EP3063712A1

EP3063712A1 - Determining a business strategy

Info

Publication number: EP3063712A1
Application number: EP13896594.2A
Authority: EP
Inventors: Kas Kasravi; Charles S. Clark; Matthew E. FUHRMAN; Timothy HARTFORD; Marie Risov; Terry J. White; Maria N YUZON
Original assignee: Hewlett Packard Enterprise Development LP
Current assignee: Ent Services Development Corp LP
Priority date: 2013-10-30
Filing date: 2013-10-30
Publication date: 2016-09-07
Also published as: WO2015065381A1; EP3063712A4; CN105849752A

Abstract

A technique includes applying at least one econometric model to data representing information about an internal environment of a business organization and an external environment of the business organization; and determining a business strategy associated with a set of ordered steps. Each step is associated with at least one action to be performed.

Description

DETERMINING A BUSINESS STRATEGY BACKGROUND

[0001 ] A business organization may use a high level plan, or strategy, to achieve one or more goals under conditions of uncertainty. Traditionally, the owners, executives and other senior leaders of a business organization develop the strategies for the organization. In developing these strategies, the senior leaders may distill and condense information about the organization and market; meet with clients and employees of the organization; hold information gathering meetings; envision multiple scenarios that may affect candidate strategies that are being considered; and so forth.

BRIEF DESCRIPTION OF THE DRAWINGS

[0002] Figs. 1A, 1 B and 1 C are illustrations of the ordering of steps of a business strategy according to example implementations.

[0003] Fig. 2 is an illustration of a strategy construct according to an example implementation.

[0004] Fig. 3 is an illustration of classifiers used in developing a business strategy construct according to an example implementation.

[0005] Fig. 4 is a schematic diagram of a computer-based system to determine a business strategy according to an example implementation.

[0006] Fig. 5 is an illustration of a workflow used by an optimization engine of the system of Fig. 4 according to an example implementation.

[0007] Fig. 6 is a schematic diagram of a physical machine according to an example implementation.

[0008] Fig. 7 is a flow diagram depicting a technique to determine a business strategy according to an example implementation.

[0009] Fig. 8 illustrates a fact base for a fictional town in which a strategy for a fictional store located in the town is determined according to an example implementation.

[0010] Fig. 9 is an illustration of example strategy constructs for the business strategy for the fictional store according to an example

implementation.

[001 1] Fig. 10 is an illustration of an initial population according to an example implementation.

[0012] Fig. 1 1 illustrates genetic operations according to an example implementation. [0013] Fig. 12 illustrates a first generation derived from the initial population of Fig. 10 according to an example implementation.

[0014] Fig. 13 depicts econometric model-derived values for the first generation of Fig. 12 according to an example implementation.

[0015] Fig. 14 is a generational fitness score trend according to an example implementation.

[0016] Fig. 15 is a generational fitness individual trend according to an example implementation.

DETAILED DESCRIPTION

[0017] One approach to develop a strategy for a business organization (a business partnership; a private corporation; a publically-traded corporation, a limited liability company; a government agency; and so forth) is to involve senior leaders and other personnel of the organization in a significant amount of planning and research in this effort. Due to the presence of the human element, the business organization may face significant challenges in remaining agile and responsive to new opportunities and unexpected problems. This may be especially true for relatively larger business organizations, where the complexity and abundance of information to be processed makes it humanly challenging to sufficiently comprehend all of the factors to optimize strategy in a dynamic and responsive fashion.

[0018] Computer-aided systems and techniques that are disclosed herein, which leverage available internal data about the internal environment of the business organization, as well as external data about the external environment of the organization, for purposes of rapidly developing and optimizing strategies for the organization in an adaptive manner.

[0019] The systems and techniques that are disclosed herein model a given business strategy as an ordered set of steps. As an example, Fig. 1 A depicts a business strategy 100, which includes an ordered sequence of sequential steps 104. For this example, the steps 104 are performed one at a time in a linear fashion.

