S-Cube Learning Package    Service Network Performance Analysis:Performance Analysis and Strategic Interactions in Service...
Learning Package Categorization                                          S-Cube                          Business Process ...
Learning Package Overview Problem Description Service Network Performance Analysis Performance Analysis of Competing Se...
Background: Service Networks Economic globalization combined with rapid technological progress has  led most service comp...
Background: Service Networks Service networks consist of interdependent companies that  use social and technical resource...
Problem Description A central problem in service network design is to analyze  participants’ behavior and optimize their ...
Study “what-if” scenarios Such as …    – What is the impact of setting optimal – for one participant - prices on      the...
The Dynamics of Emergent ServiceNetworks Capturing the dynamics of the SNs (emerging markets and business  models, networ...
Motivating Scenario: Car Repair ServiceNetwork Focus on the end-to-end process “Repair & Order” Car repair service netwo...
Motivating Scenario: Repair ServiceDescription The technicians report the car service requirements that may include  repl...
The Repair Service System ModelLearning Package: Service Networks Performance Analysis   © S-Cube - 11
Motivating Scenario: Repair Service Based on the service network model, we need to investigate  network profitability and...
Learning Package Overview Problem Description Service Network Performance Analysis Performance Analysis of Competing Se...
Overall Contribution Introduce a service performance analytics model in support of  strategic analysis of service network...
Background: System DynamicsApproach System Dynamics approach analyzes the behavior of a  complex system over time. Syste...
Definition of Terms Service Network: a set B of participants connected through  transfer of offerings that delivers value...
Customer Satisfaction (1/2) Denotes the willingness of end customers to buy the services  offered by the network Influen...
Customer Satisfaction (2/2) Without loss of generality, the following formula is used to quantify theabove measures.     ...
Participants’ Value (1/2) Each economic entity within a service network has value  when it satisfies the entity’s needs a...
Participants’ Value (2/2)                                                                  (3) ERij(TN) and ECij(TN): exp...
The Mechanism for Value Calculation The mechanism for the calculation is divided in various  hierarchical levels. Graphi...
Visualization of 1st & 2nd Level                                   TOTAL VALUE OF                                  SERVICE...
Visualization of 3rd Level                       repair materials              Repairing Materials              ordering  ...
Simulations on the Car Repair ServiceSystem (1/2) Economic entities analyzed: technicians, parts manager, and  help desk ...
Simulations on the Car Repair ServiceSystem (2/2) Participant’s value is calculated as a function of price and  time    –...
Simulation Results – Case A A. Value Optimization in Basic and Transformed Network                                       ...
Simulation Results: Observations(A’ cont. ) The dealers’ optimal mean repair price in the basic service network is lower ...
Simulation Results: Observations(A’ cont. ) Results reveal that OEM’s value in the transformed network is not higher  tha...
Simulation Results – Case B B. Sensitivity analysis of the Mean Repair Price                                         1400...
Simulations Results – Case C C. The Impact of New Entries Investigate the impact of the change of dealers letting the pr...
Simulation Results – Case D D. Participants’ Equilibrium Strategies Investigate strategic interactions and determine equ...
Learning Package Overview Problem Description Service Network Performance Analysis Performance Analysis of Competing Se...
Contribution Evaluating service systems’ performance and analyzes the  strategic interactions between competing service n...
Competing Service Networks Service Network A and Service Network B offering the same  service s    – Service Network B is...
Competing Service Networks                                                          ←The competing networks               ...
Competing Service Networks: Approach First, we seek to derive equilibrium strategies for the  providers, given that all o...
Background: Solution Approach Formalization combines concepts from service-oriented economy and game  theoretic tools    ...
Solution Approach Sequential game rather than solving the problem in an one-shot game     – At each time period, each OEM...
Scenario #1 Information revealed for each OEM in each time slot is the  number of ordered parts of its opponent for the p...
Scenario #1 Cont. & Results The price for OEM i is determined as following:    – Value is compared in two consecutive tim...
Scenario #1: Results OEM A’s value decreases and OEM B’s value increases up to month 4,  where equilibrium is reached.   ...
