Stating
The Problem
Electronic
procurement is one of the major business-to-business growth markets on the Internet.
As opposed to previous generations of procurement, which have of course been
"electronic" for many years, E-procurement refers specifically to procurement
that is based, at least in part, on the use of internet or intranet technology.
The final direction will eventually involve all parties - users, approvers,
suppliers - working from standard browsers, but not all vendors have moved away
from products that require specialized desktop software.
The
whys of electronic procurement are obvious. Organizations look to E-procurement
to reduce administrative costs and improve turnaround, and to help them exercise
control over inventory and spending. E-procurement systems also provide new
sources of supply and can lead to lower prices for the good being procured.
It is no surprise that, according to a survey conducted by Purchasing
magazine, 90% of purchasing professionals expect to be buying online by the
year 2002. The market for E-procurement tools is estimated to grow from under
$200 million in 1998 to $8.5 billion by 2003.
But
the wheres of electronic procurement may not be quite so obvious, because
E-procurement needs to be thought about in a way that is different from most
other applications. For most applications, while there may be inputs coming
from outside the company or data sent to the outside, the core of the application
- the way we think of where it is -- lies well within the enterprise. With E-procurement
the focus is on the transaction, which lies between two different organizations.
If
E-procurement were simply about two different players - a buyer and a seller
- it would not be quite so interesting, or so difficult. But each player is
really the center of a star: the buyer deals with many sellers, and the seller
deals with many buyers; see Figure 1. So the question of where E-procurement
happens is neither trivial nor obvious.
Different
Solutions
An E-procurement system can be located at the seller, at the buyer, or in between.
Figure 2 shows the first case. In this sell-side model the buyer is like
a shopper on Main Street, who visits the various vendors of interest to look
at the offerings and prices. The seller mounts software that enables each buyer
to browse and purchase. This is, clearly, a sell-side model.
The
obverse case is, obviously, the one shown in Figure 3. Here the sellers have
to bring their wares to the buyers, like Travelling Salesmen of the early 20th
century. This is a buy-side model.
In
the final case, in Figure 4, third parties set up department stores to connect
buyers and sellers. This is called the Marketplace model; it has also
been called the portal model - or, when it serves a vertical market,
the vortal model.
Trade-offs
Each model has advantages and disadvantages for the buyer (and, of course, for
the seller). The important questions revolve around who builds and maintains
the catalogs that show what products are available, how well the procurement
functions integrate with corporate procedures and back-end systems, and how
convenient the result is for the buyer's employees.
In
the Main Street model each vendor puts up its own catalog, possibly on an extranet.
A company in partnership with such a vendor can have its employees navigate
to the vendor site with their desktop browsers and make purchases. No catalog
maintenance needs to be performed by the purchasing company, which makes adoption
of such a system cheap and easy. However, integration with back-end financial
systems will not be easy. There are standards being developed, mostly XML-based
next-generations of EDI, which in theory will make it easy for any of the documents
resulting from a purchase to be read by legacy or ERP systems. The success of
that approach remains to be seen as the various proposed standards compete for
acceptance. Some of the major vendors such as Ariba and Commerce One have their
own standards, as does Microsoft. The Main Street model also suffers from the
problems of comparison-shopping and control. In order to comparison shop a purchasing
agent must navigate through a number of vendor websites, each with its own interface,
packaging and shipping arrangements and discounts. Controlling maverick purchases
is nearly impossible. The Main Street approach will work best for smaller purchasing
organizations, and for cases where the vendor product is fairly specialized.
One example of this model is a new initiative by Eastman Chemical company. Eastman
is providing low cost computers from Dell and internet access via UUNET to allow
its customers to make purchases at the Eastman site.
The
other end of the scale is the Travelling Salesman model. Here the vendors come
to the purchaser by providing data from which the purchasing company builds
and maintains its own catalog. There are many advantages for the company doing
the purchasing. Since it controls the catalog it can exercise tight control
over which products get into the catalog, and even give different employees
different product selections and quantity limits. In this way purchasing can
be brought to the individual desktop, and employees have a single interface
to deal with. On the other hand, building and updating catalogs can be a major
editorial effort. One vendor in this market, PurchasePro, installs its software
at buyer sites. Using the PurchasePro software the client can shop at both vendor
sites and at marketplaces. One of PurchasePro's customers is the hotel operating
chain Meristar.
In
the middle are the Marketplaces, third party sites that bring buyers and sellers
together into their own market space. Marketplace sites do provide the unity
of a single catalog, but are less integrated with backend systems and the buying
company's own policies. Naturally, many vendors in this space are putting significant
effort into making their solutions look as much as possible like buy-side solutions.
There
are actually two variants of Marketplace. One is the vortal, the vertical marketplace.
One example is ChemConnect, which brings buyers and sellers in the chemicals
industry together. One estimate projects that 15 percent of this $1.6 trillion
industry will be Internet mediated within the next three to five years.
The
other variant is a marketplace that covers a wide variety of product types.
The real driver here is MRO - Maintenance, Repair and Operating supplies - procurement,
which can be as much as 35% of a business' total expenditures. Additionally,
MRO buying is a frequent cause of maverick buying - buying off of the preferred
vendor list. There are estimates that up to 40 percent of MRO buying is maverick
buying, and could be costing a typical billion-dollar company as much as $10
million per year in lost vendor discounts. There are a number of players in
this space. The leaders are Ariba and Commerce One, both of whom are actively
soliciting vendors and selling their solutions to buying organizations. Others
include Concur Technologies, Intelsys, and many smaller players. There is also
competition in this space from large ERP vendors like SAP and Baan, Oracle,
IBM, HP, and others.
One
way to implement a Marketplace model is to be completely Internet based, particularly
on the buyer side. This would mean that no software resided on the buyer's own
computers and that all operations were done by means of browsers. A second approach
would put specialized software on the desktops of the purchasing company's employees,
but house network of vendors centrally on the Internet.
Figure
5 is a suggestive comparison of the three approaches. It shows graphically how
the different models compare in convenience for some important criteria. But,
as always, the devil is in the details, which change daily in this product space.
For example, integration into back-end systems is theoretically expected to
be harder with the Marketplace model than for the Main Street because the Main
Street solution is installed and customized on the buyer's site. But there is
no reason that a particular vendor couldn't make a sufficient investment so
that their Marketplace model would integrate easily with your back-end systems.
Indeed some such ability is promised by major Marketplace vendors.
Conclusion
E-procurement is the most critical short-term application area for organizations
now that Y2K efforts are winding down because the potential savings are so tremendous.
Using E-procurement for ordering MRO goods can lead to savings from five to
twenty percent. The same study reported that the cost of processing orders can
drop by 70%.
We
believe that due to the cost of catalog maintenance pure buy-side solutions,
where the buying organization builds its own catalogs, will be a rarity in the
short-term, accounting for no more than 15% of the total purchasing market,
and only a few percent of MRO purchasing. it won't make sense for most companies.
However, once there is more standardization on the form of XML used for vendor
catalogs, this could resurface as a viable option, perhaps in three to five
years. We believe that most buying organizations will gravitate toward a Marketplace
model, at least in the sense of using centralized networks of sellers with centralized
catalogs. The vendors of these products are busy building up both their vendor
networks and their technology to make this an attractive option. We believe
that more than half of users will do their purchasing through browsers, but
that specialized purchasing software installations will sit on at least 30 percent
of desktops. . Vendors themselves are keeping both options open (See TEC NEws
Analysis article: "Ariba
Reaches Out to the Little Guy" September 28th, 1999). Potential users
should do careful analysis of features and, especially, the degree of integration
they need with their back end systems.