Consumer satisfaction versus churn in the case of upgrades of 3G to 4G cell networks

3 collaborators

Lester Johnson (Advisor)

Leanne Carter (Team member)

WHAT IS IT?

This model simulates the competitive dynamics between 3G and 4G telecommunications networks in a market where consumers make adoption decisions based on bandwidth requirements and price sensitivity. The model examines how network capacity, pricing strategies, and consumer tolerance for service quality affect market share, customer satisfaction, and churn rates in telecommunications markets.

The simulation represents customers as agents who search for network services that meet their bandwidth needs within their price constraints. Network infrastructure is represented as a spatial grid where each location offers either 3G or 4G service with different capacity and pricing characteristics. The model captures key marketing phenomena including service switching behavior, customer satisfaction dynamics, and market equilibration processes.

Empirical Validation: This model has been validated through comparison with longitudinal survey data (N=971) as reported in D'Alessandro et al. (2015, Marketing Letters). The validation study demonstrated that the agent-based model accurately predicts the relative importance of coverage, customer service, and price in determining satisfaction and switching behavior, supporting the use of ABM as a cost-effective tool for understanding telecommunications market dynamics.

HOW IT WORKS

Agent Initialization: Customers (turtles) are created with heterogeneous characteristics drawn from normal distributions: - Bandwidth requirements: mean of 100 units with standard deviation of 10 - Price sensitivity (willingness to pay): mean of 20 with standard deviation of 20

Network Infrastructure: The spatial environment (patches) is divided between 3G (green) and 4G (black) network cells based on the FourG parameter: - 4G networks offer higher bandwidth capacity (200 units) but can be priced differently from 3G - 3G networks offer lower bandwidth capacity (120 units) with independent pricing - Each network cell has a capacity constraint limiting the number of simultaneous users

Customer Search and Switching Behavior: Each time step, unsatisfied customers search their local neighborhood (radius 1) for network cells that: 1. Provide sufficient bandwidth to meet their requirements 2. Are priced within their willingness to pay 3. Have available capacity

If suitable options exist, customers move to one of these cells and become satisfied (turning red). If no suitable options exist, customers move randomly and their churn counter increments.

Churn Mechanism: Customer dissatisfaction accumulates through a tolerance-adjusted churn counter. When this counter exceeds 12, customers leave the market permanently. This represents realistic customer attrition in competitive telecommunications markets where sustained dissatisfaction leads to market exit.

Equilibrium Conditions: The simulation runs until either: - All customers find satisfactory service (market equilibrium) - 50 time steps elapse (observation window)

HOW TO USE IT

Setup: 1. Use the FourG slider (0-100%) to set the proportion of 4G network infrastructure in the market 2. Set the number of Customers (0-1000) entering the market 3. Adjust the capacity parameter (0-100) to determine how many customers each network cell can serve simultaneously 4. Set the tolerance level (0-5) which affects how quickly dissatisfied customers churn 5. Configure Price3G and Price4G (0-50) to test different pricing strategies 6. Click "Setup" to initialize the model

Running: Click "go" to run the simulation. The model will execute until equilibrium is reached or 50 ticks elapse.

Monitoring: - The "happy" monitor shows the current number of satisfied customers - "Market size" monitor displays the total number of customers (decreases with churn) - "Customer mix" plot shows the distribution of customers across 3G and 4G networks over time - "Percent Happy Customers" plot tracks overall satisfaction levels - "Network Use" plots show excess capacity across network types - "Mean use of Capacity" plots indicate utilization rates for each network

THINGS TO NOTICE

Network Effects and Congestion: Observe how capacity constraints create congestion externalities. As more customers join a network cell, it may reach capacity and become unavailable to new customers, forcing them to continue searching or accept inferior alternatives.

Price-Quality Tradeoffs: Notice how the interaction between network quality (bandwidth capacity) and price affects adoption patterns. Higher-capacity 4G networks may command price premiums, but excessive pricing can drive customers to 3G alternatives or out of the market entirely.