Scenario #2 Information revealed for each OEM in each time slot is the  number of ordered parts of its opponent for the p...
Scenario #2 Cont. & Results In this case, we change the order of the criteria placing the  comparison of the number of or...
Scenario #2: Results The second scenario in which the most recently set up network imitates its  opponent’s decisions in ...
Scenario #3 Information revealed for each OEM in each time slot is the  number of ordered parts of its opponent for the p...
Scenario #3: Results OEM B changes his strategy to encounter delivery time in addition to price. This  has as a result an...
Scenario #4 We study the behavior of the weaker competitor (network B)  given that it first observes opponent’s behavior....
Scenario #4: Results The leader (network A) faces the smaller loss in value (and consequently  in the number of customers...
Learning Package Overview Problem Description Service Network Performance Analysis Performance Analysis of Competing Se...
Discussion Existing approaches and methodologies regarding the measurement of  service networks’ performance    – mainly ...
Summary The work presented contributes towards the evaluation of  performance of service systems    – By estimating value...
Further ReadingCaswell, N. S., Nikolaou, C., Sairamesh, J., Bitsaki, M., Koutras, G. D., and Iacovidis, G. Estimating valu...
Acknowledgements             The research leading to these results has             received funding from the European     ...
Upcoming SlideShare
Loading in...5
×

S-CUBE LP: Performance Analysis and Strategic Interactions in Service Networks

337

Published on

Published in: Technology, Business
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total Views
337
On Slideshare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
3
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

S-CUBE LP: Performance Analysis and Strategic Interactions in Service Networks

  1. 1. S-Cube Learning Package Service Network Performance Analysis:Performance Analysis and Strategic Interactions in Service Networks University of Crete (UoC) Marina Bitsaki, Mariana Karmazi, Christos Nikolaou www.s-cube-network.eu
  2. 2. Learning Package Categorization S-Cube Business Process Management (Performance) Analysis and Design of Service Networks Performance Analysis and Strategic Interactions in Service NetworksLearning Package: Service Networks Performance Analysis © S-Cube - 2
  3. 3. Learning Package Overview Problem Description Service Network Performance Analysis Performance Analysis of Competing Service Networks Discussion and SummaryLearning Package: Service Networks Performance Analysis © S-Cube - 3
  4. 4. Background: Service Networks Economic globalization combined with rapid technological progress has led most service companies to coordinate their corporate operations in a world of interactions and partnerships. Service companies organize their fundamental structure into service networks capitalizing on the advantage of collaboration. Service providers prefer to team up and combine their services to create new and innovative ones in the market place. – These new combined ones form a service system or a service network – There may be several service networks within a service ecosystem that offer comparable or replaceable services, and which therefore compete for market share – There may be service providers that belong to more than one service networks, possibly competing ones.Learning Package: Service Networks Performance Analysis © S-Cube - 4
  5. 5. Background: Service Networks Service networks consist of interdependent companies that use social and technical resources and cooperate with each other to create value and improve their competitive position. The concept of service networks includes a network of relationships between companies where flexibility, quality, cost effectiveness and competitiveness are better achieved than in a single company. The interaction between two participants of a network does not depend solely on their direct connection, but on the impact other relationships within the network have on them. In addition, the network can strengthen business innovation as different parts contribute services to an overall value proposition combining their know-how and core capabilities.Learning Package: Service Networks Performance Analysis © S-Cube - 5
  6. 6. Problem Description A central problem in service network design is to analyze participants’ behavior and optimize their value Predict the performance of the service network (through simulations) – Predictions about the future of the service network in order to increase its adaptability to the changes of the environment and enable network participants to determine the most profitable co-operations and attract new ones. We show that the interactions among the participants of a network force them to reach equilibrium otherwise the network will collapse – Successful predictions of the future behavior of service networks help analysts improve service network’s functionality Study the impact of strategic changes on the performance both at the level of the network as well as its participantsLearning Package: Service Networks Performance Analysis © S-Cube - 6
