Hello!
I would like to ask you about your split testing methods you're using with Facebook.
1. What's the first thing you split test?
2. Is it good idea to splittest for example AGE, IMAGE and PLACEMENT in first ad run? in results ex 30-50 ads on first run.
At the moment I'm just split testing in few runs. First I test images, then if I have some winner I'm creating new campaign and split testing age/placement/gender and etc.
Is it good way to do this?
I'd say it really depends on what you're pushing on FB. I'm currently running ads for some ecommerce sites and this is how I split-test:
1. The very first thing I split test are angles. Trying to find out whether fear, hope, avoidance of loss etc. are the best approaches in general. This includes extensive research and building out a potential customer profile, once you know what your customers are after, you tailer the angles to that, individual ad copy for each angle. After I have the angles ready, I do further research based on the customer profile + angle to find out the type of targeting/demographics/interests/behaviours that might work in theory for that particular angle.
Images are indeed the biggest factor, but I found that if you have a very solid product + extensive research before you launch anything, I would always test the angles first as they make the biggest difference later on. I split-test images while testing the angles.
E.g. Hope - "This product is super-awesome-amazing and can benefit you in that way" (2-3 images) vs Fear - "Does this happen too often? AVOID these issues asap with our product" (2-3 images) vs a combination of both (show the painful bits and the solutions all in one, 2-3 images) vs [insert client testimonials here](2-3 images) etc., anything angles related you can figure out after doing your research.
Using the above I manage to land 8-10 relevance scores right off the bat on most ads. It takes much longer to launch a campaign instead of quickly throwing stuff at the wall, but the time investment pays off.
2. Yes, I split every single ad and variable into its own campaign (not ad group) because I prefer proper split-tests and no skewing of the data whatsoever - aiming for the highest accuracy/even stats for every ad variation. I usually separate by age groups where every age group has the given angle, and a whole bunch of images for every ad. This results in a million separate campaigns and is somewhat hard to manage, but covering everything gives me a lot more confidence and tons of insights about your targeted customers later on. Same thing if you're after particular interests, it just increases the overall number of campaigns by a lot. I also split-test the the ad types too (CPC, CPM, PPE, CVRs etc.). Placements I leave for last, once you start optimising the promising campaigns, usually.
Testing the images before anything else isn't wrong too, I think.
The above is very time consuming and expensive but I found that it does pay off. This is quite in the lines of you having/promoting a solid product that you know is good/will sell and is probably not a very good method when it comes to promoting the typical affiliate offers (unless you know you have a really killer/exploding offer)