POs considered harmful for dependent demands

[Bob Haugen <linkage@interaccess.com>]

Maybe everybody already knows this, so this is the short version.
I have included this point in a different message, but wanted to
make sure it was as clear as I could make it.

I still see documents going to this list that seem to assume 
that purchase orders are the way all B2B ecommerce is done.

PO's are not a good mechanism for dependent demands,
and if they are set in stone in ebXML in such a way that
it is difficult to do business without using them, it will
need to be redone for Internet-mediated commerce.

Dependent demands are demands that are dependent on some
other demand, usually called the independent demand.
This concept comes from MRP, the predecessor (and still
included in) ERP software.

Dependent demands include the components of manufactured
products, retail replenishments, shipping for almost any 
purchased item, etc.

Purchase orders are a carryover from paper systems.
They are usually composed of a collection of line items,
often aggregating quantities over time periods.  They
have no knowledge of how the purchases items 
will be used, nor what processes and components
are required to fulfill the order.

Dependent demands, by contrast, are totally dependent
on whatever independent demand stimulated them in the
first place.  

All dependent demands should be linked to their
relative independent demand so if there are changes
anywhere in the network of activities, they can be
rippled out to the affected relatives.

For example, if a customer order for a finished good 
changes in quantity or timing or is cancelled - the
dependent demands should be changed correspondingly.

The PO is too heavy a mechanism for managing
dependent demands - something more like an
electronic Kanban or manufacturing schedule
or point-of-sale event notification would be 
better.

The same goes for invoices, which are unnecessary
for dependent demands.

Comments? Violent disagreement or agreement?

-Bob Haugen