WO2001031506A2

WO2001031506A2 - System for tracking and analyzing data changes in computer databases

Info

Publication number: WO2001031506A2
Application number: PCT/US2000/029169
Authority: WO
Inventors: Charles Nicholls; Ranvir Wadera
Original assignee: Customatics, Inc.
Priority date: 1999-10-22
Filing date: 2000-10-19
Publication date: 2001-05-03
Also published as: AU8030600A; WO2001031506A3

Abstract

A system allowing a business users and analysts to use software to track and analyze segments, joiners, leavers, and define probes and other mechanisms to facilitate real-time analysis of data changes in a computer database. Joiners are members who have recently joined a segment. Leavers are members who have recently left a segment. Probes are re-usable expressions for deriving a value from properties of a segment member, joiner, leaver or other item. A probe can be a flag, simple or complex relational expression, or other expression. An example of a probe value is the average purchase interval for a given segment. Probe values are typically tracked over time to provide meaningful analysis data. Probe expressions can be associated with segments' member, joiners, leavers and / or associated with individual members of segments and other aspects of the system. By using probes in combination with segments or individual segment members, the need for running slow, expensive queries against large amounts of detailed data is reduced or eliminated. This allows important analytical data to be available in real-time. Alerts and actions can be triggered according to probe values or other conditions, called ChangeAgents. Such alerts and actions can result in email messages being sent; or mail, fax or other communications being sent to members of a company or to customers. A graphical user interface is provided to provide an efficient and intuitive means for an analyst or operator to implement the above features. Other aspects of the system include monitors for associating rules to segments with scheduled activation, and vectors for indicating the direction that a data item (e.g., a customer) is moving over time.

Description

SYSTEM FOR TRACKING AND ANALYZING DATA CHANGES IN COMPUTER DATABASES

CLAIM OF PRIORITY The present application claims priority from U.S. Provisional Patent Application Serial No. 60/160,921, filed October 22, 1999, the contents of which is hereby incorporated by reference for all purposes.

BACKGROUND OF THE INVENTION This invention relates in general to processing of data in computer databases, and more specifically to a system for tracking and analyzing changes in computer databases. There is great commercial value to businesses in knowing the behavior of their customers. Knowledge of changes in a customer's buying habits can help the business decide whether to take action to build customer loyalty, encourage a customer to purchase more, or to prevent the customer from choosing to do business elsewhere. Such knowledge also allows sellers and distributors to keep inventories full and can provide other benefits in any database application where it is advantageous to track and analyze data changes.

A traditional approach to tracking data changes, such as customer behavior recorded in a database, is to compile data about the customers, their buying habits, and other properties. Segments are then defined to provide a general category for different classes of customers. For example, if the business is an Internet point-of-sale site, customers may be segmented by what products they buy, the length of time they have been buying products, their total purchases from the company, etc.

There is a great value in understanding why, and how many, segment members are joining or leaving a segment, and to know to which segment members are migrating. For example, customers who have just left the high value segment should trigger sales or marketing efforts directed at maintaining those customers in the high value segment. Ideally, it is better to know that a customer is tending to leave a segment before the customer actually leaves the segment so that preventative measures can be taken. However, most segmentation done today is not in real time. There is often a substantial time lag involved in identifying to which segment a new customer/member belongs, or to track the migration of particular members between segments. Today, companies carry out extensive and complex analysis manually to track segment membership (or other data) changes - but this process is time consuming, requiring a high level of skill which leads to a time delay in producing the analysis. This time lag between the data change and being able to take action on the analysis of the change is measured in weeks or months at many companies. Also, much analysis requires running slow, expensive queries against large amounts of detailed data in a database.

In most businesses today cycles need to get shorter, and the business environment more competitive. Marketing departments are not immune from this: customer retention is a major corporate priority for many businesses today, and as more business is transacted on-line, there is increased pressure to reduce the cycle time.

Thus, it is desirable to provide a system that allows faster tracking and analysis of database changes. Such a system would have widespread application and benefits in data analysis in many fields as, for example, to provide improved marketing and customer relations. Today, tracking data changes requires a high level of skill. The task is typically performed by data analysts or by staff trained in technical database skills. It is highly desirable to provide a means, via an easy-to-use graphical interface, so that non-technical business managers can do this analysis themselves.