[0020] As another example, Fig. 1 B depicts a business strategy 1 10 in which example steps 1 12 are performed in linear and parallel fashions. More specifically, the strategy 1 10 includes steps 1 10-1 and 1 10-12 that are performed sequentially in a linear fashion. Steps 1 12-3A and 112-3B are formed in parallel at the same time after the step 1 12-2. Another step 1 12-4 sequentially and linearly follows execution of the steps 1 12-3A and 1 12-3B.

[0021] As another example, Fig. 1 C depicts a business strategy 120 in which steps are performed in three parallel processing paths 130 (paths 130-1 , 130- 2 and 130-3, being depicted as examples in Fig. 1 C). More specifically, for the path 130-1 , steps 132 are performed linearly in an ordered sequential sequence. Likewise, for the paths 130-2 and 130-3, steps 136 and 138, respectively, are also performed in a sequential and linear order.

[0022] It is noted that the paths 130-1 , 130-2 and 130-3 may consume different respective amounts of time to complete. Moreover, as also depicted in Fig. 1 C, dependencies may exist between some steps, such as the depicted dependency between steps 132-4 (of the path 130-1 ) and 138-3 (of the path 130-3).

[0023] Thus, in general, a given business strategy in the context of the application, refers to an ordered sequence of steps, and the steps may be performed sequentially, and in different parallel paths; or a combination thereof. Moreover, some of the steps may be dependent on each other; and different processing paths may consume different corresponding amounts of times for the given strategy. Therefore, many variations are contemplated, which are within the scope of the appended claims.

[0024] The systems and techniques that are disclosed herein further model a given step as being defined by a set of strategy constructs. In general, a strategy construct defines an action to be performed by the business organization. For example, a given action may be an action to buy, sell, hire or relocate. Moreover, successive actions may negate one another. For example, an action to buy may be followed by an action to sell in a later step.

[0025] In general, a given action is performed on or acted upon an entity, which may be a company, stock, or an office of the business organization, as just a few examples. Moreover, the entity from which the action acts may be identified by one or more associated entity parameters, such as (as examples), a salary, a price, a quantity; a location and so forth.

[0026] It is noted that given strategy construct may be defined in many different ways. An example strategy construct 200 is depicted in Fig. 2. Referring to Fig. 2, the depicted strategy construct 200 includes an action 202 that operates on an entity 204. Entity 204, in turn, may be defined or identified by one or multiple entity parameters 205. Moreover, one or multiple resources 208 are involved with performing the action 202 on the entity 204.

[0027] The resources 208 may be, as examples, people 212, assets 214, intellectual assets 216 and business entities 218 involved with performing the action 202 on the entity 204. Moreover, as depicted in Fig. 2, the resource(s) 208 may be associated with one or multiple defining resource parameters 210. Moreover, Fig. 2 also depicts a date 206 on which the action 202 is performed for the strategy construct 200.

[0028] Thus, in general, a business strategy may be modeled, or quantified, as follows:

S = Set of SS (Set of SC(Set of A (E(PI, P2 ... ),R, D))), Eq. 1 where "S" represents a given business strategy; "SS" represents strategy steps; "SC" represents strategy constructs; "A" represents an action; "E" represents an entity; "P1 and P2" and so forth represents entity parameters; "R" represents a resource; and "D" represents a data. Therefore, the strategy S is quantified by a set of strategy steps SS, which include one or more strategy constructs SC(s), as defined by combinations of assets A, entities E, entity parameters P, resources R and dates D.

[0029] It is noted that the definition and classification of terms used in connection with the strategy, strategy steps and strategy constructs discussed herein may vary substantially from industry to industry or from organization to organization within the same industry. Therefore, in accordance with example implementations, a custom classification, or taxonomy, is used for purposes of defining the terms for any instance of the solution.

[0030] More specifically, in accordance with example implementations, custom taxonomy is used to sort relevant data for a given business strategy analysis. In general, the data may be generally sorted into two types: internal data and external data. The internal data pertains to information about the internal dynamics, or internal environment, of the business organization and as its name implies, is gathered from within the organization. For example, the internal data may include data about the organization's products, human resource data, asset data, resource, data, facility location data, financial data, financial target data and other organizational information relevant to the strategies of the organization.

