WO2001071626A2 - Marketplaces involving business rules for partially automated generation of quotes - Google Patents

Description

MARKETPLACES INVOLVING BUSINESS RULES FOR PARTIALLY AUTOMATED GENERATION OF QUOTES

CROSS-REFERENCE TO RELATED APPLICATIONS This application is related to commonly-owned U.S. Pat. Appl. Serial No.

09/498,023, entitled "System For Multi-Dimensional E-Commerce Auctions", filed February 4, 2000 and bearing Attorney Docket No. M-8388 US.

BACKGROUND OF THE INVENTION Field of the Invention

This mvention relates to business methods and, more particularly, to business methods for processing requests for offers and generating bids in online marketplaces.

Description of Related Art If buyers were perfectly informed about all the products and services available in the marketplace and the prices at which each set of products and services could be purchased, there would be no need for marketing, advertising or sales departments. In reality, however, gathering information about the various products and services offered is costly and time consuming, especially when buyers and sellers care about factors, such as geography, product quality, delivery time, and cost.

Various mechanisms have been proposed and/or implemented for simplifying the purchasing process. Any process that tries to match buyers and sellers should try to meet at least three objectives: 1) consummating deals between the buyer(s) and seller(s) who have the most to offer one another; 2) minimizing the elapsed time from the initiation of a search for a trading partner and the conclusion of a trading agreement (or delivery of the product/service); and 3) minimizing the resources (such as the value of human time needed to locate and participate in auctions) required to solve the first two objectives. No method yet meets all three objectives. Current methods only partially meet at least one of the above objectives. Minimizing elapsed time (second objective) and resource costs (third objective) are separate issues because, e.g., even an instantaneous "sealed bid" auction can entail high resource costs to buyers because buyers must find the auction and prepare and submit intelligent bids. Sellers can also incur high resource costs because bidders may demand the hiring of a third party to ensure fairness and (perhaps) secrecy of bids. Both buyers and sellers incur substantial costs when bids are "multidimensional" (i.e., the bids involve more factors than price alone). For sellers, the problem of creating an attractive multidimensional bid is more complicated than just naming a low price. For buyers, bid evaluation is more complicated than just identifying the lowest price. The buyer must either evaluate multidimensional bids manually or predefine a formula for evaluating bids.

TRADITIONAL ENGLISH AUCTIONS

English auctions are well-known auctions in which a seller gathers together buyers and sells a single item (or multiple identical items) to the highest bidder(s) where bidders publicly announce their bids and an auctioneer terminates the auction when no bids have been placed for a certain period of time. Traditional English auctions fail to completely meet any of the above objectives. It may seem that English auctions meet the "best buyer-seller match" objective because the highest bidder wins. However, the highest bidder may not be the highest possible bidder since not all potential bidders are aware of the auction. English auctions also fail to eliminate the waiting time which elapses from beginning of search to time of purchase. Potential buyers must find the auction, evaluate the product, and submit a series of bids. If their top bid fails, the potential buyers must repeat their search. The time required to enter auctions, wait for an outcome, and (possibly) repeat the process creates delay and uncertainty which can be very costly to buyers.

Sellers confront similar problems. Sellers want to sell products quickly but achieve high prices. To ensure that an auction results in a sale above a certain minimum price, many sellers set "reserve prices". But many auctions fail to achieve a bid above the reserve price. This results in additional inventory holding costs and delay in receiving cash for a sale. Asset values are typically higher for "liquid" assets because liquid assets allow asset owners the ability to quickly sell the asset at a "fair" price. Traditional English auctions also require human decision-making. For example, humans typically research potential auctions, travel to auctions, determine how much to bid, and submit bids based on the progress of the auction. The aggregate cost of human intervention increases, as the number of required human interactions in an auction grows. Thus, there is a tradeoff between selling an auctioned product to the highest valuation potential bidder and the aggregate human cost of auction participation.

ONLINE AUCTIONS

With the increasing popularity of the Internet, computer-based marketplaces are becoming an attractive business option for companies and individuals to both purchase and offer products and services. One popular class of electronic marketplace is so- called "online auction sites", which conduct auctions on the World Wide Web. Perhaps the most well known online auctioneer is eBay (http://www.ebay.com). In an online auction, sellers use an Internet-based user interface to enter information about products or services they would like to auction off. Bidders enter bids for products they would like to purchase into a different, also Internet-based, user interface.

ONLINE ENGLISH AUCTIONS

Bidding in traditional English auctions is a time-consuming process which often requires the physical presence of the bidder. Conducting English auctions (and variants of English auctions) online (through, for example, eBay) reduces the cost of submitting offers to auctions. Because bidders need not be physically present and can bid at relatively low cost, participation is higher, leading to more efficient pricing and better matching of auctioned products to buyers. As the cost of participating in an auction drops, participation rises which, in turn, increases the probability that the highest valuation potential bidder wins.

However, eBay's method does little to address the resource cost of bidding (since buyers must still find the most appropriate auction and engage in time-consuming bidding), hi some respects, online auctions are even more time-consuming, hi contrast to traditional English auctions conducted at a single location, online English auctions typically stretch over long periods of time, such as several days. Even though placing a single online bid is fast and easy (typically requiring only a few mouse clicks), bidders must frequently check current bids and adjust their bids accordingly in order to compete with other bidders. Additionally, bidders in English auctions must wait to learn the price required to win the auction and acquire the item(s). For bidders who win the auctions, this delays their purchases. For bidders who lose the auctions, it delays the time before they can enter a new auction. And all bidders suffer because time delay creates uncertainty about when the buyer will eventually find the item(s) at an acceptable price. For certain bidders (especially businesses), delay and uncertainty are particularly costly. So, to the extent that online auctions succeed in attracting large numbers of actual or potential bidders, the aggregate resource cost of operating an auction market increases.

A method eBay has adopted to address the problem of the time bidders must put in is to close auctions at a fixed time, e.g., seven days (to the minute) after the auction begins. This means that bidders do not have to watch the auction continuously, thereby saving the bidders' time. Unfortunately, bidders have little incentive to bid early in such an auction. Rational bidders can wait until virtually the last minute before submitting a bid, hoping that rival bidders have no time to respond. (This practice is common enough to have acquired a name, "sniping".) This introduces some distortions. It means the auctions have some characteristics of one-shot sealed-bid auctions; many of the theoretical advantages of the English auction form are lost. Online English auctions, therefore, have some serious drawbacks.

One innovation for reducing the time demands of participating in online auctions is "proxy bidding". Current Internet auction sites (such as eBay) have a "proxy bidding" feature, which allows a user to enter the maximum price they are willing to bid. This spares users the necessity of placing a large number of bids by hand. Instead, the system can place bids for them.

ONE-SHOT AUCTIONS

Another approach to eliminating the need to check current bids and update bids is to use a one-shot auction, in which bidders specify bids and the best bid wins. There are several common kinds of one-shot auctions, differing according to how the price is determined and whether the buyer or seller initiates the bidding.

The most familiar kind of one-shot auction is the sealed-tender sale, in which the seller initiates the auction and each potential buyer specifies the price it is willing to pay. The highest bidder wins at a price equal to the winning bid. hi business-to-business and business-to-government procurement, the auction is more commonly initiated by a buyer, whose goals include obtaining a low price. A buyer initiates this "reverse auction" by describing a particular product or service it wishes to purchase or a set of needs it wishes to fulfill. The buyer then solicits bids from sellers. In the reverse sealed-tender auction, the lowest bidder wins at a price equal to its bid.

There are also sealed bid auctions in which the price is set differently. The most famous of these is the "Nickrey auction". In a Vickrey auction initiated by a seller, buyers submit sealed bids for an item and the winning bidder wins the item but pays the price offered by the second highest bidder. William Vickrey proved that each buyer' s "dominant" (i.e., optimal) bidding strategy in such an auction is to bid its maximum price. This is true because although a buyer's bid in this auction affects the conditions under which it wins the auction, the winner's bid does not affect the price it must pay, which is the second highest price submitted, not the lowest price. A "reverse Vickrey auction" is a Vickrey auction with the roles of buyer and seller reversed. The buyer then solicits bids from sellers. The lowest bidder wins the auction and the price is set equal to the second lowest bid.