SUMMARY OF THE INVENTION The invention allows an analyst to use software to analyze segments, joiners™ and leavers™; define probes™ and by associating them to segments, joiners and leavers, to facilitate real-time analysis of data changes. A segment is a category of customers, products, suppliers etc. defined by criteria. For example, customers can be segmented by the number and amount of purchases they make in a month. Joiners are members who have recently joined a segment. Leavers are members who have recently left a segment. Probes are re-usable expressions for deriving a value from properties of a segment member, joiner, leaver or other item. A probe can be a flag, simple or complex relational expression, or other expression. An example of a probe value is the average purchase interval for a given segment. Probe values are typically tracked over time to provide meaningful analysis data. Probe expressions can be associated with segments' member, joiners, leavers and/or associated with individual members of segments and other aspects of the system. By using probes in combination with segments or individual segment members, the need for running slow, expensive queries against large amounts of detailed data is reduced or eliminated. This allows important analytical data to be available for real-time analysis. Alerts and actions can be triggered according to probe values or other conditions referred to as Change Agents™. Such alerts and actions can result in email messages being sent; or mail, fax or other communications being sent to members of a company or to customers. A graphical user interface provides an efficient and intuitive means for an analyst or operator to implement the above features. Other aspects of the system include monitors™ for associating rules to segments with scheduled activation, and vectors for indicating the direction that a data item (e.g., a customer) is moving over time. Other features are described below. One embodiment of the invention provides a method for tracking data changes in a database, the method using a computer having a display screen, processor and user input device, the method comprising using the processor to define a plurality of segments to classify the data; accepting signals from the user input device to associate a probe expression with at least one of the segments, wherein the probe expression derives a value based on one or more data items in the associated segment; and displaying the derived value on the display screen.

BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 shows the conceptual approach of the present invention; Fig. 2 shows an example of user interface 160 for assigning a probe function to a segment, or category;

Fig. 3 shows an example of a web page of a portfolio for a Segment Manager;

Fig. 4A is a first illustration of a user interface for adjusting segments; and Fig. 4B is a second illustration of the user interface for adjusting segments.

DETAILED DESCRIPTION OF A PREFERRED EMBODIMENT

The present invention provides a database system for analyzing and processing data. A preferred embodiment of the invention is used to analyze customer purchasing habits and trends, however, the invention can be applied to other applications such as tracking changes in supply chain management, risk management, product/category management, etc. A preferred embodiment of the present invention is provided in a suite of software programs manufactured and distributed by Business Objects, SA including "Customer Intelligence" and "Application Manager." Third-party software used by the preferred embodiment include "Customer Intelligence" and "Application Manager" by Ithena, Inc., and Access and Internet Explorer by Microsoft Corporation.

As will be seen, any suitable software applications can be used to implement different facets of the system of the present invention. The analytical tools, user interface, and other aspects of the invention are adaptable to any type of data - not just the customer and marketing examples shown herein. Certain aspects of the invention, such as specific user interfaces and tools, are not commercially available and have been custom-developed by the applicant. It should be apparent that the concepts comprising the system of the present invention may be implemented in many different ways using various applications, databases, computer platforms, devices, etc. Moreover, the present invention can be practiced with one or more processes or functions of the invention located remotely from the other processes or functions. For example, a user can be in one geographic location operating a user interface according to the present invention while the database is geographically distant. The user can communicate with the database user interface by, for example, using a computer network.

Fig. 1 shows the conceptual approach of the present invention. In Fig. 1, system 100 includes various definitions and constructs which are illustrated by items in the figure. Rectangles such as 102, 104, 106, 108 and 110 represent "segments" or categories that are used to classify customer entries. Such categories are typically given common names. For example, category 102 refers to a "Collector" category. Similarly, categories 104, 106, 108 and 110 refer to "Explorer," "Typical," "Low" and "Novice" categories, respectively. The present example deals with a wine retail business where it is desirable to categorize, and track, customers according to their wine-buying habits.

