How to read this
Strong means the methodology produces reliable, actionable output at your spend and industry combination. Viable means it works but with meaningful caveats — check the detail panel. Marginal means the signal-to-noise ratio is poor enough that you risk making worse decisions than if you'd done nothing. Not viable means a structural constraint (regulatory, data quality, or spend volume) rules it out.
The spend threshold for MMM is the hardest constraint: below ~$5M you don't have enough week-to-week variation across channels to separate effects from noise. Industry matters because it determines signal quality (do you have clean conversion data?), purchase cycle length (does any attribution even make logical sense?), and cookie dependency (how much of your current measurement breaks without third-party identifiers?).
The recommended method is the single best fit — not the only option. Most teams above $20M run two methods and use one to calibrate the other. Geo experiment results fed as priors into an MMM is the industry standard for large advertisers.
See the full MMM tool comparison →
Robyn, Meridian, Recast, Analytic Partners — open source vs managed, and which fits your team's data science capacity.