[0031] The external data pertains to information about the external

environment of the business organization, such as the marketplace in which the organization operates, partners of the organization, competitors to the organization and other external sources that affect the performance of the organization. More specifically, the external data may include such data as stock market data, competitive products information data, competitive product pricing data, demographic data, weather data and economic indicator data, just to name a few.

[0032] The custom taxonomy is applied to internal and external data to allow identification and comprehension of the relevant data in both of the internal and external datasets. The custom taxonomy processing may involve collection and preparation of the data, such as unstructured, temporal or geo- location data pre-processing.

[0033] Fig. 3 depicts a classification of data for an example strategy construct 300. Referring to Fig. 3, a custom taxonomy, or classification, for the strategy construct 300 may include such actions 304 as actions to acquire, divest, hire, invest, partner and litigate. Example classifiers for the entity 308 of the strategy construct 300 may include company, stock, building, patent, contract and client classifiers. For a resource 312 of the strategy construct 300, the classifiers may include employee, contractor, hardware, software, service and cache classifiers. [0034] Referring to Fig. 4, in accordance with example implementations, a computer-based system 400 that is depicted in Fig. 4 may be used for purposes of recommending a given business strategy. The system 400 includes a data preparation engine 414, which gathers internal data 404, external data 410 and applies a custom taxonomy 420 for purposes of producing consolidated data 422, i.e., for purposes of sorting through the internal 404 and external 410 data to classify the data and present the corresponding relevant data consolidated 422 for further strategy analysis.

[0035] An optimization engine 430 of the system 400 applies one or multiple econometric models 424 (depending on the implementation) to the consolidated data 422 for purposes of providing strategy steps and associated constructs 436 for a recommended strategy. Econometrics is the application of mathematical and statistical methods to economic and business activities. Econometric models are the specific models used in econometrics, which comprehend statistical and probabilistic variations to predict and optimize business performance. As depicted in Fig. 4, in accordance with example implementations that are described herein, the optimization engine 430 may apply one or multiple econometric models to determine the outcome of candidate strategy steps and the overall value of the steps in a given candidate strategy.

[0036] The evaluated values that result from the application of the

econometric model(s) may be a financial value (a revenue or profit value, as examples) or a non-financial value (student performance or job placement rate, as examples). In general, the value(s) determined by the econometric model(s) 424 are used by the optimization engine 430 for purposes of comparing different candidate strategies and selecting, or recommending, a business strategy based on this comparison, as further described herein.

[0037] In accordance with example implementations, the optimization engine 430 generates candidate business strategies and compares them with each other for purposes of determining and recommending a final strategy (a strategy to maximize profits over a two year period, for example). The optimization engine 430 may use an iterative process, in accordance with example implementations, in which the engine 430 considers candidate business strategies, evaluates the strategies, and makes further refinements, including selecting new candidate strategies, selecting one or more of the existing candidate strategies, selectively merging candidate strategies, and so forth, before ultimately recommending a business strategy.

[0038] The optimization algorithm that is employed by the optimization engine 430 may vary, depending on the particular implementation. As an example, the optimization engine 430 may use a genetic algorithm, in accordance with example implementations.

[0039] As a more specific example, the optimization engine 430 may employ a genetic algorithm that is generally represented by a technique 500 of Fig. 5. Referring to Fig. 5 in conjunction with Fig. 4, the optimization engine 430, pursuant to the technique 500 initially begins with a set 510 of possible strategy constructs and from the constructs generates (block 514) initial, candidate business strategies. From these candidate strategies, the optimization engine 430 generates (block 518) new candidate strategies via genetic operators.

[0040] It is noted that the initial candidate strategies may be randomly or heuristically generated, depending on the particular implementation. As an example of the heuristic generation of the initial candidate strategies, the optimization engine 430 may budget salaries after a hiring data and not before.

[0041 ] In general, the genetic operators may be such operators as crossover and mutation operators, which selectively combine candidate solutions and create new ones. The crossover operator combines the elements of two strategies, for example, and the mutation operator may, as an example, randomly change one or more elements of a given candidate strategy. [0042] The optimization engine 430 ranks (block 522) the candidate strategies using the econometric model(s). In this manner, in accordance with some implementations, upon application of the econometric model(s), the

"improved" candidate strategies or the strategies having respective higher ranks survive and form the next generation of candidate strategies that are considered. In accordance with some implementations, a Darwinian natural selection process guides the optimization approach, where an initial set of candidate strategies is improved over time to ultimately lead to an optimal strategy.