A serious problem with reverse auctions is that sellers may offer somewhat different items in response to a buyer's solicitation. If so, price is not the only relevant factor in the buyer' s mind. The buyer must somehow determine which combination of price and product characteristics is most appealing. Evaluating offers requires time and effort. Also, because the buyer's preferences are not known perfectly, sellers are unsure whether a certain offer is better than another. Thus, in many important applications, optimal bidding strategies are very complicated and possibly unknowable, even when the prices are set by "Vickrey" type rules. A serious drawback of both Vickrey and reverse Vickrey auctions is that the bidding process is not interactive, so each bidder must privately tell the auctioneer the lowest bid they are willing to make. The bidder who makes the lowest bid wins the auction (assuming the auctioneer is honest). However, the winner is charged with only the second lowest price (for reverse Vickrey auctions) or the charged only the second highest price (for Vickrey auctions). In theory, this yields the same outcome as the English auction. While it has the advantage of requiring only one bid from each bidder, a reverse Vickrey auction requires each and every participating seller to reveal their final offer to the auctioneer. If multiple auctions are conducted over time, this creates potential risks because such information would prove valuable to buyers and other sellers were it to be revealed due to an unscrupulous auctioneer or security flaw.

MULTI-DΓMENSIONAL AUCTIONS

The most serious shortcoming of English auctions and sealed-bid auctions, including Vickrey auctions and reverse Vickrey auctions, for procurement applications is that they apply only when bids are one-dimensional (price being the usual dimension). This is acceptable when buyers compete to purchase a single product for sale, but single dimensionality rules out auctions in which buyers state their needs and preferences and sellers compete to satisfy those needs and preferences. In many business-to-business (B2B) and business-to-consumer (B2C) applications, the buyer does not simply want a particular, precisely specified product at the lowest possible price but wants to find a product or service to meet needs that could be satisfied in a variety of ways by any of several differentiated product(s). Competition among such suppliers is not based on price alone but also on product characteristics. Similarly, sellers often care about who buys the product or service (because this affects, among other things, delivery cost and the cost of servicing) and about hard-to-pre-specify details of the product or service they wish to sell. For example, the buyer's location affects things such as delivery costs, time-to-delivery, and the cost of providing on site service. Bidding becomes even more time-consuming and costly when bids are multi- dimensional. Also, because those costs are proportional to the number of participants, they are a significant barrier to the development of large auction markets. On the time side, if request for quotation (RFQ) evaluations are imperfect and the buyer must run multiple auctions to match product/service specifications to their needs, the time needed to complete the buying process is increased. Thus, it is desirable to have rapid and inexpensive processing of RFQs to help buyers focus in quickly on the set of most likely vendors.

Conventional methods, such as those discussed below, do not satisfactorily solve the multi-dimensional matching problem. Making bids for simple items may be cost- effective, but the auctions still can take days to complete, which inhibit their use in RFQ contexts in which the buyer must buy quickly and/or needs to revise its request several times. This is not an issue for eBay because the products it sells have fixed descriptions.

ONLINE RFQs

To allow explicit competition among differentiated products in an Internet marketplace, buyers must be able to enter RFQs into a suitable user interface that allows them to express what is being sought. The RFQ must then be circulated among potential sellers, each of whom can choose to provide a quote to the buyer. For example, Internet sites Respond.com and Imandi.com have recently created systems in which sellers respond to RFQs with email offers, which buyers can accept or reject. On Via is another example of an RFQ submission web site.

By using the Internet to circulate the RFQ and the quotes, significant cost reductions are achieved. Also, because of the instant delivery of email messages, communication takes place a lot faster than in traditional, paper-based RFQ-and-quote processes. Online RFQ systems allow buyers to specify needs with more latitude and allow sellers to be flexible about how the buyers' needs are met. For example, buyers can specify in an RFQ that they need a computer system with certain minimal performance standards without specifying the particular model and configuration desired. Sellers can then review the RFQs they receive, consider whether they are worth the time required to prepare, and if desired, transmit a quote to the buyer. Such online RFQ transmission systems can also aggregate buyers online, allowing sellers to submit RFQs and quickly receive responses from many potential buyers to purchase the product or service described in the seller's RFQ. The seller's RFQ could explain the product or service in great detail or could be vague in certain areas because the seller is open to alternatives and willing to accommodate buyers' particular needs. For example, a consulting firm (seller) might want to rent out a consulting team with no current projects. The seller's RFQ might indicate the domain expertise of each consultant but be flexible about location and industry of the engagement.

One problem with existing online RFQs is that human action is required to respond to an RFQ with an offer because RFQ responses are either not automated (in which case the seller must assign a person to respond) or are fully automated in a non- useful way (in which case the buyer receives information with insufficient, non-specific detail of little use to the buyer such as the phone number of a possible seller). So, buyers using existing online RFQ systems typically experience significant delays, ranging from hours to days, from the time of RFQ submission until receipt of detailed offers.

For the buyer, the cost of submitting an RFQ is very low. In fact, online RFQs may be too easy to submit. Many "buyers" submit RFQs merely to acquire information. For example, submitting RFQs is a valuable method for benchmarking. Companies competing in the product/service space may submit RFQs without revealing their true identities in order to collect data about their competitors and the way they generate quotes. Or, an engineer designing a system based on prefabricated components might use the results of an RFQ to estimate the cost of a particular component.

The most serious problem with online RFQs is that the total human cost of responding to RFQs increases multiplicatively with the number of RFQs submitted and the number of recipients per RFQ. With present technology (which requires a human to respond to sufficiently complex RFQs), each RFQ incurs a non-negligible cost to the sellers, since sellers must at least evaluate whether it is worth providing a quote (where a "quote" consists not only of a price but can also include a set of product features and attributes, such as delivery time). This gap between the low cost to buyers of submitting RFQs and the substantial collective cost to sellers of processing the RFQs results in frustration and inefficiency on the seller side. If there is no mechanism to ensure that RFQs are genuine, sellers start ignoring the RFQs, and the system breaks down. Even if RFQs are genuine, the human cost of seller participation may still discourage some bidders or lead the auctioneer to enforce limits on who is allowed to bid. Instances have occurred where sellers quit online RFQ systems because of the "firehose problem", i.e., sellers are deluged with RFQs, many of which are not closely matched to their products and services. Automated RFQ systems can quickly become "corporate spam", e.g., a source of widely distributed emails that fill up the electronic mailboxes of the recipients.

One way to ensure that RFQs are submitted only when there is a real need and an intention to buy is to charge a fee for placing an RFQ. However, this discourages buyers from placing RFQs and reduces efficiency and popularity of the RFQ site and its usefulness to sellers.

An additional problem with RFQ systems is that the buyer must evaluate each offer before deciding which one to accept. This problem is exacerbated by the number of potential offers a buyer may receive after submitting an RFQ. For example, after submitting an RFQ to hnandi.com, a large number of responses was received, but the offers were radically different from one another, not ranked or classified in any way, and often providing little information beyond a phone number. For complex products or services, the decision making process is very difficult and time consuming. One way of dealing with this is to define a scoring rule that scores each of the offers automatically. Scoring rules have traditionally been used only in high-value B2B or government-to-business contracting because they have been complex to specify and administer. Information contained in each quote has to be entered by hand into a decision making system, and the information in the quotes may be incomplete because the structure of the quote does not necessarily force the seller to expose all details of the offer.

Yet another limitation of some existing online RFQ systems is that they require the buyer to commit to buying a particular product before knowing all details of the deal. Commitment is perhaps acceptable when auctioning off commodity products. However, it is a serious problem when firms are choosing among a set of differentiated products, because the buyer may judge that none of the proposed solutions adequately meets its needs. This problem is compounded by the fact that computerized scoring rules (which quantify how well matched a particular product is to the needs of a particular person or firm) can usually only approximate, but not exactly identify, any person or firm's true valuation of the product and/or service. Finally, existing online RFQ processes require sellers to commit to their offers, even when the chance of buyer acceptance of any offer is small. The cost involved in sorting through volumes of RFQs can result in a limited number of sellers participating in the process. As the number of sellers increases, the probability that an RFQ received by a seller results in a sale decreases. At some point, the cost of reading and responding to many RFQs no longer justifies the number of sales made, resulting in abandonment of the system.

