Paid media
Cannabis Geofencing: Fence Sizes, Dwell Time And What Actually Converts
Short answer
Cannabis geofencing works by drawing a polygon around a physical location, collecting the mobile devices that dwell inside it long enough to be real visitors, and then serving ads to those devices and their households for weeks afterward — with results measured as in-store visit lift against a holdout group.
Polygons beat radius targeting
A radius around a competitor's address usually includes a road, a parking structure and half a shopping center. A hand-drawn polygon that follows the building and its lot collects a far higher proportion of actual shoppers.
For a standalone store, fence the building plus its parking. For a store inside a strip mall, fence the storefront footprint only and accept lower volume in exchange for higher intent.
Dwell time is your quality dial
A device passing through in ninety seconds is traffic. A device present for five to ten minutes is probably a customer. Raising the dwell threshold shrinks the audience and raises its value; lowering it does the opposite.
Event fences work differently — dwell is naturally long, so the risk is collecting staff, vendors and neighbors rather than attendees. Tighten the fence to the venue footprint and the window to event hours.
Lookback and household extension
Historical audiences let you target devices seen in a location in the past, often up to around six months depending on the data source. That means a festival that already happened is still an addressable audience.
Household extension matches an observed device to its household and makes the other devices reachable. It expands reach meaningfully without widening the geography, which matters in states where broad targeting is restricted.
Measure visits, not clicks
Click-through rate on programmatic display is close to meaningless in this category. Hold back a portion of the audience, compare store visitation between exposed and withheld groups, and report cost per incremental visit.
If a vendor cannot describe their holdout methodology, treat their reported results as marketing rather than measurement.