ROI Prior

[shubhambooking2]

In analyzing the return on investment (ROI) for media channels, I find that ROI estimates are heavily influenced by specified priors. Despite expecting the posterior distribution to be primarily influenced by the likelihood of data given a specific media channel, it appears that the priors still significantly impact the results. Even attempts to introduce non-informative priors, such as by increasing variance, result in some ROI estimates being unreasonably high (e.g., 50). How can I establish ROI priors in a manner that ensures the interpretation of signal from data rather than being disproportionately influenced by the priors?

[jyklly]

It seems like there may not be enough information in the data. Please see When the posterior is the same as the prior in the Additional Considerations section of the documentation as it addresses this issue.

Please also see Channels with low spend, When ROI results are widely different depending on the prior used
and Amount of data needed for additional insights.