In the British and American markets, it is clear to us that our supply chain is not nearly as effective as we would like. Returns are high, possibly as high as fifty percent of the copies of new titles shipped for many consumer segments in the US. Returns are not as dramatic, but they are climbing, in Britain.

So what we are doing isn't working.

Vendor-Managed Inventory, by which the responsibility for choosing the content and timing of shipments to each location moves from the buyer to the seller, is the best hope to improve that performance to the benefit of the entire supply chain. VMI would be helpful whether it were offered by a wholesaler across the range of possible titles or by individual publishers for their titles alone. VMI would improve a store's performance, and sharply reduce their operating costs, whether some or all of their books were supplied this way.

Here's how VMI works. The supplier -- the publisher or wholesaler -- agrees on inventory limits with the store. Those limits can be across-the-board, or can be assigned by subject section. Of course, the limits would change with the calendar, just as a store's sales do, and just as a store tries to make its inventory do to be in step with its sales. Granularity to the level of store sections is optional, but it is a requirement that inventory limits be set for each individual retail location in any circumstance where an agreement is being made centrally for a group of stores.

Then, somehow, the publisher or wholesaler must know what actual sales are in each of the locations being controlled. In the past, like when my father first installed VMI at Doubleday in 1957 or even when we did it at Two Continents in the 1970s, the only way to do that was to send in a sales rep to take a physical count of books on hand. Today, that data is retrieved effortlessly by EPOS systems, making it efficient for an offsite supplier to get up-to-date sales information as often as desirable, even daily.

Based on the sales and inventory reporting, the supplier then calculates the next shipment and makes it. On a periodic basis, a scorecard of the actual results will demonstrate whether the supplier's control is sufficiently profitable. A store using this system with many publishers would, of course, regularly adjust the inventory limits to give more profitable publishers a larger inventory allowance, taking shelf space away from less profitable ones.

The suppliers would achieve better results to the extent that they used sound statistical prediction methods to calculate future stock needs from sales data. You need a good buying system to maximize the opportunity. They would also achieve better results if they ship more frequently, which will permit them to address out-of-stock situations faster and to avoid shipping big quantities in anticipation of future sales. It is precisely those shipments which result in a disproportionate share of the books returned. Shipping more frequently is the most obvious leverage provided by VMI, which eliminates the need to put a selling bureaucracy in touch with a buying bureaucracy to trigger a purchase ordering mechanism. That clears the most significant barrier to frequent shipment.

Obviously, VMI cannot work without controls. If the store sets no inventory limits or there is not some other control like a consignment payment arrangement, there will be no discipline to force publishers or wholesalers to pick the right books, or to avoid "loading" the store. And if there is no benchmarking of what the performance was when the store did the buying, and no scorecard of the supplier's performance, there will be no reliable way to confirm that the situation has actually been improved.

Time does note permit elaboration today on the many ways VMI actually works to increase sales and reduce returns. I would refer those of you who want more detail on that point to some of my prior speeches, which are on both Vista's and Idea Logical's Web sites. But, frankly, any technique that increases the frequency of inventory replenishment will increase sales and reduce returns. VMI offers enormous efficiencies in addition. It cuts the need for selling staff and buying staff. It eliminates the need for the elaborate marketing preparations that are now required for publishers to sell the books into the stores before publication. It enables instant publishing and distribution in ways that cannot be conceived in today's marketplace.

Our company is now deeply involved in one VMI development and on the cusp of another. Our client, The Butterick Company, is a creator of patterns for sewing garments. They have almost universal distribution into the stores that sell fabric. Starting in early 1999, the newly formed Butterick Book Distribution Company will place the books of all publishers into that marketplace in the US. The initial customer, JoAnn's Fabrics of Hudson, Ohio, has 1200 stores for which Butterick will become the exclusive books supplier. All shipments, and there will be a shipment each week to each store, will be via VMI.

Butterick could not have made this deal with JoAnn's if they had not offered VMI. The ceding of control is not a barrier with the account; it is seen as a great advantage by the account.

We have also been asked for help by the Canadian arm of a global book publishing company that wants to begin offering VMI to its domestic bookstore network in 1999.

In addition, VMI is being tentatively introduced at the margins of the US book trade by publishers at the edges of true trade publishing, specifically publishers of calendars, computer books, and mass-market juveniles. And it is tending to be introduced in non-book outlets, which would otherwise require a big investment in the personnel and systems necessary to manage book inventories, and which show little inclination to accept such substantial costs. One wholesaler, Advanced Marketing of San Diego, which is a supplier primarily to mass marketers, has begun to stock books for their customers using VMI and is claiming great success.

But while take-up of the concept seems to be increasing, it is certainly not pronounced enough to be labelled a trend. Despite widespread recognition of the ills in the supply chain, there is real built-in resistance to the concept. Chain stores and multiples have invested heavily in buying staffs, buying systems, and an infrastructure, often including warehouse and shipping capabilities, to service the status quo. That represents a core constituency in their organizations to maintain it.

And the changes that would be wrought by VMI threaten many in the publishing companies too. The whole catalog-and-sales conference, seasonal selling cycle would be changed. So would the placement of large speculative quantities of books in advance of publication, preceding any true consumer reaction. The compensations for these changes: faster and more direct access to the consumer, the ability to keep books alive easily even at low sales levels if bookstore customers are indeed buying them, will take awhile to prove out and provide relief from the discomfort of losing established professional rhythms and expectations.

VMI is a sensible idea for a supply chain that is desperately looking for one. It gets easier every day as EPOS technology spreads to universality. VMI has become is a concept that no publisher concerned about the level of its sales, the level of its returns, or its costs of doing business, can afford to ignore.