Churn Dynamics: Watch the "Market size" monitor to observe customer attrition. Markets with insufficient infrastructure or poorly matched price points will experience higher churn rates, reducing total market size over time.

Spatial Distribution: The visual display shows clustering patterns as satisfied customers (red) concentrate in network cells that best match their requirements. Unsatisfied customers (white) continue searching.

THINGS TO NOTICE

This section could give some ideas of things for the user to notice while running the model.

THINGS TO TRY

Experiment 1: Infrastructure Investment vs. Pricing - Run the model with low FourG (5-10%) and low Price4G (10) vs. high FourG (50%+) and high Price4G (30-40) - Compare total customer satisfaction and churn rates - Question: Is it better to invest in widespread infrastructure with lower margins or premium pricing on limited infrastructure?

Experiment 2: Capacity Planning - Hold pricing constant and vary the capacity parameter from 5 to 25 - Observe network utilization rates and customer satisfaction - Question: What is the optimal capacity level that balances infrastructure costs (represented by capacity) with service quality?

Experiment 3: Price Discrimination Strategies - Set different price differentials between 3G and 4G (e.g., equal pricing vs. 2:1 price ratio) - Monitor how customers sort between networks and overall market satisfaction - Question: What price differential maximizes total market coverage while optimizing revenue?

Experiment 4: Tolerance and Switching Costs - Compare outcomes with tolerance = 0 (immediate switching) vs. tolerance = 5 (high switching costs) - Observe how this affects market stability and equilibrium patterns - Question: How do customer switching costs affect competitive dynamics?

Using BehaviorSpace: The model includes a pre-configured experiment "4G-3G with churn=12" that systematically varies: - Price3G: 10 vs. 20 - FourG deployment: 5% vs. 10% - Price4G: 10 vs. 20
- tolerance: 0 vs. 1 - capacity: 5 vs. 10

This creates a 2×2×2×2×2 = 32 experimental conditions to explore the parameter space comprehensively.

EXTENDING THE MODEL

Quality of Service Metrics: Currently, bandwidth capacity is binary (sufficient or insufficient). The model could be extended to include graduated service quality measures, where customers experience different levels of satisfaction based on how well the network exceeds their requirements (Rogers, 2003).

Dynamic Pricing: Implement time-varying or demand-based pricing where networks adjust prices in response to congestion or competitive pressure, representing real-world revenue management strategies in telecommunications (Varian, 1996).

Network Infrastructure Investment: Allow operators to upgrade patches from 3G to 4G during simulation in response to demand patterns or competitive dynamics, representing strategic capacity expansion decisions (Economides, 1996).

Consumer Learning and Word-of-Mouth: Add social influence mechanisms where customers share information about network quality with nearby agents, affecting search behavior and adoption decisions (Bass, 1969; Rogers, 2003).

Multi-Attribute Decision Making: Expand customer decision criteria beyond bandwidth and price to include factors such as network reliability, brand reputation, contract terms, or device ecosystems (Fishbein & Ajzen, 1975).

Heterogeneous Usage Patterns: Model customers with time-varying bandwidth requirements (peak vs. off-peak usage) to capture more realistic demand patterns and congestion dynamics.

NETLOGO FEATURES

This model demonstrates several important NetLogo programming features and techniques:

Agent and Patch Properties: The model uses turtles-own to give each customer agent individual properties (bandwidth requirements, price sensitivity, satisfaction state, churn counter) and patches-own to assign network characteristics (bandwidth capacity, price, network type). This allows for heterogeneous populations and spatially differentiated infrastructure.

Global Variables: Seven global variables track aggregate market outcomes (Happy%, Loss, Mean3G, Mean4G, Total3G, Total4G, happy) that are updated each tick and displayed in monitors, allowing real-time observation of market dynamics.

Random-Normal Distribution: The model uses random-normal to create heterogeneous customer populations with bandwidth requirements (mean=100, SD=10) and price sensitivities (mean=20, SD=20) that follow normal distributions. This creates more realistic market segmentation than uniform random distributions and models the natural variation in consumer preferences.

