Say you decide to target younger females with a specific angle.  You might want to target females age 18-29 but split it up into 3 age groups: 18-21, 22-25, and 26-29.  Further, you might also decide to split test Caucasian and everyone else.  With 3 age groups and 2 ethnicity groupings, you would be required to create 6 campaigns total to cover all the permutations of your split tests.

Can I sum up ad data across campaigns?

POF data analysis

Since the age range of your target audience is somewhat similar, you might decide that you can get away with testing the same creatives in all 6 campaigns.  Can you then aggregate the data from all 6 campaigns to gauge the performance of the same creative that is in these campaigns so you can arrive at a decision on those creatives faster?  The short answer is yes.

To be clear, in order to properly split test your age groups and any other targeting via multiple campaigns, you must set the same CPM bid.  In theory, there is absolutely no difference between running our 6 campaigns at the same bid and running 1 campaign targeting all females 18-29 in terms of ad serving.

Obviously, the advantage that creating 6 separate campaigns does give us is the ability to see the performance per segmentation.  This splitting of the campaign into multiple parts, however, reduces the volume received per sub-campaign.  That is, the traffic is split into 6 different segments.  Therefore, you may want to add up the impressions, clicks, cost, revenue, etc., across the campaigns for the same creative if you feel that you’re gathering your testing data too slowly.

When doing so, however, you should check for and keep the ads that are actually doing well for particular sub-campaigns.  For example, just because an ad is doing poorly across 6 campaigns, it doesn’t mean it isn’t killing it in 1 of them.  That’s the whole benefit of splitting into sub-campaigns in the first place.  Another benefit, of course, is the ability to see if an entire sub-campaign is unprofitable.

How do we track the data?

To track your data for a specific creative across multiple campaigns, you just need a code to identify the creative.  As an example, a code I might come up with looks something like this:

CampaignName_0000001[002]-123

In the above example, “0000001” is an image code that I might use to keep track of the images I’m using, while “002” is a copy code that I might use to keep track of the copies I’m testing, and finally, the “-123” are merely random numbers appended to ensure that every keyword I pass to Prosper202 is unique.

In our 6 campaign example, then, you would have a “0000001[002” that identifies the exact creative combo (image and copy) in each of the 6 campaigns, but the appended “-123” would be different to keep those 6 ads identifiable through unique keywords.

If you’ve gone through the basic Excel tutorials, then you would have learned how to isolate the “0000001[002” in the POF Daily Creative Delivery report into its own column.  You can then sum up the performance data via that column (hint: use the SUMIFS function).

If you haven’t done so, please run through the  Excel tutorials, as well as check out the POF 7-Day Mastery Guide, where I go through my overall campaign tracking methodology.  You might also want to check out the Basic Campaign Viewer and use it as a base to build additional functionality such as creative tracking as described here.  The Campaign Viewer is for subscribers only, so if you haven’t, simply enter your email on the right sidebar of the blog and get your copy!

If you are confused about anything in the article, please post in the comments!

  • Mike

     Hi Tom

    Is it possible to have the Mr G’s uploader append the image name to the tracking url so I can easily search thru my images and find / colate the converting images?

    Thanks
    Mike

    • No I don’t think it’s possible with that uploader.

      All this stuff will be solved with our upcoming POF platform, but in the meantime, you’re going to have to figure it out with a combination of Flash Renamer and Excel.