Weighted Multi-Touch Attribution?

[Anonymous]

Hello,

Our SVP is curious to know if Marketo has any plans to provide a way for us to create a way to incorporate weighted multi-touch attribution in reporting. Is anything like this in the works?

I'm curious, would anybody else find this helpful?

Thanks!

Louie Frank​

[Grégoire_Miche2]

Hi Sarah,

I am not sure I would know how to set the weights and on what scientific grounds I could justify them... I would be interested to read some methods / ideas on how to do this.

I am afraid that, setting the weights would in fact drive to create precisely wrong models, that are worse than being approximately true

-Greg

[Anonymous]

This week I attended a tech talk where one company spoke about using BrightFunnel to weight their acquisition program 40%, conversion program 40% and the rest of the programs that lead interacted with had 20% split between all of them (so 4 middle of funnel programs is 5% each and 20 programs would be 1% each).

Based on the presentation, it seemed like it not only helped them to give credit for middle of the funnel programs that help drive leads to bottom of the funnel, it helped them determine what programs to run based off of where their leads generally were. A huge influx of new leads one quarter might mean more spending on TOF programs but in a quarter or two, spending might need to be more MOF which previously would not have received spending because there was no insight into which programs were actually working or pushing leads along.

Something like that would be really powerful in Marketo.

[Grégoire_Miche2]

Hi Christina,

I agree these approaches are key, and the case you are making on how it can be used to balance the marketing plan between TOFU and MOFU programs is excellent.

My point was that one does not necessarily need to change the weights to do it. Marketo's program analyzer already provides a great insight on multi-touch efficiency. In Marketo, the user just does not have the possibility to change the respective program weights. (AKAIK, they are all equal).

The split you are describing (40%-40%20%) seems to me completely arbitrary, as arbitrary as the "all programs are equals" that Marketo provides. I would really love to understand the underlying theory and how one would know how to set the weights.

-Greg

[Anonymous]

Hi Gregoire,

The 40-40-20 model is a standard called "Position Based" or "U shaped attribution" (scroll halfway down this page​ for info on it). Although you don't have to use those exact numbers, they are the default set by Google Analytics. The idea is that the first and last touch be weighted more, but they don't have to necessarily be 40%, most articles recommend that you adjust the numbers over time based on what makes sense for your company.

I think it would depend quite a bit on the conversion time of the industry that you're in. If you have a very long cycle to conversion you likely have more MOF programs and splitting 20% might not be enough, so lowering first and last touch to 20% and sharing 60% across your middle programs would still keep their individual percentages below 20% (so that first and last touch are still weighted more). However, if you're in a much faster qualification cycle, 40-40-20 might make sense, especially if they might only hit 3-5 programs overall.

[Grégoire_Miche2]

Hi Christina,

Thx for the links! Very useful.

-Greg