[0043] Thus, based on the ranking 522, the optimization engine 430 selects (block 524) the best currently available candidate solution and decides (decision block 530) whether an optimal strategy has been determined, or found. As an example, the optimization engine 430 may deem that an optimal strategy has been detected when the population remains static across multiple iterations. To avoid being trapped in a local minima, the process may be repeated multiple times. Upon determining that an optimal strategy has been found (decision block 530), the optimization engine 430 reports the recommended strategy, pursuant to block 536.

[0044] In accordance with example implementations, the computer-aided business strategy determination may be performed on a physical machine 600 that is depicted in Fig. 6. The physical machine 600 is an actual machine that is made up of actual hardware 610 and actual machine executable instructions 660, or "software." As examples, the physical machine 600 may be a client, a server, a laptop, a tablet, a desktop, and so forth.

[0045] The hardware 610 of the physical machine 600 may include non- transitory memory storage devices (semiconductor storage devices, optical storages, magnetic-based storage devices, and so forth), which form a memory 614 of the physical machine 600. The hardware 610 may further include one or multiple Central Processing Units (CPUs) 612 and, as a further example, one or multiple network interfaces 616. In general, the CPU(s) 612 may execute instructions that are stored in the memory 614 for purposes of performing one or more of the techniques that are disclosed herein pertaining to determining and recommending a strategy for a business organization.

[0046] As depicted in Fig. 6, the machine executable instructions 660, or software, may include instructions that when executed by the CPU(s) 612 form the optimization engine 430, the data preparation engine 414 and may form the econometric model(s) 424. Moreover, the machine executable instructions 660 when executed by the CPU(s) 612 may form various other software components of the physical machine 600, such as, for example, an operating system 662.

[0047] It is noted that although Fig. 6 depicts the physical machine 600 as being contained within a "box," or rack (as an example), the physical machine 600 may be physically distributed over multiple locations and thus, may be a distributed computing system, in accordance with further implementations. Thus, many implementations are contemplated, which are within the scope of the appended claims.

[0048] Thus, referring to Fig. 7, a technique 700 in accordance with example implementations includes processing (block 704) candidate business strategies to evaluate candidate business strategies based at least in part on at least one or multiple econometric models, internal data representing information about the internal environment of the business organization and external data representing information about the external environment of the business organization. Each candidate business strategy is associated with a set of ordered steps, and each step is associated with at least one action to be performed. The technique 700 includes recommending (block 708) a business strategy based at least in part on the evaluation.

[0049] As a more specific example, the computer system 400 of Fig. 4 may be used to determine a business strategy for a fictitious general store (called "Gstore" herein) that is located in a small town (called "Stown" herein). For this example, the computer system 400 determines a strategy to maximize the revenue of Gstore. A fact base (described below) that is available to the computer system 400 provides the parameters of this problem. In this manner, the computer system 400 acquires internal 404 and external 410 data representing this fact base, such as (as examples) information about the population of Stown, the needs of Stown, the employee costs, the local competition and other factors.

[0050] The optimization engine 430 applies a genetic algorithm and uses an econometrics model 424 to the fact base to assess strategies defined in a genetic population for this example. Moreover, for this example, the econometrics model is based on the transactions in a single month.

[0051] Fig. 8 is an illustration of the fact base about Stown. In this regard, Fig. 8 depicts two tables 800A and 800B of factors 810 in the fact base and their corresponding values 814. As can be seen from Fig. 8, some of the factors 810 have associated dollar values, whereas other factors, have corresponding percentages. In this manner, the supply for books, for example, is one factor 810 that has a value of $5000, whereas another factor 810 pertains to inventory costs, which is percentage of the total costs.

[0052] Fig. 9 depicts example strategy constructs 900 for the Gstore for this example. Each construct 900 for this example includes an action 910 to be performed on an entity 914, which is characterized by one or multiple options, or entity parameters 918. For example, a given construct 900 pertains to an action of opening the store (910-1 ), which operates on a store entity 914-1 and is associated with parameters 918-1 of the possible opening times. As another example, a given construct 900 pertains to an employee-based action 910-2, which involves hiring employees (910-2 and 914-2) and is involved with parameters 918-2 specifying the possible number of employees to be hired.