Thus, there is a need for an efficient online marketplace that allows buyers to submit complex requests for quotes cost-effectively, sellers to provide quotes cost- effectively and quickly, and buyers to evaluate these quotes efficiently.

SUMMARY OF THE INVENTION

The present invention addresses the foregoing by allowing sellers to pre-process the request for quotation (RFQ) using an automated "business rule". Sellers embed, within their business rules, the criteria determining whether the business rule generates a binding (i.e., "firm") offer or a non-binding (i.e., "indicative") offer in response to an RFQ. In some markets, the initial round of bidding may admit only indicative bids. Though not binding on sellers, indicative offers are useful because they help to generate a short list of potential sellers to the buyer for the final auction, reducing the total cost of the process for both buyers and sellers. Following the initial round of bidding, if a transaction is not closed based on one of the firm bids, indicative bids may be used to screen bidders in any of several ways. The following three examples, which are not exhaustive, illustrate some of the range of variations in the use of indicative bidding when all initial bids are indicative. First, the indicative bids may be communicated almost instantaneously to the buyer, who may invite sellers, based on their indicative bid, to engage in additional rounds of bidding or negotiations. This may be especially useful in situations where the buyer finds it valuable to refine its request based on the bids received. As a second possibility, the indicative bids may be kept secret and used only by the market maker (i.e., the entity that is running the RFQ method) to determine which sellers to invite to participate in a round of firm bids. A third possibility is that some information, such as rank order information, from the indicative bidding stage may be communicated back to the sellers, allowing them to decide whether to incur the cost of making a firm bid.

Automated bidding using business rules replaces the human cost of manually generating bids (a cost which is incurred each time an RFQ is submitted) with the cost of programming an automated rule (which is typically only a one time cost). This approach has four benefits. First, many sellers can participate in each auction. This allows even large auctions to be run at low total cost to all sellers and raises the probability that the winning seller will offer the buyer the most attractive deal. Second, each seller can participate in many auctions using the same business rule for each auction. The ability to reuse business rules saves sellers time and encourages them to specify their business rules precisely. Third, indicative bidding greatly expands the applicability of auction processes driven by business rules by allowing even imperfect rules to add significant value through improved screening and matching, without committing bidders to unwanted deals. Fourth, indicative bidding gives suppliers a migration path to fully automated auctions in markets where those are ultimately feasible by allowing imperfect business rules to add significant value to the process and by providing an incentive for bidders to improve their rules in order to become finalists in manual auctions they are likely to win and profit from.

It should be noted that the invention is described in terms of buyers submitting RFQs and sellers replying to RFQs for simplicity. However, the invention is also applicable when sellers submit RFQs to sell products and buyers reply to RFQs. Thus, both buyer-initiated and seller-initiated auctions are suitable for use with the present invention. Furthermore, the use of the term "RFQ" differs from standard usage in two respects. First, traditional RFQs list desired or required features but do not allow the buyer to express and quantify flexibility on features. For example, the buyer typically cannot specify a demand like "I need the product within 5 days. I will pay $50 extra for overnight delivery and $20 for 2 days delivery." The RFQs of the present invention allow more latitude in specifying preferences by incorporating information about willingness to substitute between product features and willingness to pay for certain features. Second, traditional RFQs do not ask buyers to state a desired price. However, it may be advantageous in some circumstances for buyers to state in advance either a binding maximum price or a price which indicates in broad terms their willingness to pay. A firm attempting to purchase, for example, a corporate jet may have a maximum budget of only $20 million. If so, they might want to state a price of $18 million so that sellers do not bother offering jets that cost $50 million. Thus, buyer RFQs of the present invention can list desired features (and willingness to pay for those features) with or without a price.

After a buyer submits an RFQ, a database of potential sellers is accessed and offers are generated automatically based on the details of the RFQ and characteristics of the individual sellers (including current inventory levels).

Offers are generated with "business rules" specified by each seller for each type of product. Business rules may indicate, along with details of conditions under which various offers will be made, whether an offer is a firm offer or an indicative bid. Once offers are generated, they can be processed in any of the several ways described above. One possibility, if both firm and indicative bids are allowed, is that the bids can be listed in order of preference, based on a scoring system, and transmitted in full or in part to the buyer. Based on this list, the buyer can do one of three things. First, the buyer can immediately purchase any firm offer on the list. Second, the buyer can contact any seller on the list and try to complete a sale. And, third, the buyer can select a subset of sellers (e.g., those sellers who submitted the top ten offers) that the buyer wishes to request more offers in an additional bidding round. These sellers are then notified electronically that they have been invited to provide another round of offers, either binding or indicative. Bids beyond the first round may be generated by a human or by a computer function, but requests for post first round offers are accompanied by a time limit for receipt of responses (for example, 24 hours). These offers may be collected, sorted, and transmitted back to the buyer all at once, or each offer may be transmitted back to the buyer as soon as it arrives at the market-maker. After each non-final-round of bidding, the buyer may invite a subset of bidders to submit another set of bids or to submit final, binding bids. This can go on indefinitely until the buyer buys something, gives up, or requests a final round of binding offers. hi some embodiments, requesting final binding offers obligates a buyer to buy something. In such cases, a sufficient number of sellers (perhaps only one) may be required to make offers "close enough" to their previous round offers or else the buyer is released from its commitment to accept an offer. In another embodiment, buyers may attach messages and documents (e.g., text fields, text documents, diagrams, or photos) corresponding to each of their responses to each non-final bid. For example, a buyer could tell one seller that they want a more powerful engine, tell another seller that they want a lower price, and tell a third seller that they want faster delivery. Sellers are free to ignore such information or to use it to modify their earlier offers. Sellers may also attach messages and documents to their offers. For example, a seller might attach a photograph of a forklift, testimonials from satisfied customers, a FAQ sheet, a map indicating the product' s location, a URL where additional information can be found, or a shipping schedule. Such messages provide additional information important to buyers and/or sellers not captured in a simple list of product attributes. The permissible messages may be constrained in some way if it is important to conceal buyers' and/or sellers' identities. By having initial offers generated automatically by business rules and screened before sellers must provide firm, binding offers, sellers who are unable to make strongly desirable offers are spared the time and effort of generating binding offers and, possibly, of intermediate rounds of indicative bids. This reduces both the cost to the buyer of having to evaluate numerous final offers and the cost to the seller (in time and effort) of having to generate a final offer that the buyer is highly unlikely to accept. This time and cost savings encourages a wider set of sellers to participate in the initial stage of bidding, thereby enabling sellers to search more widely for potential customers, and increasing the likelihood that the buyer will find the seller that is best able to meet its needs. Business rules can be defined by the individual sellers, based on the seller's own criteria, such as current inventories of various product types and quality levels. A business rule can be as simple as a firm offer by a particular merchant to sell a particular product at a particular price. If a company places its entire catalog into a database, each product-price pair becomes a simple business rule. The company could, however, place more complicated business rules. For example, a merchant with excess supplies of staplers might submit a business rule that automatically reduces the staplers' price two percent per day until all the staplers are sold. Another method of entering business rules is via an RFQ. When a buyer submits an RFQ, they quickly receive a list of indicative (or firm) offers. But in most embodiments, they are not required to purchase from the list of offers. They may instead decide to leave their RFQ in the buyer RFQ database as either a firm or indicative offer to buy. If a seller later submits an RFQ closely matching the buyer's RFQ, the two would be notified that their RFQs are a close match. If the buyer left its RFQ in as a firm offer to buy, the seller would be instantly notified that the buyer has committed to buy on the terms of its original RFQ. Thus, an RFQ is another simple, but important, form of business rule. The RFQ states the general terms on which the buyer is willing to buy. Should the buyer not buy, the RFQ can remain in the buyer RFQ database for matching to seller RFQs submitted in the future. (There is a similar seller RFQ database containing unfulfilled seller RFQs, downloaded product catalogs, etc.)