  7. 7. Study “what-if” scenarios Such as … – What is the impact of setting optimal – for one participant - prices on the performance of the other participants as well as the entire network? – What is the impact on the performance if a new participant suddenly enters the service network? – Are there any equilibrium strategies among the participants that eliminate their conflicts of interests? What about emergent service networks? – The dynamic environment in which service networks emerge, and especially on connectivity and profitable cooperation that play an important role in value creation, must be studied.Learning Package: Service Networks Performance Analysis © S-Cube - 7
  8. 8. The Dynamics of Emergent ServiceNetworks Capturing the dynamics of the SNs (emerging markets and business models, networks, consumers, regulators, etc.) is to a large extent an open problem today. Special interest in analyzing and predicting how existing market interactions can give rise to emergent SNs, with possibly business sector specific patterns: – develop models of value creation and destruction in networks and of value optimization. – use game theory to understand behavior, evolution of competing SNs – show how the drive for value optimization and the constant change of the environment(new businesses enter the ecosystem and others die, innovative services and products appear, etc.) push the network to waves of restructuring in order to remain competitive. – develop criteria and indicators of change that can serve as input for business process and alliances redesign and realignment.Learning Package: Service Networks Performance Analysis © S-Cube - 8
  9. 9. Motivating Scenario: Car Repair ServiceNetwork Focus on the end-to-end process “Repair & Order” Car repair service network linking four types of participants: – an Original Equipment Manufacturer (OEM) (e.g., Volvo), - Content Packager - Help Desk Experts – Dealers (with repair facilities), - Technicians - Parts Manager – Suppliers (Supply Chain Suppliers & Third-party Suppliers) – CustomersLearning Package: Service Networks Performance Analysis © S-Cube - 9
  10. 10. Motivating Scenario: Repair ServiceDescription The technicians report the car service requirements that may include replacing teardowns, warranty replacements and collision repairs. – Once authorized the automotive technician will scrutinize failure symptoms, detect faulty parts, order parts and perform the repair. Ordering parts is a complex process that involves asking advice from expert technicians from the OEM, including acquiring information about parts under warranty, and getting approval from the dealer’s part manager. – The part manager then checks local inventory for the required part, and if necessary checks the stock at the OEM or supplier stocks, and eventually places an order. The part manager may either use third-party suppliers or suppliers from certified supply-chain suppliers - The quality of the OEM parts, catalogues, and OEM support services influences how many OEM parts will be ordered and used for a car repair and how many parts will be used from Third Party Suppliers (TPS), and how many customers will go to OEM dealers or to TPS dealers. OEM obtains parts from certified supply-chain suppliers (SCS)Learning Package: Service Networks Performance Analysis © S-Cube - 10
  11. 11. The Repair Service System ModelLearning Package: Service Networks Performance Analysis © S-Cube - 11
  12. 12. Motivating Scenario: Repair Service Based on the service network model, we need to investigate network profitability and gives answers to the following – Determine the conditions under which it is profitable for a firm to participate in the network and identify the factors that influence its value. – Identify key stone participants (participants that create the most value for the network). – Determine participants’ optimal strategic decisions (cooperating with someone or not, joining the network or not, etc.)Learning Package: Service Networks Performance Analysis © S-Cube - 12
  13. 13. Learning Package Overview Problem Description Service Network Performance Analysis Performance Analysis of Competing Service Networks Discussion and SummaryLearning Package: Service Networks Performance Analysis © S-Cube - 13
  14. 14. Overall Contribution Introduce a service performance analytics model in support of strategic analysis of service network changes and improvements. – Study “what-if” scenarios Propose a simulation model to evaluate the long-term impact of changes to resources and predict the performance of service networks. Contributing to the definition and solution of value optimization problems with respect to service prices Simulations have been performed taking into consideration the basic and transformed service system model as presented in [1] and forms the basis of the proposed methodology.Learning Package: Service Networks Performance Analysis © S-Cube - 14