Each customer entry, symbolized by an oval, such as customer 120 includes descriptive information about a specific human customer. The descriptive information can include values, parameters, text or other attributes and will vary from application to application. For example, for purposes of illustration, the customers can have attributes as follows: location, average monthly purchase dollar amount, number of different types of units purchases, number of purchases per month since the customer has been associated with the system, time that the customer has been in the system, etc. One database implementation may store such characteristics as attribute- value pairs. So, for example, the attribute "location" could have the value "San Francisco." Another implementation may use a record format where each customer is a record that has pre-defined fields which, in turn, contain values such as numbers, text, pointers to other fields or values, etc. In general, the specific database representation is not critical to implementing the present invention. In Fig. 1, each category is defined by one or more criteria. As shown in Fig. 1, the "Collector" category is defined by the criteria of customers' having average purchases over $200 a month AND having purchased at least two different types units per month on average. Likewise the "Explorer" category has the criteria of customers having average monthly purchases of between $200 and $50 per month AND where the customers' have purchases of a number of different types greater than two.

The "Typical" category requires the customer have an average purchase below $50 per month. The "Low" category is for customers who have not purchased any bottles of wine in the last two months. Finally, the "Novice" category is for customers who have been in the system less than two months.

Note that other rules for placing customers into categories may come into play other than the explicit criteria discussed above. For example, the categories may be parsed from bottom to top so that any customers fitting into a lower category will not be checked to see if they will fit into an upper category. In some implementations of the system, a customer may fall within two or more different categories. Still, other customers may not fit into any pre-defined categories at all. Various possibilities can be used to handle this situation as appropriate for the specific application. For example, a miscellaneous "Catch- All" category can be defined. It is important to note that such segmentation categories need not be restricted to customers only. They can include schemes for risk analysis, supply chain optimization etc.

Given the category definitions shown in Fig. 1 , new customers start out in the "Novice" category. After two months in the "Novice" category, the system will have obtained at least some information about customer 120's buying habits. Customers are moved automatically into the "Low" category if they have not made any purchases in two months. This is shown by the arrow extending from customer 120. Other customers, who have made purchases within their first two months in the system, may be re-assigned to the "Typical" category. Still other customers may immediately, after two months, be placed into the "Explorer" or "Collector" categories. Similarly, customers can migrate between any of the categories based on their changing characteristics. Since these characteristics usually include some aspect of the customer's buying habits, the migration into, and out of categories, is indicative of customer's behavior. Further, as is discussed below, predictions can be made and trends identified which are useful in letting the business operating the system of the present invention know how to proceed in order to maximize customer revenue and to keep customers.

One possibility is that customers leave the system completely. This is illustrated in Fig. 1 where customer 122 is removed from the system. This can occur, for example, where a customer closes their account, where a customer is taken out of the system by the system operator for purposes of special handling, etc. Customers can belong to multiple segments. The preferred embodiment also allows for overlapping segments where one or more criteria of one segment overlap with another segment.

Another concept of the present invention that is illustrated in Fig. 1, is the use of "probes." A probe is a function, usually in the form of a relational expression that used one or more customer characteristics. Additionally, probe functions can use category criteria, other probe functions, or any other value, characteristic, function or relationship that is provided by the system of the present invention. Probes are useful to set up a function that outputs useful information. For example, Probe 130 of Fig. 1 can be set up to compute the average expenditures of customers in the "Explorer" category each month. Thus, probe 130 would automatically execute on the first day of a month to compute the previous months total expenditures of the customers in the "Explorer" category. Note that probe 130 only uses customer characteristics within a single category. Also, probe 130 would need to know which customers reside in the "Explorer" category. This can be handled in various ways in the database which may provide, for example, a list of pointers to all customer objects within the "Explorer" category, a flag, or tag id associated with each customer record or regular object that is in the "Explorer" category, etc.

Probe 132 illustrates a probe function which uses characteristics of customer objects residing in multiple different categories. Probe 132 may be assigned the task of tracking expenditures of customers who reside in San Francisco. In this case, probe 132 can provide daily, monthly, etc. totals for expenditures of customers in particular geographic regions. Note that the geographic location attribute is not even one that is used in defining the categories Fig. 2 shows an example of user interface 160 for assigning a probe function to a segment, or category.

