20-Minute Dispatch Window

Measuring real dispatch departure

Target Last Mile needed a clearer read on when drivers actually left the sortation center after loading. The existing pickup timestamp was self-reported, and the old geofence was too broad to isolate real departure from time spent inside the wider sortation-center area.

My role was to validate whether geo signals could support the 20-minute dispatch window launch and make the load-time metric more trustworthy.

What improved

I compared manual pickup reports against geo-based exits from tighter polygons around the sortation-center garage or building. I also filtered sessions with bad location coverage, long ping blackouts, and anomalous garage pings before using the signal.

• Replaced broad geofence timing with tighter sortation-center polygons, cutting inflated geofence time out of the load-time metric.
• Filtered unreliable geo sessions with bad location coverage or anomalous pings before they reached the launch readout.
• Validated the launch rule on the cleaned signal, confirming geo-based departures lined up with the dispatch window.

Why it mattered

The launch needed a metric people could trust. This work showed where manual timestamps were early, where geo data was reliable, and where it needed to be filtered out. That gave the team a cleaner way to validate the 20-minute dispatch window across Target Last Mile.