  15. 15. Background: System DynamicsApproach System Dynamics approach analyzes the behavior of a complex system over time. System dynamics tools allow modelers to succinctly depict complex (service) networks, visualizing processes as behavior-over-time graphs, stock/flow maps, and causal loop diagrams. These models can be tested and explored with computer simulation providing for example better understanding of the impact of policy changes (e.g., through animation of (service) systems) and facilities for sensitivity analysis. Examples of such tools : iThink, Vensim, and PowerSim.Learning Package: Service Networks Performance Analysis © S-Cube - 15
  16. 16. Definition of Terms Service Network: a set B of participants connected through transfer of offerings that delivers value to them Offerings are treated as services composed of participants’ interactions and co-operations to provide a final service to a set C of end customers. pij: price partipant i charges participant j for offering its services and rij : denote the service time of the interaction between participants i and j. Corner-stone for calculating value: customer satisfaction – Customer satisfaction is affected by the time and priceLearning Package: Service Networks Performance Analysis © S-Cube - 16
  17. 17. Customer Satisfaction (1/2) Denotes the willingness of end customers to buy the services offered by the network Influences the increase or decrease of new entries SATij(TN): satisfaction of participant j for consuming services from participant i at the end of the time interval [TN-1, TN] – As a variation of the American Customer Satisfaction Index, operationalized through three measures: - q1 is an overall rating of satisfaction, - q2 is the degree to which performance falls short of or exceeds expectations, - q3 is a rating of performance relative to the customer’s ideal good or service in the category.Learning Package: Service Networks Performance Analysis © S-Cube - 17
  18. 18. Customer Satisfaction (2/2) Without loss of generality, the following formula is used to quantify theabove measures. (1)  [x]: the integer part of x  βks ,γks: parameters determining the effect of price pij and time tij respectively on qk. Satisfaction Index Function (2) where wk are weights that indicate the importance of each measure qk.Learning Package: Service Networks Performance Analysis © S-Cube - 18
  19. 19. Participants’ Value (1/2) Each economic entity within a service network has value when it satisfies the entity’s needs and its acquisition has positive tradeoff between benefits and sacrifices – Gains and losses captured by the relationships between participants in order to compute value Expected Profit definition: – Epij(TN) of participant i due to its interaction with participant j to be the expected value of participant i in the next time interval [TN, TN+1] increased (or decreased) by the percentage change of the expected satisfaction ESATij(TN) in the next time intervalLearning Package: Service Networks Performance Analysis © S-Cube - 19
  20. 20. Participants’ Value (2/2) (3) ERij(TN) and ECij(TN): expected revenues and costs respectively for the next time interval. Thus, the value Vi(TN)of participant i at the end of time interval [TN-1,TN] is the sum of its realized profits(revenues minus costs) and the expected profits that come from its relationships with all other participants. Total value of the network is the sum of value of each participantLearning Package: Service Networks Performance Analysis © S-Cube - 20
  21. 21. The Mechanism for Value Calculation The mechanism for the calculation is divided in various hierarchical levels. Graphical representations are generated from iThink tool – Each nodes represents a module that calculates the value of a participant – Arrows represents the dependencies between modules - Green arrows represent the impact a module has on another module – A module encloses a sub-system that calculates the value of the module = 2nd hierarchical level - Complex variables inside the module are presented as modules tooLearning Package: Service Networks Performance Analysis © S-Cube - 21
  22. 22. Visualization of 1st & 2nd Level TOTAL VALUE OF SERVICE NETWORK DEALERS REVENUES DEALERS REVENUES.REVENUES OF THE DEALERDEALERS VALUE OEMs VALUE DEALERS VALUE + DEALERS EXPECTED DEALERS EXPECTED PROFITS.EXPECTED PROFITS - PROFITS OF THE DEALER DEALERS COSTS.COSTS OF THE SUPPLIERS VALUE DEALER END CUSTOMERS VALUE DEALERS COSTS 1st hierarchical level of value mechanism 2nd hierarchical level – dealer’s value Learning Package: Service Networks Performance Analysis © S-Cube - 22