In Fig. 2, segments are shown at 162 and probes, or "measures," at 164. To associate a probe function with a segment the user selects the segment by clicking on the segment to highlight it, then selects the probe function in a similar manner. Box 166 shows the probes that are presently on the selected segment. Naturally, probes can be added to, or deleted from, segments. They can also be associated with members, joiners or leavers of segments. Probes can be pre-defined expressions or can be created by users/analysts. Additional enhancements to probes can be selected. For example, user interface 160 shows that a "Period Type" can be selected, various "Smoothing" functions (to decrease the effect of transient anomalies on the probe results) can be applied, etc.

In a preferred embodiment, Probes are defined in the Customer Intelligence application. These may be based on measures created in Business Objects Designer A probe is a re-usable aggregated measure associated to a segment.

Typically, a probe's instantaneous values are stored over time. They can be flags, simple or complex expressions. Probes can be associated not only to segment members, but also to joiners and leavers of segments (discussed below). A probe value could be something such as the average purchase interval for a given segment over the last six months. This value in itself is interesting, but becomes highly valuable to marketers, and other business managers, when it is tracked over time. To be able to track how the purchase interval changes based on marketing campaigns that have been run, competitors actions in the marketplace, or just to note its changes over time is very high value intelligence to the marketer. For example, a probe might be used to store monthly revenue from particular segments of the customer base. This information then can be used in any number of reports, such as (1) to view trends (especially when applied to segment joiners and leavers); (2) to forecast future revenues by segment; and (3) to compare segments. Note that these tasks can be performed without having to query against the detailed transaction records except for once a month, when the probe value is calculated. The present invention allows customers to be segmented, and re- segmented (or "re-assigned" to categories) in real time

The present invention tracks the movement of customers between segments (referred to as segment migration). Additionally, predictions as to when customer movement between segments is likely to take place. This can be invaluable to a company to know when a purchaser is moving from a higher-buying category to a lower- buying category. Or when a customer may be about to leave the enterprise altogether. Preventive actions in the form of promotional campaigns, discounts, etc., can be used as incentives to try to keep the customer, or customer group, in the higher-buying segment. A preferred embodiment of the invention refreshes the assignments of customers to segments on a frequency which is appropriate to the task at hand - daily, for example this could be as frequently as daily in some cases.

The present invention uses the concepts of "Joiners" and "Leavers" of segments. Joiners are members who have entered a segment. Leavers are members who have left a segment. By comparing recent joiners with recent leavers, one is able to get early warning about the quantity and direction of change in a segment's population well before such change becomes detectable in the statistics for the larger segment population. There is also great value in understanding why, and how many, segment members are joining and leaving, and to which segments they are going. For example, customers who have just left the high value segment are clearly in need of some sales / marketing action being taken.

Joiner / Leaver comparison analysis (or "migration") also gives companies valuable early warning about the future make up of a given segment. By comparing the values for the same probe applied to the joiners and leavers of a segment provides a basis for quickly understanding and detecting changes in the complexion of segment membership long before the change is reflected in the overall membership.

For example, if the average lifetime value of customers joining the high profit segment is lower than the average for the whole segment, then over time the whole segment value will decrease. Equally, if the leaver average lifetime value is higher than the overall segment and higher than the joiners value then t is clear that over time the overall value of the high profit segment will decrease even more. This gives very high value early warning to business management long before the change to the high profit segment would otherwise be detected. This type of analysis can be simplified further by defining a probe which is calculated based on a calculation between probes on joiner, leaver or the main segment. This is called a 'probe on probe' or 'differential probe' where you are calculating a probe value based on other probes. So for example Probe A = Probe B for segment leavers less probe B for segment joiners. Similarly, sudden changes, such as could be caused by a competitive challenge, can be quickly detected, if one has set up a smart segmentation scheme and defined probes appropriately on the segments. This enables organizations to rapidly detect changes in their competitive position (as might be caused by a competitor targeting their customer base with a special offer) much earlier than would otherwise be possible.

Migration analysis is especially interesting in the area of segmentation. This is where companies study the movement of segment members, understand why they are moving, where they are going, what impact that has on current and future business opportunities. Such movement can also help in understanding potential changes, i.e. detect changes/early warnings of changes within segment membership. Normally, this would be very complex analysis, which is time consuming.

