There was some interesting discussions in Slack’s analytics community about the topic I recently wrote a post: Basics of pages, marketing channels and tracking codes in Adobe Analytics. Reporting window allocation gives credit to all traffic sources (within reporting date range), but what if you want to give credit to only channels or tracking codes before the converting visit?
I’ll take example directly from the discussion we were having in Slack. Visitor has 3 different visits and every visit is containing a new campaign code (a, b and c). If we looked event with “reporting window allocation” all of those three campaign codes would get value of “1”. Great, but sometimes it would be useful to understand difference between last touch vs. what were sources/campaings that helped to get the conversion. Could there be some campaign that rarely is the last touch before conversion, but without exactly that campaign we would anyway have much less conversions? This is possible to analyze with so called “event assists” calculations.
Previous post I had user with 3 different visits and want to test this “event assists” with my example. Here is the calculation to get the metric. Just change event to whatever you want to use. At first, you choose to count all conversions with event you like and then you just reduce events that happened with the last visit user had. You have to use those correct allocations it to work!
My visitor converted on Natural Search. Perfect, my new “assists metric” is working, because it does not give any credit to my last touch channel.
And looks good also for my tracking codes, because “facebook-campaign-123” was the last tracking code (even though, it was on the second visit and not in the third)
Of course, it’s little different situation when you have thousands of visits with different campaign codes, so you can’t know what were full paths for user journeys and which channels tend to work best together. Not a real attribution models, but again, you can get some good insights playing around with different allocations. Hmm, what else could I try? When you have lots of data then usually conversion metrics brings the best insights. Let’s try event assists as conversion % metric. I did this kind of calculation:
Could that work? Let’s check my data:
Seems to be working just fine. Email and social networks were the two channels that assists to make my conversion and because those have two other visits then conversion 50% is correct. (forget that referring domains data, might be some testing I did) This way you could get more insights which channels are very useful, even though not bringing conversions right away. NOTE: I INCLUDED ALL THE VISITS (ALSO LAST TOUCH VISITS) AND THAT’S WHY MY CALCULATED METRICS IS NOT PERFECT, BECAUSE IT COUNTS ALL VISITS AND THIS MAKES CONVERSION % LOOKS SMALLER THAN IT REALLY SHOULD BE (because I won’t count conversions in the last visit, but count visits anyway, makes sense?) I guess difficult if not possible to see last touch report with conversion metric excluding last touch visits!? 🙂
With tracking code you can use calculated metrics including settings as “exclude last touch”. I did little testing with these different metrics on another report suite where I had more data and let’s look one campaign code that has enough data for different calculations:
Visits, form submissions and conversion are basic stuff. These we usually investigate when we want to see which campaign codes are rockin and making conversions. Tracking codes has memory for one month (or whatever you have used in admin settings). However, if you have multiple campaigns rockin at the same time then there are lots of situations where new tracking codes overwrite the previous one and you might miss some good tracking codes that are very good to give the assistance for conversions.
Code (above) is also helping to generate 26 additional conversions, that’s pretty awesome. If I calculate conversion % based on all visits I get 1,33% conversion. However, I need to exclude also last touch visits (after that I get 369 visits) to get the real conversion % for these “assists conversions” and that is 7,05%. Wow, cool! So I definitely need to continue to use this campaign too because it helps so much other campaigns to generate conversions. Yes, there is some complexity and if you noticed that I made some kind of mistake then please share your thoughts, thanks!
Here are calculations for my “Visits – Excluding last touch visits” and “Form Submission OPV – Assists only (RWP %) (excluding last touch visits)”. Hah, that’s a short name for a metric.
Ps. Adobe also has small and simple example for “order assists” calculation:
Although, bit confusing to me, because in the screenshot you can see “participation” and that does not exists today. Maybe it a screenshot before release and it was updated to “reporting window participation”? It should be “reporting window participation” to work, and not just “visit participation”, right?