Download
on time on demand delivery n.
Skip this Video
Loading SlideShow in 5 Seconds..
On Time, On Demand Delivery PowerPoint Presentation
Download Presentation
On Time, On Demand Delivery

On Time, On Demand Delivery

205 Vues Download Presentation
Télécharger la présentation

On Time, On Demand Delivery

- - - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript

  1. On Time, On Demand Delivery Chris Kantarjiev Time, Place & Space 20 September 2006

  2. At the tone, the time will be 1999 • 28.8kbps • 640x480 or 800x600? • What’s a cookie? • 500MHz? Wow!

  3. “Manifesto”

  4. Dec 1996: ISR Oct 1997: Oasis Virtual Supercenter Feb 1998: VP Eng hired May 1998: ChrisK hired Dec 1998: Christmas at the DC May 1999: First delivery Jun 1999: Public launch Nov 1999: IPO Feb 2000: 3000 orders/day Mar 2000: 2.0 cloned to ATL Jun 2000: Buy HomeGrocer Aug 2000: 2.0 cloned to CHI Jul 2001: Close the doors

  5. Region overview • 60 mile diameter • DC is 330,000 sq ft, costs $30MM • 50K SKUs • 4.5 miles of conveyors • Frozen ice cream, fresh turkeys • 14 delivery zones, 4+ vans/zone • Target 8000 orders/day, $100/order

  6. Lessons, part 1 Customers don't know where they live Maps don't know where they live, either Couriers are not drivers and don't know area Letting the customer dictate your schedule is a bad idea Knowing what to measure is hard

  7. The big picture • Front Office • Webstore • Customer Relationship Management • Transportation Subsystem • Order Management System • Order Fulfillment System

  8. The Promise • Must consider five things: • Inventory • Van driver time slot • Van capacity • Truck capacity • DC capacity

  9. Delivery flow • Define zones • Geocoding • Subzone assignment • Schedule delivery • Order confirmation • Cutoff • Fill to order • Pick • Stack • Truck load • Download routes • Van load • Delivery (unpack, returns, receipt) • Special situations

  10. Scheduling is much harder than routing • FTO was very slow, couldn't be started early • truck routes were 'fixed' but changed every week • personnel scheduling was done by hand • van resources were fairly fixed 7 days out • interesting anomalies from van and truck size • never got our heads around driver breaks

  11. Yield management • Make the map smarter • Special events • Construction • Realistic speeds • Variable pricing • Short window but not wide choice • Remove FTO

  12. What went wrong? • too much technology love • needed 5/hour couldn't do 4/hour • too customer centric • didn't charge enough for delivery • opened with a too-large service areas • DC too rube goldberg... chill chain/FTO 3250/day max, based on 8000 seayco server that handled tote diverters would crash • we started with 4000 orders/day, but we poisoned the well. • shifted from carriage trade to basic supplies, but never had same day delivery many SKUs does not match many stores • hard to get customers to plan ahead • hard to shift demand in time

  13. Lessons, part 2 • Don’t design for perfection • More vans not necessarily better • Open slowly, can’t go back • Think big in the right places • Optimize the entire chain, if you can • Bigger windows • Too much technology • Density matters!

  14. Make the map smarter Most traffic data companies focus on outliers (accidents, etc.) 80% of observations within dark blue band

  15. Traffic data sources • Government sensors • Cell phones • Tracked vehicles

  16. Traffic data providers