[0053] Other constructs 900 may involve actions 910-3 to sell. As illustrated in Fig. 9, a given sell action 910-3 may operate on various entities 914. For example, a given construct 900 may involve Gstore selling (910-4) baked goods 914-3 and may be associated with possible dollar values 918-3 for the resulting revenue. As another example, another construct 900 may involve Gstore the selling (910-3) may involve selling home delivery items 914-4 and may involve a binary-based decision of yes or no, as depicted by parameters 918-4.

[0054] Fig. 10 depicts an example initial population 1000 of individuals 910, or candidate strategies, which are considered by the optimization engine 430 to begin an iterative process that the optimization engine 430 uses to determine the recommended business strategy for purposes of maximizing the net revenue of Gstore. It is assumed that reasonable margins and profits are a function of revenues.

[0055] The optimization engine 430 uses a genetic algorithm for this example to generate candidate strategies (also called "individuals" herein in the context of the genetic algorithm) and assess the net revenues of these candidate strategies with an econometrics model 424. In particular, the optimization engine 430 determines a consequence of each strategy step, determines the overall outcome of each candidate strategy and determines a net revenue value for each candidate strategy. The net revenue for this example is the measure of the "fitness" of each candidate strategy, or individual.

[0056] For this example, the net revenue is used as the natural selection factor in the genetic process. As discussed herein, although the initial population 1000 defines an initial set of six candidate strategies, or individuals, for this example, the optimization engine 430 may derive additional candidate strategies using mutation and crossover operations to produce additional candidate strategies for evaluation.

[0057] For the econometrics model 424 for this example, the optimization engine 430 determines the total sales, based on the available goods for sale and the market demand; determines additional sales due to the advertising budget; determines additional sales due to home delivery option; determines preliminary gross revenues by adding up the revenues; determines actual gross sales based on the opening hours (by percentage); determines the costs of the employees; and determines inventory costs for unsold goods. The econometrics model 424 further causes the optimization engine 430 to determine net revenues by subtracting the employee costs, the advertising budget and unsold goods from the actual gross revenues.

[0058] It is noted that the econometric model 424 is relatively simple for this example for illustrative purposes. In general, the econometric model 424 may substantially more complex in accordance with further implementations.

[0059] Moreover, for this example, the genetic algorithm causes the optimization engine 430 to maintain a population of six individuals, or candidate strategies, at the beginning of each generation (or "iteration"). The individual chromosomes of the genetic algorithm are the strategy constructs, and the values are selected from the options list. Each individual has a unique number (see Fig. 10) for identification purposes.

[0060] In accordance with an example implementation, for this example, the genetic algorithm uses the following optimization process. First, constructs and values are randomly assigned to each member of the population to produce the initial population 1000 of Fig. 10. The six individuals are then randomly paired into three pairs; and three new offsprings are created by randomly using genetic crossover and mutation operators.

[0061] The optimization engine 430 applies the econometrics model 424 to each of the nine individuals, so that the individuals may be ranked according to respective net revenues. The genetic algorithm then causes the optimization engine 430 to select the six highest ranking individuals and discard the remaining individuals to derive six new individuals (or "candidate strategies").

[0062] The above-described process then repeats for another generation by randomly pairing the six individuals into three pairs and proceeding as described above. This process continues through one or multiple iterations until the populations rankings stabilize. After this stabilization occurs, the optimization engine 430 selects the highest ranking individual as the recommended strategy for Qstore.

[0063] As a more specific example, Fig. 1 1 depicts an illustration 1 100 of individuals 1010-1 and 1010-3 of the initial population 1000 of Fig. 10 by randomly pairing the individuals 1010-1 and 1010-3 together to produce an offspring 1010-7. As illustrated in Fig. 11 , this random pairing uses mutations 1 104 in which parameters of neither individual 1010-1 nor 1010-3 are selected and crossover operations 1 108 in which the parameters for one of the individuals 1010-1 and 1010-3 are selected over the other.

