A customer order may be picked entirely by one worker; or by many workers but only one at a time; or by many at once. The appropriate strategy depends on many things,but one of the most important is how quickly must orders flow through the process.For example, if all the orders are known before beginning to pick, then we can plan efficient picking strategies in advance. If, on the other hand, orders arrive in real time and must be picked in time to meet shipping schedules then we have little or no time in which to seek efficiencies.

    A general decision to be made is whether a typical order should be picked in serial (by a single worker at a time) or in parallel (by multiple workers at a time). The general trade-off is that picking serially can take longer to complete an order but avoids the complications of co¨ordinating multiple pickers and consolidating their work.

    A key statistic is flow time: how much time elapses from the arrival of an order into our system until it is loaded onto a truck for shipping? In general, it is good to reduce flow time because that means that orders move quickly through our hands to the customer, which means increased service and responsiveness.

    A rough estimate of the total work in an order is the following. Most warehouses track picker productivity and so can report the average picks per person-hour. The inverse of this is the average person-hours per pick and the average work per order is then the average number of pick lines per order times the average person-hours per pick. A rough estimate of the total work to pick the skus for a truck is the sum of the work-contents of all the orders to go on the truck. This estimate now helps determine our design: How should this work be shared?

     If the total work to pick and load a truck is small enough, then one picker may be devoted to an entire truck. This would be a rather low level of activity for a commercial warehouse.
     If the total work to pick and load an order is small enough, then we might repeatedly assign the next available picker to the next waiting order.
     If the orders are large or span distant regions of the warehouse or must flow through the system very quickly we may have to share the work of each order with several, perhaps many, pickers. This ensures that each order is picked quickly; but there is a cost to this: Customers typically insist on shipment integrity, which means that they want everything they ordered in as few packages as possible, to reduce their shipping costs and the handling costs they incur on
receipt of the product. Consequently, we have to assemble the various pieces of the order that have been picked by different people in different areas of the warehouse;
and this additional process is labor-intensive and slow or else automated.
     For warehouses that move a lot of small product for each of many customers,
such as those supporting retail stores, order-picking may be organized as an
assembly-line: The warehouse is partitioned into zones corresponding to workstations,
pickers are assigned to zones, and workers progressively assemble each
order, passing it along from zone to zone.
    Advantages include that the orders emerge in the same sequence they were released,
which means you make truck-loading easier by releasing orders in reverse
order of delivery. Also, order-pickers tend to concentrate in one part of the
warehouse and so are able to take advantage of the learning curve.
    The problem with zone-picking is that it requires all the work of balancing an
assembly line: A work-content model and a partition of that work. Typically this
is done by an industrial engineer.
Warehouses tend to use combinations of several of these approaches.