Business rules are applied to each new RFQ to generate new offers automatically. The buyer receives a list of offers in a very short time, thereby reducing the amount of time needed to complete a transaction.

An important advantage of automated bidding of this sort is that it allows a virtually unlimited number of sellers to bid in early bidding rounds without sellers incurring a high cost of generating initial bids or the buyer incurring a high cost of evaluating many offers.

Two factors motivate sellers to set their rules accurately. First, sellers who make spurious offers may acquire a reputation that damages their chances in subsequent auctions. The market-maker may facilitate reputation formation by recording and disseminating each auction participant's participation history. Sellers who renege on binding offers or frequently submit binding offers far worse than their previously offered indicative offers may find it harder to get invited to make binding offers. Second, because the initial round or rounds of bidding serve an important screening function for the sellers as well as the buyer, sellers who bid honestly are able to focus their bidding resources on those contracts that they have the best chance of winning profitably. Inaccurate bids at the first stage lead sellers to waste bidding resources on contracts for which their chances of winning profitably are small.

In one embodiment, once a buyer has requested final, binding offers and those offers have been sorted, the buyer is obligated to accept one of the offers, provided that a sufficient number of offers (perhaps even one) meet the minimum requirements set by the buyer or the market-maker. In another embodiment, the buyer can request partially binding offers (i.e., binding on sellers) but still enter into further negotiations with sellers. The sellers retain the option of adjusting their offers (e.g., price, quality, features) within a fixed set of parameters or by approval of the buyer. The sellers have incentive to negotiate because they know that the buyer is obligated to buy from a short pool of sellers, in which they are included, h addition to the buyer being bound by an RFQ, some embodiments may require sellers to be bound by the offers generated from their business rules. This provides extra incentive for the buyer to review the offers in detail and possibly enter into genuine negotiations, but it also discourages sellers from automating bidding on large contracts and from setting their business rules to make aggressive bids. Different embodiments can use various combinations of binding offers and RFQs at different stages of the process. And, as stated above, every enumerated embodiment has an otherwise identical embodiment but with the roles of buyer and seller reversed.

The present invention provides buyers and sellers with a mechanism which: (1) lets them define the product(s) and/or service(s) they desire in as much detail as they wish, (2) returns to buyers a list of the most appealing prospects and offers, possibly ranked and/or scored as to their predicted relevance, and possibly including detailed information about each option (or hyperlinks or other references to each option), and (3) allows buyers to purchase one or more products and or services and/or conduct electronic or offline negotiations with one or more of the sellers offering high ranking products. Buyers may be required to accept offers meeting certain minimum requirements during various stages of the process, and sellers may be required to submit binding offers during various stages of the process, especially if the buyer assures sellers that the buyer will accept one from a relatively small pool of offers.

This invention will be more fully understood in light of the following detailed description taken together with the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 is a block diagram of a computing system to carry out the inventive technique;

Fig. 2 is a block diagram of the architecture of the computing system of Fig. 1; Fig. 3 is a block diagram showing a network topology to carry out the inventive technique;

Fig. 4 is a flow chart illustrating steps according to one embodiment of the present invention;

Fig. 5 is a flow chart illustrating steps according to one example of the present invention; and

Fig. 6 shows a block diagram of the system according to one embodiment of the present invention.

Use of the same reference symbols in different figures indicates similar or identical items.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

Fig. 1 is a block diagram of a computing system for executing computer executable process steps according to one embodiment of the present invention. Fig. 1 includes a host computer 10 and a monitor 11. Monitor 11 may be a CRT type, a LCD type, or any other type of color or monochrome d splay. Also provided with computer

10 is a keyboard 13 for entering text data and user commands, and a pointing device 14 for processing objects displayed on monitor 11.

Computer 10 includes a computer-readable memory medium such as a rotating disk 15 for storing readable data. Besides other programs, disk 15 can store application programs including web browsers by which computer 10 connects to the Internet and the request for quote (RFQ) processing system according to one aspect of the present invention.

Computer 10 can also access a computer-readable floppy disk storing data files, application program files, and computer executable process steps embodying the present invention or the like via a floppy disk drive 16. A CD-ROM interface (not shown) may also be provided with computer 10 to access application program files and data files stored on a CD-ROM.

A modem, an integrated services digital network (ISDN) connection, or the like also provides computer 10 with an Internet connection 12 to the World Wide Web (WWW). The Internet connection 12 allows computer 10 to download data files, application program files and computer-executable process steps embodying the present invention's RFQ processing system.

Fig. 2 is a block diagram showing the internal functional architecture of computer 10. As shown in Fig. 2, computer 10 includes a CPU 201 for executing computer-executable process steps and interfaces with a computer bus 208. Also shown in Fig. 2 are a WWW interface 202, a display device interface 203, a keyboard interface 204, a pointing device interface 205 and disk 15.

As described above, disk 15 stores operating system program files, application program files, web browsers, and other files. Some of these files are stored on disk 15 using an installation program. For example, CPU 201 executes computer-executable process steps of an installation program so that CPU 201 can properly execute the application program.

A random access main memory ("RAM") 206 also interfaces to a computer bus 208 to provide CPU 201 with access to memory storage. When executing stored computer-executable process steps from disk 15 (or other storage media such as floppy disk 16 or WWW connection 12), CPU 201 stores and executes the process steps out of RAM 206.

A read only memory ("ROM") 207 is provided to store invariant instruction sequences such as start-up instruction sequences or basic input/output operating system (BIOS) sequences for operation of keyboard 13. Fig. 3 shows a typical topology of a computer network with computers similar to computer 10, connected to the Internet. For illustration purposes, three computers X, Y, and Z are shown connected to Internet 302 via Web Interface 202 through a gateway 301, where gateway 301 can interface N number of computers. Web interface 202 may be a modem, network interface card or a unit for providing connectivity to other computer systems over a network using protocols such as X.25, Ethernet or TCP/IP, or any device that allows directly or indirectly, computer-to-computer communications.

It is noteworthy that the invention is not limited to a particular number of computers. Any number of computers that can be connected to the Internet 302 or any other computer network may be used.

Fig. 3 further shows a second gateway 303 that connects a network of web servers 304 and 305 to Internet 302. Web servers 304 and 305 maybe connected with each other over a computer network. Web servers 304 and 305 can provide content to a user from database 306 and 307. Web servers 304 and 305 can also host the present RFQ processing system such that buyers and sellers may access the RFQ processing system according to the present invention. Also shown in Fig. 3 is a client side web server 308 that can be provided by an Internet service provider.

With the system described above or any other suitable system, the RFQ process of the present invention can be implemented. A few of the embodiments are described below.

A well-implemented computerized process for RFQs has multiple advantages compared to ordinary physical or electronic searches. For example, it spares buyers a great deal of search effort and sellers a great deal of bid preparation effort, and it also greatly improves the match for both. Partially automated RFQs are valuable for buyers because they reduce the need to search manually through manufacturer catalogs, travel to physical stores or warehouses, place telephone calls, send faxes, etc. to evaluate available offerings on multiple criteria. Instead, buyers can simply submit a single online RFQ and rely on software algorithms to evaluate and rank the offers received. Since seller offerings are located in a database (or collection of accessible databases), the computer can sort through all the offerings, rank them according to criteria determined by the RFQ and scoring algorithm, then return a list of offers or sellers matching the RFQ most closely.

After the initial round of fully automated bidding, the mechanism can allow further bidding rounds in which sellers may enter bids manually or via business rules. For the initial round of automated bids, search results are returned to the buyer within seconds or minutes rather than days as with traditional paper-based, voice-based, email- based and fax-based RFQs. This is particularly advantageous for buyers who need to acquire products and/or services as quickly as possible. If any bids are "firm" rather than "indicative" offers, the buyer can buy immediately. And even if only indicative bids are offered, the buyer receives an instant report listing those sellers most likely to offer the best deals.