  23. 23. Visualization of 3rd Level repair materials Repairing Materials ordering on order receiving materials Inventory selling materials Ordering supplier Logic lead time Consumer Demand Logic MEAN REPAIR TIME rate of service CUSTOMER DEALERS COST requests SATISFACTION poisson s numbers of parts DEALERS ordered per month FIXED COST.DEALERS FIXED COST average number of parts TOTAL COST OF per repair PURCHASES DEALERS FIXED COST mean price per repair TOTAL COST OF average OEM price PURCHASES.COST total purchases Pd per part Po 3rd hierarchical level – dealer’s costLearning Package: Service Networks Performance Analysis © S-Cube - 23
  24. 24. Simulations on the Car Repair ServiceSystem (1/2) Economic entities analyzed: technicians, parts manager, and help desk – each of which is offering its labor as a service to the service system Rates of offerings and payment flows are measured per month over a period of about 30 months Service requests, s, are strongly affected by end customer satisfaction, since satisfied customers attract new customers to enter the network. Without loss of generality, service requests are produced by the Poisson distribution with mean es being the output of the function: (4) – where a2>2a1>0 so that es is an increasing function of SAT in the range [0,1].Learning Package: Service Networks Performance Analysis © S-Cube - 24
  25. 25. Simulations on the Car Repair ServiceSystem (2/2) Participant’s value is calculated as a function of price and time – Optimal level is determined with respect to price Investigating Nash equilibrium strategies between OEM and the dealer – OEM’s and dealers’ strategies are defined according to mean profit rates a and b of selling parts and repairing services. – ps, p0, pd be the mean prices set by the suppliers, OEM and dealers respectively for offering their services (5) (6) Assumption: Rest network participants (apart from the OEM and the dealer) do not affect the decisionsLearning Package: Service Networks Performance Analysis © S-Cube - 25
  26. 26. Simulation Results – Case A A. Value Optimization in Basic and Transformed Network ← a. Comparison between basic and transformed Network. P* is the mean repair price that maximizes the dealers’ and OEM’s value ← b. OEM’s value in basic (1) and transformed (2) network at common mean repair pricesLearning Package: Service Networks Performance Analysis © S-Cube - 26
  27. 27. Simulation Results: Observations(A’ cont. ) The dealers’ optimal mean repair price in the basic service network is lower than in the transformed service network, since the mean repair time (that affects value) decreases, so the dealer charges his customers less. Consequently, the dealer is forced to increase the mean repair price in order to increase its revenues. Nevertheless, at the optimal mean repair price, dealers’ value is less in the transformed network since the customer satisfaction has decreased as well (higher charges). OEM’s value is much higher in the transformed network than in the basic one. This is explained by the fact that the mean repair time decreases and the customers are more satisfied (at OEM’s optimal mean repair price). In addition, OEM in the transformed network has much lower mailing and labor costs. In both networks OEM’s value at dealer’s optimal mean repair price (111 and 116 respectively) is very low compared to OEM’s value at his optimal mean repair price. This means that OEM will never be satisfied to offer its services at prices that reach dealer’s optimal level. Dealers’ value at OEM’s optimal mean repair price is higher in the transformed network, since OEM’s optimal price is lower (218).Learning Package: Service Networks Performance Analysis © S-Cube - 27
  28. 28. Simulation Results: Observations(A’ cont. ) Results reveal that OEM’s value in the transformed network is not higher than that of the basic network from the 1st month. – It dominates after 10-12 months, when both networks offer their final services at their optimal mean repair price (Fig b.). – When both networks offer their services at common prices in the range of 80 to 350, the transformed network dominates the basic network at month 8 to 17. The total value of the transformed network (32.190.040.300) is maximized at mean repair price 216 and is higher than that of the basic network (28.593.400.000) which is maximized at mean repair price 223. – End customers are more satisfied and OEM (the keystone participant) has managed to cut costs at a great extend in the transformed network. – Optimal mean repair price for both service networks is very close to the optimal mean repair price of OEM, since OEM contributes the largest part of the total value of the network.Learning Package: Service Networks Performance Analysis © S-Cube - 28
  29. 29. Simulation Results – Case B B. Sensitivity analysis of the Mean Repair Price 14000000 ← Dealers’ difference of value in basic and Difference of value in two networks 13500000 transformed network 13000000 12500000 12000000 11500000 111 112 113 114 115 116 117 118 119 Price As the mean repair price increases, the difference between the dealers’ value in the basic network and that in the transformed network is smaller. This is justified by the fact that, although the service requests decrease, the mean repair price increases resulting in a decrease of the total valueLearning Package: Service Networks Performance Analysis © S-Cube - 29
