Spends as priors

[christosvis]

In Lightweight MMM we could use scaled media costs as priors with the scaler, which can be a good starting point if you don't have other info about the priors

costs = cost_scaler.fit_transform(unscaled_costs)
mmm.fit(media=media_data, extra_features=extra_features, media_prior=costs, target=target, number_warmup=1000, number_samples=1000, number_chains=2)

Is there a way to do this in Meridian?

[cpulavarthi]

Hello @christosvis,

Thank you for contacting us!

Meridian currently doesn’t support setting scaled media costs as priors similar to how it is done in Lightweight MMM. Meridian allows you to incorporate existing knowledge about your media performance through ROI priors. This allows you to leverage prior ROI information for each channel from incrementality experiments, industry benchmarks, or other domain knowledge.

If you have no knowledge or data to guide the prior selection, we recommend that you use default priors for the parameters while defining a Meridian model. You may find the details of default prior distributions in Meridian linked here.

You can also set custom priors in Meridian based on your past experiment data in Meridian. A detailed guide to the same is available in our documentation on the Setting Custom ROI priors page.

Feel free to reach out for any further queries regarding Meridian.

Thank you

Google Meridian Support Team