It is noted that the description involves buyers (requesters) submitting RFQs and sellers (responders) responding to RFQs via "business rules" for ease of description. However, sellers can also submit RFQs for a "seller-driven" case, as opposed to the "buyer-driven" case described herein. The most general implementation of the invention entails a large number of buyers and a large number of sellers. Buyers or sellers can submit RFQs when they desire to buy or sell something. A newly submitted buyer RFQ will be matched against a database of seller "business rules". Previously submitted seller RFQs that have not yet been withdrawn constitute one class of seller "business rules". The description refers to buyer RFQs matched against a database of seller RFQs. However, the invention can also apply to seller-initiated RFQs, such as when a newly submitted seller RFQ will be matched against a database of buyer "business rules", where one class of buyer "business rules" can be previously submitted buyer RFQs that have not yet been withdrawn. The present invention involves a multi-step process. In step one, a buyer submits an RFQ, which is routed by the centralized marketplace to potential sellers for instant quotes and/or is matched against the centralized marketplace's database of business rules which generate instant quotes. Results returned from the database of seller "business rules" (i.e., rules which tell the centralized marketplace which offers for a product/service a particular seller is willing to make in certain situations) should automatically be returned to the buyer who submitted the RFQ. The results are at least partially based on the criteria specified by the RFQ. Additional bidding rounds are explained below. Business rules, indicative and firm bids, the user interface, and the auction are discussed in detail below.

BUSINESS RULES

A "business rule" is a program or piece of software which, when given an RFQ, calculates the terms (e.g., price, warranty, delivery time, delivery method, Mid return policy) a seller is willing to offer at a particular time to a particular buyer on a specified quantity of a particular product and/or service with various attributes and features for delivery to a particular location, either as a firm offer or as an indicative bid. One simple business rule is to simply enumerate all available product configurations with their associated catalog prices and offer to sell on demand the one that scores highest according to the scoring function associated with a particular buyer's RFQ. An example for the use of a scoring function is disclosed in U.S. Pat. Appl. Serial No. UNKNOWN, entitled "System for Multi-Dimensional E-Commerce Auctions", filed February 4, 2000, and incorporated by reference in its entirety. Another business rule would simply enumerate all unfulfilled and unrevoked RFQs residing in the RFQ database and offer them as "firm" or "indicative" bids, depending on the buyer's (or seller's) indicated preference. Automated business rules can be more sophisticated than simply enumerating all possible product/service configurations and setting a fixed price. More sophisticated business rules can depend upon buyer-specific characteristics and/or competitive market conditions. For example:

(1) Dynamic pricing is possible using automated business rules that adjust offers based on current inventory levels. The software implementing the business rule can query the real-time product inventory database, determine the current inventory level, look up the price adjustment factor (in dollars or as a percentage adjustment) as a function of current inventory level, and add or subtract the price adjustment factor to/from the standard price. The adjusted price could then be submitted as the business rule's price offer for customer consideration. (2) Prices and features can be preset to change automatically over time or when pre- specified conditions become true. For example, a seller seeking to eliminate excess inventory could preset price to fall two percent per day until the excess inventory is sold. (3) Volume discounts are possible when buyers desire to purchase more than one unit of a product. A business rule could define a schedule of prices based on how many units are purchased. A single unit might cost $100; two units might cost $90 each; three units might cost $80 each; four or more units might cost $70 each.

(4) Pricing and product features can be based on buyer-specific characteristics when the business rule receives information about the customer submitting an RFQ. Firms can, for example, offer discounted prices to regular customers or to buyers who own competitive products ("competitive upgrade" discounts are common among competitive software products).

(5) Pricing and product features can be based on competitive market conditions prevailing at that moment. For example, if a competitor has just offered a product at a lower price, with extra functionality, or with a bonus product, this information could be made available to other firms' business rules, allowing them to revise their offers. Competitive firms would therefore be able to adjust their current offers to what their competitors are currently offering or have recently been offering. Similar functionality might be implemented by raising price after sales rise and lowering price after sales fall. A business rule may take into account none, one, or any combination of the above features. For example, volume discounts might be based not on how many units are being purchased in the current purchase but rather based on total purchases by a single customer during a period of time. If a particular customer had previously purchased one unit of the product described in item (3) above and decided the following week to purchase two additional units, the business rule could determine a price of $140 ($80 x 3 - $100) instead of $180 ($90 x 2) by taking into account the particular buyer's previous purchase.

Business rules may be even more complicated. For example, if the seller desires to price based on its current inventory, it might grant a 10% discount on all current offers. But this discount may be exclusive or inclusive of other discounts. A buyer might be offered only the most valuable single discount or might be granted multiple discounts.

A business rule implicitly defines the tradeoffs a firm is willing to make to sell a particular product. So, business rules can be stated as a base configuration and price along with a price associated with each adjustment from the base configuration. For example, a firm selling timber faces higher costs to provide a better grade of wood, a larger piece of wood, faster delivery, etc. It could state (relative to the base price for a piece of oak "2x4") the price increment (decrement) it charges to supply each grade of wood better than (worse than) its standard offering (oak). Similarly, it might have a price schedule for larger piece of wood which states how much more expensive it is to buy a "3x5" instead of a "2x4" oak. "Standard" delivery might be 8 to 10 days, but "overnight shipping" might be available at a cost of 15% of the total value of the sale. Prices for product feature upgrades (or downgrades) may depend on which other product feature upgrades (or downgrades) have been selected. For example, the base price for an oak "2x4" might be $ 10. A maple "2x4" might cost $5 more than an oak "2x4". An oak "3x5" might cost $5 more than an oak "2x4". However, a maple "3x5" might cost $20 more than an oak "2x4".

Business rules can be implemented in any sufficiently powerful programming language. A preferred embodiment uses Java as a programming language. Other possibilities include C, C++, Perl, or rule-based languages such as Prolog. A business rule can be implemented at a generic level and be customized by a seller using seller- specific data. For example, in the case of the simple business rule mentioned above which simply enumerates all catalog products and prices, a seller's product catalog is sufficient data for customizing the business rule. More complex business rules require sellers to specify how they will adjust offers based on factors such as purchase volume, customer-specific information, prevailing competitive market conditions, and current inventory levels. These more complex business rules obviously require interfaces between the business rule and the appropriate data source. For purchase volume, the business rule needs access to the RFQ. For customer-specific information, the business rule requires access to buyer data (perhaps specified in the RFQ, but more likely available through the auction engine). To adjust offers based on prevailing competitive market conditions, the business rule may have access to recent historical market information, either raw information or aggregated/summary information such as the average percentage discount off of the catalog price offered on the previous ten products sold in the category. Lastly, to adjust offers based on current inventory levels, a business rule must access data on current inventory levels. This could be achieved in real time by linking to back office software (such as SAP); it could be achieved in near real time by frequently updating a database with product inventory levels. If this latter method is used, its usefulness can be improved by using a software program that keeps track of how many units have been sold since the most recent database update and/or estimates the remaining quantity by extrapolating based on depletion rates.

An added advantage of flexible business rules is that it allows merchants to get started easily and quickly (by simply dumping their catalog data into the seller RFQ database) while simultaneously allowing merchants the flexibility to gradually increase the complexity, sophistication, and effectiveness of their business rules over time.

INDICATIVE AND FIRM BIDS

Seller bids generated by business rules in response to a buyer RFQ (and any later non-final round of bidding requested by the buyer) can be either "indicative bids" or "firm bids". An "indicative bid", which may only generally describe the seller's offer, is a non-binding offer. On the other hand, a "firm bid" is a binding offer by the seller to sell at the proposed terms, the first round of bidding, this decision to submit an indicative or firm bid is made by the seller through the seller's business rule. A seller could, for example, have a business rule that makes firm offers to sell any number of Palm™ VIIs for $400 apiece to any and all buyers. A buyer who submits an RFQ to purchase a personal digital assistant with wireless Internet connection would receive back a list of the most likely deals. Some offers may be indicative; others may be firm offers.

