are in the same vertical or at least cater to the same demographics
Or you can ignore part of this and split test different niches or verticals etc ... but you need to keep the other factors unchanged.
- You can split different verticals or niches or offers ... but just on one placement or traffic source ... and never all at once, that's not a split test anymore.
To put it in a simple way, the result of a proper split test should be a winning "something" (offer, ad, LP) when the other conditions (factors) are kept intact.
For example ... choose a GEO, choose a traffic source and split test 10 offers over there to find out which OFFER converts the best.
In your example, you want to test 3 GEO which alone makes a split test impossible, since you will be using different LPs for each of them, so let's just call it a general TEST.
You will need to create 3 separate campaigns for this, targeting several GEOs in one campaign is nonsense ... each GEO has different bids, different placements/sites, different demo breakdown etc ...
Each of these campaigns will have 3 LPS and 3 Offers ... set these up in your tracker and give them even rotation %. Each campaign will have 9 possible paths, so your math was correct, in total, there are 27 paths to test, 9 per campaign.