Spatial Search with In-Radius: The patches in-radius 1 construct efficiently identifies candidate network cells within a customer's search radius, representing localized geographic search behavior. The model then filters this neighborhood using nested with clauses to find patches meeting both bandwidth and price requirements.

Agent State Changes: Agents change visual representation (color from white to red, facial expression in the observer window) when they transition from unsatisfied to satisfied states, providing immediate visual feedback on market equilibration.

Agent Life Cycle: The model demonstrates agent creation (create-turtles) and deletion (die), where unsatisfied customers eventually leave the market through the churn mechanism, dynamically adjusting market size.

Conditional Movement and Search: The model uses ifelse with multiple conditional filters to implement sophisticated search-and-match behavior. Customers evaluate bandwidth availability, price constraints, and capacity limits before making location decisions, representing multi-attribute decision-making.

Time-Step Simulation: The model uses NetLogo's tick counter to simulate contract periods, with the churn mechanism triggering after 12 ticks (representing a 12-month contract). This demonstrates how to model time-dependent processes and contractual obligations.

Multiple Plot Types: The model includes four different plots that update dynamically: - Customer mix across network types (line plot) - Percentage of satisfied customers over time (line plot) - Network capacity utilization (multi-line plot) - Mean network usage (comparative line plot)

Real-time Monitors: Multiple monitors display current values of key metrics (number of happy customers, total market size), providing immediate feedback during model execution.

Interactive Controls: Five sliders allow users to manipulate model parameters (FourG coverage, number of customers, capacity, tolerance, Price3G, Price4G) and observe outcomes in real-time, making the model suitable for scenario exploration and teaching.

BehaviorSpace Integration: The model includes a pre-configured BehaviorSpace experiment that systematically varies all five parameters across 32 conditions, demonstrating NetLogo's capability for automated factorial design experimentation. This allows for comprehensive parameter space exploration and sensitivity analysis.

Efficient Counting and Filtering: The model uses NetLogo's count primitive with complex filtering conditions (e.g., count turtles with [pcolor = 0]) to efficiently calculate aggregate statistics about customer distribution and network utilization.

Color Coding for Categorical Variables: The model uses patch colors (green=3G, black=4G) and turtle colors (white=unsatisfied, red=satisfied) to create intuitive visual representations of network types and customer states, making the spatial dynamics easy to interpret.

RELATED MODELS

NetLogo Models Library: - Segregation Model (Schelling): Similar agent-based approach to examining how individual preferences aggregate into macro-level patterns - Traffic Basic: Demonstrates congestion dynamics and capacity constraints similar to network congestion in this model - Virus on a Network: Shows diffusion processes on network structures, analogous to technology adoption patterns

Telecommunications and Technology Adoption Models: This model relates to agent-based simulations of technology diffusion and network effects in telecommunications markets, though specific NetLogo implementations may vary.

CREDITS AND REFERENCES

D'Alessandro, S., Johnson, L., Gray, D., & Carter, L. (2015). Consumer satisfaction versus churn in the case of upgrades of 3G to 4G cell networks. Marketing Letters, 26(4), 489-500. https://doi.org/10.1007/s11002-014-9284-3 [This paper presents the validation of this NetLogo model through comparison with longitudinal survey data from 971 participants. The study demonstrates that the ABM accurately predicts key parameters including coverage, customer service, and price effects on satisfaction, providing empirical support for using agent-based modeling as a cost-effective method for understanding consumer behavior in telecommunications markets.]