Upon receiving a list of initial offers, a buyer can purchase any of the firm offers on the spot with no negotiation required. Alternatively, the buyer can contact any seller which submitted an offer and try to strike a deal. Or, the buyer can request another round of bidding after eliminating some of the initial bidders (sellers) from consideration. hi later bidding rounds, the decision to make a firm or indicative bid could be made by a person instead of an automated business rule. If a seller submits a firm bid, it promises to honor the terms of its offer if the buyer decides (within a specified time period) to purchase the product or service. In this case, the buyer can buy instantly if the buyer so desires. Making a firm offer may raise the chance that the offer will be accepted because it saves the buyer time. If a seller instead submits an indicative bid, the seller makes no promise to honor the terms of its offer to sell the product or service. Sellers may prefer to submit indicative bids because they reserve the right to review and modify the terms of their offer before committing to any deal. This enables the seller to delay being precise about the terms of the deal until the buyer is qualified as a likely buyer.

Buyers are free at any time to strike a deal with any seller they desire. For example, after receiving the initial list of bids, the buyer may accept any firm offers. And, if the market-maker allows buyers to see seller identities (or sellers to see buyer identities), the two sides may be allowed to strike a deal "offline", which they may be required to report to the market-maker.

After business rules have generated offers in response to the RFQ and the bids have been ranked, the buyer may accept the terms of one or more seller offerings. If the buyer wishes, the buyer can request a second round of bidding in which the buyer receives quotes from one or more sellers that submitted previous round offers, i.e., those initial offers generated automatically in response to the RFQ. At any time, the buyer can engage in negotiation with any seller, either electronically, offline, or in any combination thereof. The buyer can be guided in its negotiations by its lists of seller offerings. The first round thus helps the buyer narrow the field of potential sellers and products to a manageable list of top contenders while letting sellers avoid entering negotiations with buyers if numerous sellers have indicated a willingness to offer better terms. Further, the terms of the deals offered by round one bidders constitute a useful set of background facts for further discussions between the buyer and various vendors. USER INTERFACE

Buyers enter RFQs in a structured format that allows for partially or completely automated processing. The RFQ format varies by industry and product and is specialized to contain particular attributes that describe a certain class of products or services.

In one embodiment of the invention, this information is entered into an HTML form that guides the buyer through providing the information in the required format. In addition to being industry and product specific, this form may vary by the buyer. Certain types of buyers may be used to or most comfortable with a particular way of providing the required information. For example, some buyers might prefer to express their preferences in general terms (e.g., to state that the CPU speed of a computer is "not very important" to them but that the price of a computer is "very important"), whereas others might want to quantify their preferences more precisely (e.g., to state that they will not consider purchasing computers with CPU speed less than 400 MHz (in mathematical scoring terms, computers with slow CPUs will receive a score of minus infinity) and that they place a $100 value on each additional 50 MHz of CPU speed above 400 MHz).

Instead of providing offers manually in response to individual buyer RFQs, sellers use a seller interface to define a set of business rules that are applied automatically by a computer system whenever a new RFQ is received. Automatic response by business rules is always used in the first round of bidding to speed up the process of qualifying bidders. In subsequent rounds, firms may choose to use automatic response by business rules or manual response by humans. Their decision may depend on factors such as the time provided for making an offer, the dollar value of the transaction, the number of bidders accepted at each bidding round, the cost of reading an RFQ and formulating and transmitting an appropriate offer, and the difficulty of formulating business rules capable of making reasonably appropriate offers.

Even though defining a business rule is significantly more costly than providing an individual offer, the business rule can be applied repeatedly and automatically on various RFQs from possibly different buyers at negligible per-use cost. Because this eliminates the need for human intervention during the first round of bidding, the total cost of processing an RFQ is lower than with existing methods. Only sellers whose business rules result in a close match to the RFQ will incur a non-trivial cost to respond to the RFQ. This significantly reduces, but does not always eliminate, the usefulness of fees and other disincentive mechanisms in discouraging buyers from submitting large numbers of RFQs. The present invention allows for such fees and/or other methods. Disincentive mechanisms, such as fees, should be adopted to the point at which the benefit to the buyer of submitting another RFQ less the fee equals the cost to all sellers who must respond to the extra RFQ plus any fee the sellers might receive for responding to the RFQ (which might come out of the fee paid by buyers for submitting an RFQ).

AUCTION

Once the initial bids have been made, the buyer and sellers may be given full or partial information about the outcome of the auction. For example, they may be informed of the full list of bids, or just the bids of the top-ranked sellers, or just the ranking of the sellers. The information provided can be different for different sellers in ways that may depend on the bid ranking, and the buyer may be provided different information from any of the bidders. For example, the buyer could be informed only about the offers of the top-ranked sellers and the sellers could be informed only of the score of the best rejected bid in the previous round and the number of sellers still under consideration.

Another implementation would allow the market-maker to keep offers confidential from all parties and screen sellers based on their indicative bids rather than report details of any bids. In electronic markets, as in physical markets, bidders will frequently want to avoid tipping their hands at the indicative stage. Some market- makers will choose to apply a criterion to the bids to narrow the initial field of sellers to a shorter list of "qualified" sellers. The market-maker might then report to each qualified seller limited information about the outcome of the previous round of bidding, such as: "You are now one of seven bidders still in contention and are asked to make an indicative or firm offer by 2pm tomorrow" or "You are now a finalist according to the pre-announced market rules and are asked to make a firm offer by 2pm tomorrow." Negotiation would begin in particular earnest after the final round, when firm offers have been made by sellers, especially if the buyer has pre-committed to purchase from one of the sellers. This process does not require use of anything as sophisticated as the Vickrey auction at the non-final bidding stage because indicative bids do not establish an actual offer or determine the final terms. Simple sealed-tender offers are suitable with secret indicative bids, hi some embodiments, the market-maker receives non-final- round bids in confidence and uses them only for selection.

When a buyer submits an RFQ, the resulting offers are available to the buyer almost immediately (i.e., within seconds or minutes). A scoring function that is parameterized using the data in the RFQ is used to assign a scalar score value to each offer. This allows offers to be compared directly, greatly simplifying the decision making process for the buyer.

A tradeoff is faced in designing the interactive RFQ process. On the one hand, it is desirable that information flows freely. The RFQ process is a valuable part of decision-making for both buyers and sellers. This suggests that there should be no commitment imposed on buyers and sellers unless the buyer explicitly requests only binding offers and (possibly) binds itself to purchase from the list of sellers. Otherwise, buyers should not be required to make any purchase after submitting an RFQ, and sellers should be permitted to submit non-binding indicative offers. On the other hand, this lack of commitment might create undesirable incentives. If requests and bids are not binding, buyers and sellers have little motivation to be sincere when making initial requests and bids.

Depending on the objectives in any particular application, one or the other goal could be met, but not both, or both could be partially met. A workable automated RFQ procedure represents a compromise between these incompatible objectives. The RFQ procedure according to the present invention provides the following: (1) features that address both of these goals, (2) flexibility, so that the goals can be weighted differently in different applications, and (3) allows bidders to request only final, binding offers if they so choose. In one embodiment of the present invention, the buyer can request binding offers from a subset of sellers. Sellers must follow through on all offers made. When all offers have been generated, the buyer is presented with a list of binding offers from sellers, the list being sorted by the score value of the offers. That way, offers best matching the buyer's request are displayed at the top of the list, allowing the buyer to quickly decide which offers to consider seriously, i a more specialized scenario, the buyer is obligated to accept one or more of the offers, provided at least one of them has a score value that exceeds a reserve value previously stated (perhaps secretly) by the buyer. This special case has the advantage of inducing sincere bidding, but at the cost of reducing the information-provision role of the bidding process. Buyers might be reluctant to use this method if they are unsure what is available. hi another embodiment, a first round of bidding via business rules is used to generate non-binding, indicative bids (or, should a seller choose, binding bids) that are used to narrow down an initially large group of seller-product pairs to those seller- product pairs that are most capable of and willing to make competitive offers in response to a particular RFQ. This can be done repeatedly (but using a mix of human- generated offers and automated business rules) until the number of potential seller- product pairs has been reduced to a level determined either before the bidding round or by the buyer following the bidding round. At some point, the buyer may request some remaining sellers to each provide a binding offer. This offer may be carefully calculated by a human or determined by a computer algorithm. (For example, the computer could be asked to resubmit the exact same offer as in the indicative bidding round.