The present invention allows alerts and actions to be triggered based on probe function results, migration conditions and other events. Such ChangeAgents can take several forms; one is where a probe value has reached a particular threshold. Another form is a ChangeAgent that sends a list including segment members, joiners and leavers to another system. One example action can be to alert a marketer, analyst or other person by email when the purchase interval for a segment increases by more than 5 days. This enables marketers to manage much larger numbers of segments since if any business rule is triggered they will immediately be notified that corrective action needs to be taken. Equally, rules can be attached to changes in probe values enabling marketing campaigns (or any other actions as defined in business rules) to be triggered automatically in real time as soon as the segment, or an individual in the segment, changes behavior.

Alerts and rule-based business actions are not peculiar to probes but can also be applied to other business changes as well. Alerts and actions can send email messages to lists of members to raise the appropriate awareness within the company. Equally they can be used to trigger actions. For example, alerting can be by telephone, fax, regular mail, etc. Customers who have left a high value customer segment can automatically be put into a corrective promotional campaign triggered in the call center. The present invention provides yet more mechanisms for performing analysis. "Rules" can be defined which are re-usable conditional expressions, based on probe or other values. Rules can also have associated actions. "Monitors" are the associations of rules to segments with scheduled activation. A "vector" refers to the direction an individual is moving over time, in terms of their behavior pattern. A vector shows the direction and magnitude of change in a customer's behavior with respect to others in the same segment. This can be used, for example: to identify customers whose behavior is changing in a negative direction from the company's standpoint; or to identify customers whose behavior is changing in a negative direction and exceeding an "expectation" derived from the members of the segment(s) to which the identified customers belong; or to identify customers at risk of leaving a high value segment.

By summarizing information on vectors for the membership of a segment, we can also identify the need to refine a segment definition or create a new segment to differentiate high affinity customers from those with lower affinity who perhaps belong in a different category. The important point about vectoring is that it takes into account not only the most recent customer attributes or aggregations thereof, but it also uses information about past behavior to differentiate between normal variation in individual behavior and an incipient change in an individual behavior pattern. As an example, assume that the average inter-purchase interval for a segment is 25 days with an estimated standard deviation of 5 days. A member of this segment has an average inter-purchase interval of 28 days. The recent history of inter-purchase intervals in days for this member looks like: 23, 30,32,24,25,29, 26, 35, 28, 38, 29, 42. To better see the trend for this individual, we can calculate a simple moving average, which looks like: 28.3, 28.6, 27,26, 26.7, 30, 29.7, 33.7,31.7, 36.3.

By calculating a "smooth" of this history, it becomes apparent that the trend for this customer is toward longer inter-purchase intervals. Approximately 95% of members in this segment have inter-purchase intervals in the 15 to 35 day range. So, since the last smoothed value for this customer is 36.3, this indicates a clear "at risk" situation in comparison with others who have similar buying patterns. However, this may be too late to be the optimal point for changing behavior. One way of trying to identify the best point to take action at an individual level would be to find inflection points in the history. A simple approach to ascertaining this might be to define alert conditions such as detecting when an increase in adjacent smoothed values of more than 10%. Using this rule, action would have been initiated when the person's smoothed inter-purchase interval jumped from 26.7 to 30. Alternatively, both the average values for the reference segment and the history of values for the individual may be used in combination.

Portfolio management provides the ability for various stakeholders of a business to track and analyze analytical information of interest to them. This information may be related to their customers, segments, products, sales volumes etc. It has a web interface, which is customizable by all users of the application. The portfolio consists of several categories which have analytical analysis associated to it. E.g. Campaign analysis, Segment Migration etc. The portfolio also handles alerts so a user can view alerts that are directed to them.

Different users of the system have different portfolios which reflect their responsibilities - hence a segment manager will be responsible for managing the performance of a group of segments. But the western regional sales manager will have responsibility for maximizing revenues from all segments in the western region. This enables a wide range of different types of users to optimize their particular areas of the business while all are using common definitions and measures facilitating consistency and coordination in the business.

Fig. 3 shows an example of a web page of a portfolio for a Segment Manager.

