How to Read Algorithmic Attribution Numbers

[JayGr]

I'm a little confused how to read the breakdown of my algorithmic data and couldn't find any online explanations.

In this case, during this time period, it would say that Influencer should be credited with $200K more revenue than last touch gives credit.

However, when I breakdown the channel by more specific marketing tags to understand why, I get tags which don't even apply to the specified channel (which makes sense why they all have $0s in last touch).

But then why is "google" which is typically only used by Paid Search showing up as a breakdown on Influencer? How should I read this?

So did Influencer add $129K in value to search during this period? Or did search add $129K in value to Influencer? Did the combination of the two generate $129K in additional value?

Or is there some other way to understand what I'm getting here?

[Josh__Stephens]

I'm not sure what exactly the utm_source dimension is.  But you should use marketing channel detail to break out marketing channel.  You should see values that make sense.

Other dimensions, such as tracking code/campaign have differing scopes.  For example, tracking code/campaign only track channels with CID and will persist until expiration or replaced by another CID.  Marketing Channel also encompasses organic channels and others without CID. 

[JayGr]

UTM source is a Google Analytics marketing tag value which we have on our URLs and have included within Adobe as an eVar. I used that as an example because it's impossible for a marketing tag to show up in a visit for which that marketing tag couldn't have existed per our Processing Rules. 

So more simply, how is a breakdown of a channel showing values which couldn't have existed in that channel's visit? But since it is, what is the attribution model communicating is happening? And what is the value of what is being communicated for business decisions?

[Josh__Stephens]

Hi @JayGr - in your last touch column, I see there aren't any values which is what you expect.  When using algorithmic attribution, you're using multi-touch attribution.  If you use channel detail as the breakout dimension, each channel will be broken out by the appropriate details.

[JayGr]

Thanks for that suggestion, but channel detail gives me the same result as UTM Source. 

Which still leaves me at the original question....

Now all the growth comes from other channel's (and visit's) tags, so perhaps I am getting concentric circle and when these two channels appear in a journey, the impact is much more positive?

But when you flip it, all the biggest losers are tags which would align from that visit, so maybe then when a journey only contains that channel it's not as good?

Either way, I'm still guessing if this is the right way to do it. And if it is, I still don't know if I should do more of the channel or more of the channel detail.