Theoretical Foundations:

Bass, F. M. (1969). A new product growth model for consumer durables. Management Science, 15(5), 215-227. https://doi.org/10.1287/mnsc.15.5.215 [Foundational work on innovation diffusion and technology adoption modeling]

Economides, N. (1996). The economics of networks. International Journal of Industrial Organization, 14(6), 673-699. https://doi.org/10.1016/0167-7187(96)01015-6 [Network effects and competitive dynamics in telecommunications markets]

Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behavior: An introduction to theory and research. Addison-Wesley. [Multi-attribute decision-making framework applicable to service choice]

Rogers, E. M. (2003). Diffusion of innovations (5th ed.). Free Press. [Classic framework for understanding technology adoption and diffusion processes]

Varian, H. R. (1996). Differential pricing and efficiency. First Monday, 1(2). https://doi.org/10.5210/fm.v1i2.473 [Theoretical foundation for price discrimination in information goods and services]

Additional Marketing and Telecommunications Literature:

Bolton, R. N. (1998). A dynamic model of the duration of the customer's relationship with a continuous service provider: The role of satisfaction. Marketing Science, 17(1), 45-65. https://doi.org/10.1287/mksc.17.1.45 [Customer retention and churn modeling in service industries]

Kim, M. K., Park, M. C., & Jeong, D. H. (2004). The effects of customer satisfaction and switching barrier on customer loyalty in Korean mobile telecommunication services. Telecommunications Policy, 28(2), 145-159. https://doi.org/10.1016/j.telpol.2003.12.003 [Empirical analysis of switching behavior in telecommunications]

Shapiro, C., & Varian, H. R. (1998). Information rules: A strategic guide to the network economy. Harvard Business School Press. [Strategic analysis of pricing and competition in network industries]

Model Development: This model was developed for research purposes to examine competitive dynamics in telecommunications markets during the transition from 3G to 4G networks. The model has been peer-reviewed and validated through empirical research published in Marketing Letters (D'Alessandro et al., 2015).

Authors: - Steven D'Alessandro, Charles Sturt University - Lester Johnson, Charles Sturt University & Melbourne Business School
- David Gray, Macquarie University - Leanne Carter, Macquarie University

The model demonstrates how agent-based modeling can provide cost-effective and time-efficient insights into complex consumer behavior, particularly for understanding technology adoption, pricing strategies, and customer retention in network markets. It can be used for educational purposes to explore strategic marketing decisions or as a foundation for further research into telecommunications market dynamics.

For questions or comments about this model, please contact the authors through the NetLogo Modeling Commons or refer to the published validation study in Marketing Letters.

Version: 10 (price-focused variant) NetLogo Version: 5.0.4 or higher License: This model is available for educational and research use

Comments and Questions

Please start the discussion about this model! (You'll first need to log in.)

Click to Run Model

turtles-own [ my-bandwidth happy? churn my-price ]
patches-own [ band-capacity price]
globals [ Happy% happy Mean3G Mean4G Loss Total3G Total4G]

to setup
  ;; (for this model to work with NetLogo's new plotting features,
  ;; __clear-all-and-reset-ticks should be replaced with clear-all at
  ;; the beginning of your setup procedure and reset-ticks at the end
  ;; of the procedure.)
  __clear-all-and-reset-ticks
  set-default-shape turtles "person"
  setup-patches
  setup-turtles
  clear-all-plots
end 

to setup-patches
  ;; (for this model to work with NetLogo's new plotting features,
  ;; __clear-all-and-reset-ticks should be replaced with clear-all at
  ;; the beginning of your setup procedure and reset-ticks at the end
  ;; of the procedure.)
  __clear-all-and-reset-ticks
  ask patches [set pcolor one-of [green black]]
  ask patches
    [
      ifelse (random 100 < FourG)
      [
        set pcolor black
        set band-capacity 200
        set price price3G
      ]
      [
        set pcolor green
        set band-capacity 120
        set price price4G
      ]
    ]
end 

to setup-turtles
  create-turtles customers [
    set color white
    setxy random-xcor random-ycor
    set size 1
    set my-bandwidth 100 + random-normal 50 10
    set my-price 10 + random-normal 10 20
    set happy? false
    set churn 0

