Recommended Minimum Trait/Segment Size for Model?

[robert_yocum]

Is there a recommended minimum audience size to use for creating Models?  We were trying to see if there is documentation on this, but couldn't find it.

We want to document how high a success metric needs to be before the Model feature creates a useful Look a Like model?  For instance, can we safely state that if you don't have at least 200 devices in your trait or segment feeding the model, you should wait the build the model until there is at least that level of success?

Thanks in advance.

[shweta_singh · Accepted solution]

Hi robert_yocum​,

There is not any recommended minimum audience size for creating Models but make sure that there's enough overlap between the baseline trait/segment and the data source used for modelling. We can't model anything if none of the profiles from the baseline are found in the input data source.

We would always recommend looking at data sets with higher overlap with ESRI.

Thanks,

Shweta

[robert_yocum]

Noted. Is there a minimum target overlap we should use as a baseline?

Thank You,

Robert Yocum | Marketing Technology

Esri | 380 New York St | Redlands, CA 92373 | USA

T 909 793 2853 x2942 | M 909 633 3747 | ryocum@esri.com | esri.com

THE SCIENCE OF WHERE ™

[shweta_singh]

Hi robert_yocum,

Again there is no number but yes overlap should be in thousands.

Regards,

Shweta