Using Google Trends statistics as input data

[victoriatascau]

Are Google Trends statistics a potential substitute to Google Query Volume data? If yes, I would consider the resulted column in the dataset a confounding variable and place it in control variables, since it is not intervenable. Is that correct?

[dattatreyam23]

Hi @victoriatascau,

Thank you for contacting us!

Yes, your reasoning seems to be on the right track. Here is a brief confirmation of how your logic aligns with the official Google Meridian documentation:

• Google Trends as a GQV Substitute: Google Trends is a valid substitute for Google Query Volume (GQV). GQV data is already delivered as an indexed value rather than raw query counts, and scaling a control variable has no influence on model fit. Because Google Trends provides a similar indexed scale (0–100) representing search interest, it does fulfill the criteria. (See here)

• Classification as a Control Variable: Query volume represents inherent consumer demand, making it a classic confounder for paid search. Because you cannot directly intervene or control organic user search behavior, it is strictly a non-intervenable control variable, not a treatment variable. Controlling for it prevents the model from overestimating the causal effect of your paid search ads.

However, we would like to point out a few things:

• If your model separates brand and generic paid search channels, be sure to pull separate Google Trends data for brand vs. generic terms and input them as distinct control variables.

• Scale your Trends data by geo-population using the control_population_scaling_id argument.

Hope this helps. Feel free to reach out to us with any further queries regarding the same.

Thank you,

Google Meridian Support Team