Why are Free channels not available at the "Support" and "Campaign" level within MTA reports?
What is Multi-Touch Attribution (MTA) Analysis and How Does it Work?
Multi-touch attribution (MTA) analysis examines the order and sequence of interactions a user has with different touchpoints before converting.
This method assigns a relative weight to each interaction based on predefined rules, thus making it possible to determine the actual contributions to conversions.
Why are there limitations in comparing different levels or dimensions like leverage, medium, or campaign in MTA analysis?
MTA algorithms treat each touchpoint, like a campaign, independently, assigning credit based on its specific performance.
When we analyze data at a particular level, such as the campaign level, the algorithms treat each unique variable (in this case, each campaign) independently.
This means that if we compare campaigns to each other, the algorithms evaluate them individually and give them credit based on their respective performance.
Why are Free channels not comparable at the "Support" and "Campaign" level?
Because these channels do not benefit from “Support” and “Campaign” dimensions.
Let's imagine we need to compare a general dimension such as a marketing channel (e.g. SEO) with hundreds of unique paid campaigns, each with its own attributes.
If we were to apply a generic campaign name to the entire SEO free channel and compare it to these different campaigns, the results would be misleading and inaccurate.
Algorithms would not be able to nuance the sub-dimensions of the SEO channel, which would lead to biased and unreliable conclusions.
How can I adapt my analysis?
To address this issue, it is essential to analyze the data in the context of the chosen dimension.
This means that you should focus your analysis on variables that contain the same level or dimension of interest, whether it is a lever, a medium or a campaign.
To limit the perception of discrepancies between your MTA analyses, we recommend focusing on the percentages of the total attributed to each variable rather than the raw values.
The percentages still won't add up, but it can help you step back and see the results from a broader perspective.
Putting into practice
In the example opposite, the SEA, when analyzed at the Channel level, is attributed 300 conversions out of a total of 1000 (30% of the total).
However, in the Support level analysis, Google Ads alone is attributed 360 conversions, even though the total is lower (600 conversions).
The missing conversions are those that only contained free courses.
This example illustrates the impossibility of comparing levels with each other.
Each analysis is only relevant within its own dimension.
These two analyses based on the same initial data set are correct within their respective scope.