  30. 30. Simulations Results – Case C C. The Impact of New Entries Investigate the impact of the change of dealers letting the price unchanged so that the end customers are motivated to remain in the network. – Competitive network: the network formed of a new group of dealers. – Calculation based on the optimal mean repair price for the transformed network: 216 Results: – Dealers’ value (31.527.812) is lower in the competitive network compared to the transformed one (35.481.031), since the new dealers’ cost is higher due to the complementarities they offer. – OEM’s value increases (from 29.793.000.000 to 31.713.504.020) due to the increase of the service requests. – The total value of the network increases from 32.190.040.300 to 32.792.529.000. Observation: a change in the network that improves its performance may affect positively some participants and negatively others. Naturally, dissatisfied participants abandon the network causing side effects to the others.Learning Package: Service Networks Performance Analysis © S-Cube - 30
  31. 31. Simulation Results – Case D D. Participants’ Equilibrium Strategies Investigate strategic interactions and determine equilibrium strategies of OEM and dealers. – 1st Experiment: OEM’s optimal profit rate calculation at a given profit rate for the dealer. - Simulations show that when the dealer increases its profit rate (e.g., from 6% to 10%), OEM’s optimal choice is to decrease its optimal profit rate (from 24% to 21%). Conversely, if OEM increases its profit rates (e.g., from 14% to 21%), the dealer optimally decreases its profit rate (from 15% to 10%). – 2nd Experiment: Calculates a set of equilibrium strategies for OEM and the dealer: - at dealer’s profit rate of 10% the optimal OEM’s profit rate equals 21%. - Conversely, at OEM’s profit rate of 21% the optimal dealer’s profit rate equals 10%.Learning Package: Service Networks Performance Analysis © S-Cube - 31
  32. 32. Learning Package Overview Problem Description Service Network Performance Analysis Performance Analysis of Competing Service Networks Discussion and SummaryLearning Package: Service Networks Performance Analysis © S-Cube - 32
  33. 33. Contribution Evaluating service systems’ performance and analyzes the strategic interactions between competing service networks. The initial model is extended to account for competitors of the existing service network. Conditions are properly defined to account how two competing networks co-exist, the types of information the networks share and various strategies chosen by the competing participants. We simulate those strategies and observe the behavior of our system over time in terms of profits and market shareLearning Package: Service Networks Performance Analysis © S-Cube - 33
  34. 34. Competing Service Networks Service Network A and Service Network B offering the same service s – Service Network B is a new entry in the market, while Service Network A has already its own customers. Existing Customers may move from A to B and vice-versa according to their satisfaction New customers may choose a provider based on service price, service time and provider’s reputation. The two networks might share a common participant (e.g., same supplier)Learning Package: Service Networks Performance Analysis © S-Cube - 34
  35. 35. Competing Service Networks ←The competing networks interacting through their customers Customers may abandon a network if they are not satisfied from dealers’ services. Even though OEMs are not directly connected to customers, their actions affect dealers’ decisions which in turn affect customer satisfaction resulting in the restructuring of market share Investigate strategic behavior of the two OEMsLearning Package: Service Networks Performance Analysis © S-Cube - 35
  36. 36. Competing Service Networks: Approach First, we seek to derive equilibrium strategies for the providers, given that all other participants of their own networks have revealed their actions. Second, we investigate the evolution of the two networks over time in terms of survivability and dominance in the market. Simulations were conducted based on four scenarios- strategies which will be presented next – Simulations examine whether these strategies reach an equilibrium and form initial predictions for the evolution of networks over time in terms of dominance and survivability. – The time of all simulations is measured in months and specifically for 60 months – The initial price is set for OEM A to be higher (pA(1) = 200) than that of OEM B (pB(1) = 190) since the new competitor needs to provide incentives to the forthcoming customers.Learning Package: Service Networks Performance Analysis © S-Cube - 36