Alternatively, the offer could automatically be made more or less competitive by a predetermined amount.) Human intervention following the initial bidding round results in more robustness for inaccuracies in the rule-generated offers. This simplifies the process of defining the business rules and reduces the risk associated with poor choice of computerized business rules. h yet another embodiment, buyers receive a list of indicative quotes, and final, binding quotes are provided by sellers only if the buyer commits to accept at least one of said binding quotes. This ensures that the buyer's request is serious, without incurring any cost to the buyer. hi the case in which offers are not binding and/or buyers are not committed to actually making a purchase, additional devices might be added (depending on the specific application). For example, buyer sincerity might be assured by charging the buyer a fixed fee, small relative to the total value of the purchase, for making a RFQ. Seller sincerity might be assured by:

(a) Requiring that those sellers who are selected for a second or subsequent round of bidding, beyond the indicative bidding round, must make offers that are at least as good as those they made in the indicative round, where "at least as good" is measured by the buyer's pre-announced scoring rule. This provision may, however, be relaxed at the buyer's discretion, i.e., the buyer may agree to consider an offer that, according to the scoring rule, is inferior to an earlier offer. (b) Allowing sellers selected for a second or subsequent round of bidding to adjust their offers so long as they are not "significantly worse" for buyers than the seller's original offer, where "significantly worse" is measured by the buyer's pre- announced scoring rule. As in point (a), this provision could be relaxed at the buyer's discretion. (c) Giving buyers information about past behavior of sellers. If a buyer claims that a seller, having been selected as a result of its bid in the indicative bidding round, subsequently makes an inferior offer (in effect reneging on its earlier bid), then that information is made available to buyers in subsequent RFQs in which the seller participates. In all embodiments discussed above, sellers may or may not be permitted to attach messages to their offers and buyers may or may not be permitted to attach messages to their requests for further bids. Such messages could provide much greater detail about the buyer's needs or the seller's product(s) and/or service(s) than can be automated in the auction engine. Buyers could also use such messages to indicate areas of particular concern to them about particular offers. For example, they might like an offer except for "7 day delivery." The buyer might send a message indicating that they really need the product within 3 days. The seller could take that message into account when formulating another offer. Such messages might be conveyed only when buyers request final, firm bids or when sellers submit final, firm bids. Fig. 4 is a flow chart illustrating steps according to one embodiment of the present invention, described in various sections above. The various blocks in the flow chart are arranged to illustrate which of the "parties" performs each step. The "parties" to this process are the buyer, each seller, and the marketplace or computer program, hi step 401, the buyer enters an RFQ, which gets transmitted to the marketplace in step 402. In step 403, the marketplace creates a scoring function based on the criteria in the RFQ and ranks or weighs various qualities of the product or service in order of importance. For each seller in the marketplace having suitable business rules for the RFQ, the marketplace, in step 404, applies the scoring function, hi step 405, the list of bids from suitable sellers is ranked, with the most attractive bid at the top of the list.

The buyer then receives and evaluates this list in step 406. As described above, the list can be grouped into sellers having bids within certain scoring ranges. The buyer then decides in step 407 whether there are any bids on the list that the buyer is willing to accept or commit to. If no, then the process ends. However, if there are bids the buyer may be willing to accept, the buyer selects those bids in step 408 for further negotiations, at which time the buyer is bound to accept a bid from one of the remaining sellers. (This obligation can be conditional on the final bids being within a certain delta of the initial acceptable bids.) These selected bids are transmitted to the marketplace, which prompts the corresponding sellers in step 409 to generate binding offers. Each seller then generates binding quotes or offers, which can be done manually or with a computer program, in step 410 and transmits the offers to the marketplace. In step 411, the marketplace sorts this list of binding offers, either using the same scoring function of the earlier list or some other function. The resulting list, which is shorter than the previous list, is then transmitted to the buyer in step 412, where the buyer then reviews this shortened list and selects an offer to accept. The buyer transmits this information to the marketplace in step 413, which then accepts full or partial payment from the buyer. This can be accomplished by an automatic debit from the buyer's pre-approved credit card or some other method. In step 414, the marketplace notifies the seller, either by email, fax, or another suitable methods, that the seller's offer has been accepted. The seller receives the bid and payment terms from the marketplace in step 415 and subsequently delivers of the product or services to the buyer in step 416. Fig. 5 is a flow chart illustrating a more specific example of a process that implements one of the invention's embodiments described above. The other embodiments of the invention can be implemented in an analogous way. This example is only intended as an illustration of the general process. Details of actual implementations of the invention may vary. An example is provided below. EXAMPLE

LUMBERCENTRAL.COM is a (fictional) online marketplace for lumber using the inventive technique. It runs a fully automated software application which implements the invention and does not require any human intervention except from buyers and sellers. A number of lumber wholesalers participate in this marketplace. They have provided a set of business rules that are capable of generating quotations when given a request for quotes (RFQ). The software that runs the online marketplace implements business rules.

A buyer intends to purchase a certain quantity of lumber. He visits LUMBERCENTRAL.COM in step 501 using a web browser, h step 502, the buyer fills out an online form with a specification of the dimensions, wood type (maple, oak, etc.) and quantity requested. Also, the buyer provides a subjective rating of how important the quality rating of the lumber is to him and an indication of how urgently delivery is required. In addition, the buyer specifies how the goods are to be delivered and a penalty he would like to charge associated with late delivery. When the buyer feels that the contents of the form he filled out adequately represent his needs, the RFQ is submitted to the marketplace in step 503.

In step 504, the marketplace reads in the RFQ information and uses the information to parameterize a scoring function in step 505 that maps each seller's offer to a rational number. The better an offer matches an RFQ, the higher this number.

(What constitutes a "better" match in this context depends on the choice dimension, as specified by the buyer. A buyer who wants maple obviously prefers maple to other woods. However, a buyer who indicates that they place a low valuation on wood quality would still prefer the highest quality wood. Even if a buyer is not willing to pay much to get high quality wood, an offer with better quality wood would be considered a

"better" offer.) The marketplace then applies each of the seller's business rules to the RFQ in step 506, which results in a list of offers. A short time after the buyer submits the RFQ, seller offers are presented to the buyer in a sorted list in step 507, sorted by the score calculated using scoring function calculated from the data in the RFQ. If any offers are "binding" rather than indicative, the buyer can buy instantly. In this case, however, all bids are indicative, so the buyer then chooses perhaps fifteen offers in step 508 that scored highly and also otherwise seem advantageous to the buyer. For example, next to each offer he sees the seller's historic deviation between scores of final offers and scores of initial, indicative bids. A seller whose final bids average, for example, 20% higher than the initial bids, is less attractive to him.

At this point, the buyer could narrow the list further and request another round of non-final bidding. Instead, however, if time is critical to the buyer, the buyer can proceed directly to requesting final quotes.

By submitting another online form in step 509, the buyer specifies a list of sellers from whom he would hke to receive additional quotes. To each request for quotation, the buyer can attach messages specific to the seller, such as "I like your offer, but I really need to receive delivery within three days." The buyer can attach such messages by attaching documents, typing in text boxes, or selecting options from pulldown menus. These messages are passed via the central marketplace to sellers who are asked to submit final quotes.

