When you collect as much data as Amazon, you can start attributing incremental worker time spent picking/carrying down to the SKU level. This means that cost of

When you collect as much data as Amazon, you can start attributing incremental worker time spent picking/carrying down to the SKU level. This means that cost of worker time is less for things like books (a picker can carry many and move quickly) while a 60" TV might require two pickers moving slowly. Better attribution of cost means more margin to play with and cost savings that can be pushed onto customers in a fairer sense - especially if your pricing is automated. Interestingly Kiva may then selectively reduce the costs associated with bulky goods as they're probably better at moving big stuff and travel faster/shorter.