[0064] Fig. 12 depicts a compilation 1200 of nine individuals 1010 produced by the pairing of individuals one and three, two and six, and four and five from the initial population 1000. Fig. 13 depicts a corresponding econometrics analysis 1300 of the nine individuals 1010 of Fig. 12. As can be seen from Fig. 13 in conjunction with Fig. 12, for the first generation (i.e., the first iteration), the individual 1010-3 is the "fittest" individual having a net revenue of $4,542. The individual 1010-3 is thus chosen, along with the next five highest ranked individuals 1010 to begin the next iteration, as described above.

[0065] For this example, the score of the fittest individual stabilizes by the ninth iteration, or generation. Fig. 14 depicts a generational fitness score trend 1400 showing a fittest individual score 1404 for each generation and a corresponding average population score 1408 for each generation. Fig. 15 depicts a generational fitness individual trend 1500 showing a fittest individual 1504, second fittest individual 1508 and third fittest individual 1512 for each generation. By generation six, for this example, the fittest individual is number 22.

[0066] While a limited number of examples have been disclosed herein, those skilled in the art, having the benefit of this disclosure, will appreciate numerous modifications and variations therefrom. It is intended that the appended claims cover all such modifications and variations.

Claims

What is claimed is: 1 A method comprising:

processing candidate business strategies on a processor-based machine to evaluate the candidate business strategies based at least in part on at least one econometric model, internal data representing information about an internal environment of a business organization and external data representing information about an external environment of the business organization, wherein each candidate business strategy is associated with a set of ordered steps and each step being associated with at least one action to be performed; and

based at least in part on the evaluation, using the processor-based machine to determine a business strategy.

2. The method of claim 1 , wherein processing the candidate business strategies comprises using the at least one econometric model to determine a financial outcome for each of the candidate strategies.

3. The method of claim 1 , further comprising:

based at least in part on the evaluation, selectively selecting and merging the candidate business strategies and repeating using the at least one econometric model in at least one more processing iteration by the processor-based machine to determine an optimum business strategy.

4. The method of claim 1 , wherein processing the candidate business strategies on the processor-based machine and determining the business strategy using the processor-based machine comprises applying a genetic optimization algorithm.

5. The method of claim 1 , further comprising:

using the processor-based machine to classify the internal data and the external data into categories relevant for the candidate business strategies.

6. The method of claim 1 , wherein each step is associated with a set of constructs, and each construct being defined by a set of actions performed up entities using resources.

7. The method of claim 1 , wherein processing the candidate business strategies comprises heuristically sequencing the steps of at least one of the candidate strategies.

8. An article comprising a non-transitory computer readable storage medium to store instructions that when executed by a computer cause the computer to:

apply a least one econometric model to data representing information about an internal environment of a business organization and an external environment of the business organization; and

determine a business strategy, the determining comprising optimizing a set of ordered steps associated with the business strategy, wherein each step is associated with at least one action to be performed.

9. The article of claim 8, the storage medium to store instructions that when executed by the computer cause the computer to use the at least one econometric model to evaluate candidate sets of ordered steps.

10. The article of claim 9, the storage medium to store instructions that when executed by the computer cause the computer to, based at least in part on the evaluation, selectively select and merge the candidate sets of ordered steps.

11. The article of claim 8, the storage medium to store instructions that when executed by the computer cause the computer to apply a genetic optimization algorithm to determine the business strategy.

12. An apparatus comprising:

a classifier comprising a processor to classify internal data representing information about an interna! environment of a business organization and classify external data representing information about an external environment of the business organization; and

an optimizer comprising a processor to:

evaluate candidate business strategies based at least in part on at least one econometric model and the classified internal and external data; and

based at least in part on the evaluation, determine a business strategy.

13. The apparatus of claim 12, wherein the optimizer applies genetic selection using the at least one econometric model as a fitness function to determine an optimum business strategy.

14. The apparatus of claim 12, wherein each step is associated with a set of constructs, and each construct being defined by a set of actions performed on entities using resources.

15. The apparatus of claim 12, wherein the optimizer randomly sequences the steps of at least one candidate business strategy of the candidate business strategies.