  37. 37. Background: Solution Approach Formalization combines concepts from service-oriented economy and game theoretic tools – two-player game: service providers make choices for prices and service times who compete for market share, while assuming only one type of customers A pure strategy si for provider i (i = A, B) is defines as the pair si = (pi, ti), where pi is the price for the service k and ti is the service time needed to provide service k. Let S be the set of all pure strategies for each provider. Nash equilibrium is a pair of strategies, one for each provider, in which each provider is assumed to know the equilibrium strategies of his opponent, and no one benefits by changing only his own strategy unilaterally In our game, s* = (s1*, s2*)is a Nash equilibrium in pure strategies if for each s1, s2 ε S, value satisfies the inequalities: (7) (8)Learning Package: Service Networks Performance Analysis © S-Cube - 37
  38. 38. Solution Approach Sequential game rather than solving the problem in an one-shot game – At each time period, each OEM exploits information revealed by his opponent and makes decisions on which prices to charge dealers and which delivery times to complete dealers’ orders for parts for the next time period. We define the following: – vi(t) be the value at time slot t, – ni(t) be the number of ordered parts at time slot t, – pi(t) the mean price per part at time slot t, – di(t) the mean delivery time at time slot t for OEM i.Learning Package: Service Networks Performance Analysis © S-Cube - 38
  39. 39. Scenario #1 Information revealed for each OEM in each time slot is the number of ordered parts of its opponent for the previous time slot. We consider symmetric strategies (same rule for both OEMs), where delivery time di(t) is left fixed for the whole duration of the experiment and price pi(t) is changed according to the algorithm 1 (where i, j = A, B, i≠j and ε>0).Learning Package: Service Networks Performance Analysis © S-Cube - 39
  40. 40. Scenario #1 Cont. & Results The price for OEM i is determined as following: – Value is compared in two consecutive time slots and the number of orders with that of its competitor. – If we observe a loss in the value and a higher market share than our opponent, then we decide to increase price by a small increment since we can afford a small reduction in the number of customers but we will gain more revenues aiming at a total increase of value. Results:Learning Package: Service Networks Performance Analysis © S-Cube - 40
  41. 41. Scenario #1: Results OEM A’s value decreases and OEM B’s value increases up to month 4, where equilibrium is reached. – That is, neither OEM A nor OEM B are willing to change the derived prices followed by their strategies after month 4. The intuition behind this result is that higher prices for OEM A imply higher service prices for dealers, thus, decreasing customer satisfaction. As a consequence, a considerable portion of market share has moved from OEM A to OEM B so that symmetric strategies have created networks of comparable size.Learning Package: Service Networks Performance Analysis © S-Cube - 41
  42. 42. Scenario #2 Information revealed for each OEM in each time slot is the number of ordered parts of its opponent for the previous time slot and additionally for OEM of network B, the price of the previous time slot set by OEM of network A. The strategy for each OEM is not symmetric (the two OEMs follow a different strategy), taking delivery time to be fixed for the whole duration of the experiment and price to be changed as shown in the following:STRATEGY FOR A STRATEGY FOR BLearning Package: Service Networks Performance Analysis © S-Cube - 42
  43. 43. Scenario #2 Cont. & Results In this case, we change the order of the criteria placing the comparison of the number of orders first. Additionally, player B uses its opponent’s former prices to gain competitive advantage when the performance of the network it participates is observed to be decreased. Results:Learning Package: Service Networks Performance Analysis © S-Cube - 43
  44. 44. Scenario #2: Results The second scenario in which the most recently set up network imitates its opponent’s decisions in case of unstable situations, has similar results as the first one. Customers of network A join network B in presence of lower prices up to month 4 at which an equilibrium is reached. Despite the same shapes of value curves between the two networks under the two scenarios, the actual value level for each network is increased in the second scenario. This is due to the fact that the prices change more aggressively in the second scenario without implying losses in value.Learning Package: Service Networks Performance Analysis © S-Cube - 44