The present invention provides Customer portfolio management, which is the ability to manage a portfolio of customers with the goal of managing the corporate- customer relationship so as to increase the profitability of the enterprise. A customer portfolio is represented as a customer segment, possibly with sub-segments. The Portfolio Manager is responsible for controlling all marketing campaigns with the customers in his portfolio, even if such campaigns are initiated elsewhere in the company. Software support for portfolio management may include, but is not limited to the following: associating a segment, and, possibly, a set of sub-segments, with a portfolio manager (possibly in a portfolio management hierarchy); defining business strategy and objectives for the segment and associate probes and reports to monitor progress against these objectives; define campaigns that the segment should belong to in a manner accessible programmatically by campaign management software; defining rules for disambiguating the boundaries of portfolios that would otherwise overlap; tracking costs by portfolio and by programs designed to influence portfolio performance; defining portfolio segment membership differently for targeting than for performance monitoring - the latter excludes recent joiners (not exposed long enough) and includes recent leavers (leaving may be a desirable or undesirable outcome of portfolio management-initiated programs); navigation support in portfolio hierarchies; roll-out of policies (contact management, performance measurement, etc.) from higher level portfolios to sub- portfolios; analysis reports for monitoring, comparing and contrasting sub-portfolio performance; and propagation upwards of alerts on sub-portfolio performance indicators. A model is a mathematical formula used in predicting customer behavior. This model may have been created using some third party data mining tool. It is very important to keep track of the accuracy of models since they inevitably degrade in performance over time. By defining probes that represent the difference between expected performance based on a model and actual performance, we are able to evaluate model performance degradation on a segment by segment basis, and automatically detect when they are no longer useful or valid.

An important aspect of obsolescence detection is that the model may continue to hold for the members of one segment, but not for those of another. An example is when a model is created and used to score the customer base in terms of expected revenue. A segment is defined based on high values of this score. A smoothing probe is defined that measures actual revenue per head generated by the members of the segment. When the value of this probe drops below a user-specified threshold, an alert is generated indicating that the model is no longer accurately reflecting actual revenue generation.

If segment membership and migration is stored over time then it is a simple matter to produce an individual footprint of each customer and how their value has changed over time. This provides marketers with valuable additional information about each customer for targeting campaigns, for tailoring service levels to each customer, and to understanding the effectiveness of promotional campaigns over time (note: to do this optimally requires also tracking each corporate-initiated customer contact that occurred so as to match stimuli to response).

Fig. 4A illustrates a user interface for adjusting segments. The user interface allows segment value ranges to be easily changed while providing visual feedback about the adjustment.

In Fig. 4A, curves 202, 204 and 206 correspond to revenue, profit and cost, respectively. Legend 208 shows this correspondence. In the preferred embodiment, color is also used to identify the different curves and a color bar to the far right of each row in the legend helps to identify the curve. Additional curves can be included and displayed in the legend by using the horizontal scroll bar. For example, a "risk factor" plot or values can be displayed. This plot can be derived from one or more customer values, or characteristics, to show the probability that a customer is going to move out of their current segment. Scales for each of the curves are shown for the vertical axis at the right side of the interface. Column 210 corresponds to revenue, column 212 corresponds to profit and column 214 corresponds to cost. Thus, the plot for revenue, curve 202, ends at about 50k; the plot for profit, curve 204, ends at about -1; and the plot for cost, curve 206 ends at about 457.

The horizontal axis shows the value that can be varied. This is indicated in box 230, as "Age." The position of sliders 220, 222, 224 and 226 indicate the boundaries of each segment. In other words, the sliders define the range of ages for each segment. For example, slider 220 is positioned at 15 years (assuming the horizontal rule's major ticks are 5 years, each) so that the first segment has the criterion of an age range of 0-15 years. The averages for revenue, profit and cost are marked at the center of the defined age range by their corresponding symbols.

Similarly, the sliders 222, 224 and 226 define four other segments with age ranges of 15-52, 52-67, 67-75 and 75-90. At the middle of each segment age range are marked the average for revenue, profit and cost for that age range. Midpoint 240 corresponds with the 15-52 year range segment defined by sliders 220 and 222. Midpoint 242 corresponds to the 52-67 year range segment defined by sliders 222 and 224.

Fig. 4B shows the user interface of Fig. 4A after slider 222 has been moved to the left. This can be done, for example, by allowing a user/analyst to click and drag the slider with a mouse or other user input device coupled to a computer that runs the user interface.