  ]
end 

to go
  if count turtles = count turtles with [happy? = true]
  [stop]
  if ticks > 49 [stop]
  ask turtles [search-for-bandwidth]
  wait 0.2
  ask turtles [leave]
  do-plot-network-use
  do-plots
  do-plot-happy
  do-results
  do-plot-mean-use
  tick
end 

to search-for-bandwidth
if not happy?
    [
    let my-neighborhood patches in-radius 1 with [ band-capacity >  [ my-bandwidth ] of myself  and count turtles-here < capacity ]
    print (word "Agent " who " looking at " my-neighborhood )
    ;let my-neighborhood1 patches in-radius 1 with [ price <  [ my-price ] of myself  and count turtles-here < capacity ]
    let my-neighborhood1 my-neighborhood with [ price <  [ my-price ] of myself  and count turtles-here < capacity ]
    print (word "Agent " who " looking at " my-neighborhood1 )

    set churn churn - tolerance + 1
    ifelse (count my-neighborhood1 > 0)
    [
      let my-target one-of my-neighborhood1
      setxy [ pxcor ] of my-target [ pycor ] of my-target
      set happy? true
      set color red
    ]
    [
     fd 1
     left random 45
     right random 45
    ]
    ]
end 

to leave
 if churn > 12 [die]
end 

to do-plots
             set-current-plot "Customer mix"

               set-current-plot-pen "4G"
               plot count turtles with [pcolor = 0]

              set-current-plot-pen "3G"
              plot count turtles with [pcolor = 55]

              set-current-plot-pen "total"
              plot count turtles
end 

to do-plot-happy
  set-current-plot "Percent Happy Customers"
  plot (count turtles with [ color = red ] / count turtles) * 100
end 

to do-results
 set Happy%  ((count turtles with [ color = red ] / count turtles) * 100)
 set Loss ( customers - count turtles )
 set Mean3G (count turtles with [ pcolor = 0] / ( count patches with [pcolor = 0]* capacity ))
 set Total3G count (turtles with [ pcolor = 55 ])
 set Total4G count turtles with [ pcolor = 0 ]
 set happy count turtles with [happy? = true]
 set Mean4G (count turtles with [ pcolor = 55 ] / ( count patches with [pcolor = 55]* capacity ))
 type FourG type " "  type Customers  type " " type capacity  type " " type tolerance  type" " type " " type Price3G type " " type Price4G  type " " type Happy% type " " type Loss type " "
 type Total3G type " " type Total4G
 type Mean3G  type " " type Mean4G type " " type happy
end 

to do-plot-network-use
   set-current-plot "Network Use"
   set-current-plot-pen "Total Excess Capacity"
    plot count patches with [capacity - count turtles-here > 1 ]
   set-current-plot-pen "3G Excess Capacity"
   plot count patches with [pcolor = 55 ] with [capacity - count turtles-here  > 1 ]
    set-current-plot-pen "4G Excess Capacity"
    plot count patches with [pcolor = 0 ] with [capacity - count turtles-here > 1 ]
end 

to do-plot-mean-use
  set-current-plot "Mean use of Capacity"
  set-current-plot-pen "Mean use of 3G"
  plot (count turtles with [ pcolor = 0 ] / (count patches  with [ pcolor = 0 ] * capacity))
  set-current-plot-pen "Mean use of 4G"
  plot (count turtles with [ pcolor = 55 ] / (count patches  with [ pcolor = 55 ] * capacity))
end

There is only one version of this model, created 2 days ago by Steven D'Alessandro.

Attached files

File	Type	Description	Last updated
Consumer satisfaction versus churn in the case of upgrades of 3G to 4G cell networks.png	preview	Preview for 'Consumer satisfaction versus churn in the case of upgrades of 3G to 4G cell networks'	2 days ago, by Steven D'Alessandro	Download

This model does not have any ancestors.

This model does not have any descendants.

NetLogo

Consumer satisfaction versus churn in the case of upgrades of 3G to 4G cell networks

3 collaborators

Close

Tags

Close

WHAT IS IT?

HOW IT WORKS

HOW TO USE IT

THINGS TO NOTICE

THINGS TO NOTICE

THINGS TO TRY

EXTENDING THE MODEL

NETLOGO FEATURES

RELATED MODELS

CREDITS AND REFERENCES

Comments and Questions

Attached files