  45. 45. Scenario #3 Information revealed for each OEM in each time slot is the number of ordered parts of its opponent for the previous time slot. The strategy for OEM of network A is the same as in scenario 2 but OEM of network B follows a different strategy that is considered to be more risky since it takes into account the variability of mean delivery time.Learning Package: Service Networks Performance Analysis © S-Cube - 45
  46. 46. Scenario #3: Results OEM B changes his strategy to encounter delivery time in addition to price. This has as a result an aggressive increase in customer satisfaction, since customers are given the opportunity to choose a relatively small increase in price at a shorter service time. This entails an increase in value for OEM B up to month 8 after which an equilibrium is reached. On the contrary, OEM A faces a loss in its value in the absence of time. At month 4, the value of OEM B becomes higher compared to the value of OEM A showing that at this time slot the number of customers of the second network gets larger than that of the first one. The flexibility practiced by OEM B gave him the opportunity to change the balance of the market to its own benefit.Learning Package: Service Networks Performance Analysis © S-Cube - 46
  47. 47. Scenario #4 We study the behavior of the weaker competitor (network B) given that it first observes opponent’s behavior. Again, the number of ordered parts in the two networks is common knowledge to both of them. OEM of network A (price leader) determines its own price following the same strategy as in scenario 1. According to this price, OEM of network B (price follower) sets its price for the next time slot according to the rule given below:Learning Package: Service Networks Performance Analysis © S-Cube - 47
  48. 48. Scenario #4: Results The leader (network A) faces the smaller loss in value (and consequently in the number of customers) compared to the other scenarios. This is explained by the fact that it is the first to choose a price and predict its opponent’s choice that will be based on this price. Equilibrium is reached at month 6 with OEM A having a significantly larger value than that of OEM B.Learning Package: Service Networks Performance Analysis © S-Cube - 48
  49. 49. Learning Package Overview Problem Description Service Network Performance Analysis Performance Analysis of Competing Service Networks Discussion and SummaryLearning Package: Service Networks Performance Analysis © S-Cube - 49
  50. 50. Discussion Existing approaches and methodologies regarding the measurement of service networks’ performance – mainly focus on describing models that represent inter-organization exchanges. – they do not study strategic behavior of network participants that would result in value optimization. – competing networks that co-exist in ecosystem has not been properly studied. Comparing to previous work, – the estimation techniques has been improved – Competitors of an existing service network are taken into consideration in order to define: - the conditions under which two competing networks co-exist - the types of information the networks share - various strategies chosen by the competing participants. – a powerful simulation tool, iThink, has been used to perform experiments and analyze dynamic “what-if” questionsLearning Package: Service Networks Performance Analysis © S-Cube - 50
  51. 51. Summary The work presented contributes towards the evaluation of performance of service systems – By estimating value – By studying the impact of strategic changes on the performance both at the level of the network as well as its participants, – By introducing an analytical model and an associated simulation tool have been introduced to optimize value. – By defining and solving optimization problems in respect to service prices. – By analyzing strategic interactions between competing networks co- existing in the ecosystem and interacting with one another to their own benefit is also introduced.Learning Package: Service Networks Performance Analysis © S-Cube - 51
  52. 52. Further ReadingCaswell, N. S., Nikolaou, C., Sairamesh, J., Bitsaki, M., Koutras, G. D., and Iacovidis, G. Estimating value inservice systems: A case study of a repair service system. IBM Systems Journal, 47, 1 (2008), 87-100.Voskakis, M., Nikolaou, C., Bitsaki, M., and van de Heuvel, Willem-Jan. Service Network Modeling andPerformance Analysis. In Sixth International Conference on Internet and Web Applications and Services (2011).M.Bitsaki, C. Nikolaou, M.Voskakis, Willem-Jan van den Heuvel, K. Tsikrikas: Performance Analysis andStrategic Interactions in Service Networks. Submitted in Journal On Advances in Networks and Services.Learning Package: Service Networks Performance Analysis © S-Cube - 52
  53. 53. Acknowledgements The research leading to these results has received funding from the European Community’s Seventh Framework Programme [FP7/2007-2013] under grant agreement 215483 (S-Cube).Learning Package: Service Networks Performance Analysis © S-Cube - 53
  1. A particular slide catching your eye?

    Clipping is a handy way to collect important slides you want to go back to later.

×