In Fig. 4B, slider 222 has been re-positioned from 52 to 40. This redefines the segments having midpoints 240 and 242 as having age ranges of 15-40 and 40-67, respectively. Note that the markers for these midpoints have shifted, as have the corresponding plots. Note that any number of different values/criteria can be set in this manner. For example, a pull-down menu corresponding to box 230 can be used to select other values such as customer income, # purchases per month, etc.

Thus, the user interface of Figs. 4A and 4B provides a simple and effective method for setting criteria to define segments while also providing instantaneous feedback to the user on the effect of the criteria changes to the segments.

Although the present invention has been described with reference to specific embodiments thereof, these embodiments are but illustrative, and not restrictive, of the present invention. For example, the invention can be applied to track any data where the data is arranged in segments in accordance with the present invention. The scope of the invention is to be determined solely by the appended claims.

Claims

WHAT IS CLAIMED IS:

1. A method for determining changes in a database, the method using a computer system, the method comprising defining two or more categories for items, wherein a category is defined by one or more criteria; associating at least one value with each item; assigning items to the defined categories based on whether an item's associated values meet a category's criteria; detecting when a given item's associated values change; re-assigning the given item to a new category if the changed values meet the new category's criteria; and performing an action in response to the step of re-assigning the given item.

2. The method of claim 1, wherein the database stores changes in a customer's buying habits, wherein an item corresponds to a customer.

3. The method of claim 1, wherein the substep of performing an action includes storing information to identify the given customer and new category.

4. The method of claim 1, wherein the substep of performing an action includes notifying an operator of the computer system that the re-assignment has taken place.

5. The method of claim 1, further comprising using the computer system to predict that the given customer will be re-assigned; and storing information to identify the given customer as a customer that is likely to change categories prior to the step of re-assigning the given customer to the new category.

6. The method of claim 1, wherein the substep of performing an action includes using the computer system to automatically send an email over a network.

7. The method of claim 1, wherein the substep of performing an action includes using the computer system to automatically dial a telephone number and playback a pre-recorded message.

8. The method of claim 1, wherein the substep of performing an action includes using the computer system to automatically send a fax.

9. The method of claim 1, further comprising using the computer to keep a record of the history of re-assignments of the given customer, wherein the record includes each category to which the given customer has been assigned.

10. A method for categorizing customers' in order to assess their buying habits, the method using a computer system, the method comprising defining two or more categories of customers, wherein a category is defined by one or more criteria; electronically accepting information from a new customer; associating at least one value with the new customer based on the electronically accepted information; assigning the new customer to a defined category based on whether the new customer's associated values meet the category's criteria; detecting when a the new customer's associated values change; and re-assigning the new customer to a new category if the changed values meet the new category's criteria

11. The method of claim 10, wherein the electronically accepted information includes the new customer's address.

12. The method of claim 10, wherein the electronically accepted information includes the name of the new customer's place of work.

13. The method of claim 10, wherein the electronically accepted information includes the new customer's telephone number.

14. The method of claim 10, wherein the electronically accepted information includes the new customer's zip code.

15. A method for detecting characteristics of customers' buying habits, the method using a computer system, the method comprising defining two or more categories of customers, wherein a category is defined by one or more criteria; associating at least one value with each customer, wherein the values describe the customer; associating one or more parameters with one or more customers, wherein the parameters describe the customers' buying characteristics. assigning customers to the defined categories based on whether a customer's associated values meet a category's criteria; detecting when a given customer's associated values change; re-assigning the given customer to a new category if the changed values meet the new category's criteria; and creating a relational condition based upon one or more customers' associated parameters; and performing an action when the relational condition is met.

16. The method of claim 15, wherein an associated parameter can be an interval of time between a customer's purchases.

17. The method of claim 15, wherein an associated parameter can be the value of a product purchased by a customer.

18. The method of claim 15, wherein an associated parameter can be an average of multiple parameters.

19. The method of claim 15, wherein an associated parameter is calculated at predetermined time periods.

20. The method of claim 15, wherein a relational condition can include other relational conditions.

21. The method of claim 15, wherein a relational condition can include values and parameters from multiple customers.

21. The method of claim 14, wherein a result can be computed from values and parameters from customers.

22. The method of claim 21, wherein a possible result is the total revenue from substantially all purchases by substantially all customers in a given category.

23. The method of claim 1 stored as instructions in a computer- readable medium.