In this marketplace setup, requesting a set of final binding quotes commits the buyer to accepting at least one of the final offers that are to be generated by the sellers. By submitting the second online form, the buyer enters a legally binding agreement with the marketplace to purchase from one of the final bidders. Upon receipt of the second form, which requests final offers from sellers in step

509, the marketplace transmits the RFQ to the sellers chosen by the buyer. Sellers can then submit final binding offers in step 510. Sellers have, for example, 48 hours, to submit a final, binding offer to sell the desired lumber to the buyer (though the buyer is free to accept any final offer at any time). A final offer must contain all details of the deal, including price, material, shipping, etc. A final offer may also (at the seller's discretion) provide additional information to the buyer. By returning this binding offer to the marketplace, the seller enters a legally binding agreement to deliver the goods under the terms described in the final offer.

When a final offer is received by the marketplace, the offer is immediately transmitted to the buyer. The buyer can accept any final offer at any time, hi this case, however, the 48 hour deadline elapses before most sellers have submitted final offers. Sellers who fail to submit final-round offers automatically commit to the initially non- binding offer made in the prior non-final bidding round. The marketplace stores offer information in a database that is used to calculate statistics on each seller's bidding track record. For example, the marketplace records the discrepancy between a seller's final round bid and its prior round bid. This is used to update the average discrepancy between that particular seller's final round bids and their prior round bids.

After the 48 hour period ends, the marketplace, in step 511, notifies the buyer by email of all the final offers. In step 512, the buyer is given a specified period of time to choose one of the offers. The buyer may or may not be permitted to negotiate further with sellers before making their purchase decision. When the buyer has chosen, the buyer submits a third online form indicating the full terms of the selected offer. The marketplace in step 513 automatically deducts the amount of the down payment of the transaction from the buyer's account and then contacts the selected seller in step 514 via email to notify the seller that the offer has been accepted. In step 515, the seller delivers the goods as specified in the offer.

Fig. 6 shows a block diagram of components for implementing the present invention. Fig. 6 shows a user interface (UT) 600, utilized to generate a request from a potential buyer. UI 600 can be accessed by a buyer via the Internet on computer 10 (Fig. 1) connected to the Internet or to a network. Generally, a buyer's request may include the buyer's information, an attribute set type that describes the product or service required by the buyer, a scoring function and any other information that facilitates the auction.

Buyer information is input into buyer's UI 600 and sent to a request generator module 601 that generates the buyer's request or RFQ and specific scoring function from the RFQ. Request generator 601 uses buyer information entered via UI 600 and any historic information available in a database 600A. Database 600A can be compiled using historic marketing data that provides customer tastes and preferences.

Request generator 601 sends the buyer's RFQ and scoring function to an offering generator 603 that is linked to an offering database 604 and a business rule repository 606. Offering database 604 is populated by using seller's information entered via a seller's UI 605, and/or from historic data stored in a seller's database 605 A. UI 605 can be accessed by a seller on a computer connected to the Internet or to a network.

A business rule uses data and rules provided by the sellers to construct each seller's offers, in response to the buyer's request containing a scoring function.

Business rule repository 606 provides various types of business rules. A seller can choose the most suitable one for a given product and parameterize it. Different business rules may use different algorithms to generate the offers.

After offering generator 603 receives the buyer's RFQ and scoring function, offering generator 403 obtains seller's information from offering database 604 and also identifies the business rule corresponding to the buyer's scoring function and RFQ. Offering generator 603 generates a set of offers in response to the RFQ. The set of offers is based upon seller's information, buyer's RFQ and scoring function, information from the offering database 604 and the business rule provided by business rule repository 606. Offering generator 603 sends the offers to an auction engine 609.

Auction engine 609 also receives the buyer's RFQ and scoring function from request generator 601.

Auction engine 609 applies the scoring function to each offer and generates a list of offers ranked in order of desirability to the buyer. Auction engine 609 sends the list of offers to an offer presenter 611 that presents the list to the buyer, such as by email via the buyer's UI 600. If the buyer decides that at least one of the offers appear to be acceptable, the buyer selects the one or more acceptable offers and notifies offer presenter 610 via buyer UI 600. Offer presenter 610 transmits this information to auction engine 609, which relays the information to seller's UI 605 and requests each corresponding seller to submit a binding offer. Each seller, via seller's UI 605, generates a binding offer, which is transmitted to auction engine 609, either directly (manually generated) or via offering data base 604 and offering generator 603 (computer generated). Auction engine 609 then ranks the offers using a scoring function or other criteria and transmits a list of ranked offers to offer presenter 610. Offer presenter 610 supplies the list to the buyer, via buyer's UI 600. The buyer then selects the desired offer and transmits that information to offer presenter 610, which notifies the appropriate seller via auction engine 609 and seller's UI 605. Upon completion of payment, the product or services are delivered to the buyer.

The above-described embodiments of the present invention are merely meant to be illustrative and not limiting. It will thus be obvious to those skilled in the art that various changes and modifications may be made without departing from this invention in its broader aspects. Therefore, the appended claims encompass all such changes and modifications as fall within the true spirit and scope of this invention.

Claims

CLAIMS We claim:

1. A method of performing a request for quote (RFQ) process involving a requester and a plurality of responders, comprising: receiving an RFQ from a requester, wherein the RFQ comprises information about the requester's needs and the requester's preferred terms for purchasing or selling a product or service; applying business rules obtained from at least one of the plurality of responders to the RFQ; automatically generating quotes based on the business rules; and storing the quotes for further processing.

2. The method of Claim 1, wherein applying business rules commits the responders corresponding to the business rules to all quotes generated therefrom.

3. The method of Claim 1, further comprising sending the quotes to the requester.

4. The method of Claim 1, further comprising parameterizing a scoring function from the information in the RFQ to assign a numerical score to each responder's quote.

5. The method of Claim 4, further comprising sorting quotes based on the results of parameterizing the based on requester preferences and sending the sorted quotes to the requester.

6. The method of Claim 1, further comprising: ranking the quotes in order of desirability to the requester; transmitting the ranked quotes to the requester; selecting, by the requester, a set of suitable quotes; requesting, from each responder corresponding to a suitable quote, a binding quote; and generating, from each of the corresponding responders, a binding quote.

7. The method of Claim 6, further comprising automatically sorting each of the binding quotes.

8. The method of Claim 7, further comprising transmitting the sorted binding quotes to the requester and selecting one of the binding quotes.

9. The method of Claim 6, wherein the desirability is automatically ranked using a scoring function.

10. The method of Claim 6, wherein the selecting binds the requester to accept one of the quotes from the selected responders.

11. The method of Claim 6, wherein the generating is manual.

12. The method of Claim 6, wherein generating is computerized.

13. The method of Claim 10, wherein the requester is bound only if a predetermined number of final quotes are better than or within a pre-determined range of previous indicative quotes from the requester.

14. A method of facilitating online transactions, comprising: upon receipt of a request for quote (RFQ) from a requester, automatically generating a list of offers from a set of business rules; ranking the list of offers based on a scoring function, wherein the scoring function indicates the desirability to the requester of a specific offer; providing the list of offers to the requester; selecting, by the requester, a number of possible offers from the list of offers; transmitting, to responders corresponding to the selected offers, an intent to buy; generating, by each corresponding responder, a binding offer; and transmitting the binding offer to the requester.

15. The method of Claim 14, wherein the generating is manual.

16. The method of Claim 14, wherein the generating is computerized.

17. The method of Claim 14, wherein the set of business rules is unique to each responder.

18. The method of Claim 14, wherein the number of possible offers is less than the list of offers.

19. A computerized system for facilitating online transactions between a requester and at least one responder, comprising: a request generator, wherein the request generator generates a requester's request for quote (RFQ); an offering generator coupled to the request generator, wherein the offering generator, in response to the RFQ, automatically generates a plurality of offers based upon responder data stored in an offering database; and an auction engine coupled to the request generator and offering generator, wherein the auction engine generates a list of offers from the plurality of offers.

20. The system of Claim 19, wherein the offering database includes business rules corresponding to the at least one seller.

21. The system of Claim 19, further comprising a requester interface coupled to the request generator and the auction engine, wherein the requester interface indicates which offers from the list of offers are acceptable.

22. The system of Claim 21, further comprising a responder interface coupled to the offering generator and the auction engine, wherein the responder interface transmits information about a binding offer in response to the list of acceptable offers.