Varying revenue per conversion with channel

[aadityachouhan-ah]

Hi all,
Is there a way we can set different revenue per KPI for different channels?
The objective is to utilise channel level LTV (from internal attribution) for analysis on top of conversion which is used as the KPI.

[cpulavarthi]

Hello @aadityachouhan-ah,

Thanks for reaching out!

In an MMM study, “conversions” will most likely be the single target KPI representing the total outcome of all marketing efforts, thus often not having channel-level granularity. The media exposure metrics of an MMM study would be metrics such as clicks, views, or impressions and they have the channel-level granularity. Also the media spend will have channel-level granularity.

Meridian is designed to work with a single, aggregated target KPI column that reflects the total KPI (in your case conversions) observed across all channels for each time period and geographical area. Because of this, we currently allow you to set the revenue_per_kpi multiplier for different time periods and geographies, as we understand these values can vary significantly based on when and where conversions occur. You may check our documentation on Input Data for more information regarding how Meridian accepts data, what should be the dimensions of each type of data input and what are the transformations performed on the input data while loading it into an InputData object in Meridian.

Since Meridian accepts the target KPI as a total value (not broken down by individual channels), setting a specific revenue per conversion per channel wouldn't align with how the data is structured for the target KPI. If you have a specific use case where setting revenue_per_kpi at a channel level would be beneficial for your analysis, please share more details so that we may explore it further.

Do get back to us if you have any further questions or suggestions regarding Meridian.

Google Meridian Support Team

[aadityachouhan-ah]

Thanks @cpulavarthi for your response.
I am using Meridian with conversions (user activations) as the KPI and not providing any cumulative revenue per conversion. So, I don't get the RoI analysis as the KPI is not revenue. From internal attributions (though not very accurate, as generally the case is), we have an idea how much value a user coming from a particular channel generates over a fixed period of time (the Life Time Value or LTV). What I wish to do is to scale the actual KPI which is conversions with LTV of the corresponding channel to convert my output to revenue for RoI analysis. Also, I am using national level data and the LTV is considered fixed for the considered time duration for a given channel.
Please do let me know if any further detail is required.

[cpulavarthi]

Hello @aadityachouhan-ah,

Thank you for the clarification!

If KPI itself can be segmented in terms of LTV you can fit separate models per segment; otherwise Meridian does not support breaking down revenue_per_kpi on a per-channel basis. However, you can either incorporate LTV directly into the model (say, by using LTV as a KPI), or conduct custom analysis by extracting incremental outcome and/or ROI in the Analyzer module.

Feel free to reach out if you have any further queries regarding this!

Google Meridian Support Team