Emerging Technologies and Business Value

Introduction

Businesses compete not only through products and services but also through the technologies used to design, produce, distribute, and improve them. Cloud computing, mobile platforms, blockchain, the Internet of Things (IoT), machine learning, and generative artificial intelligence (AI) have each altered assumptions about cost, speed, scale, and customer experience. Yet calling a technology “emerging” does not guarantee that it will create business value. Many technologies attract attention but fail to reach reliable, economical, or responsible deployment.

This chapter establishes a foundation for evaluating emerging technologies in business analytics. It explains what makes a technology emerging, how technologies evolve and diffuse, why adoption follows recognizable but imperfect cycles, how platform economics shape competition, and how value is created and measured. It also distinguishes four related but different capabilities: automation, analytics, machine learning, and generative AI.

The central managerial question is not “What can this technology do?” but “Which valuable business problem can it solve, under what conditions, and with what risks?”

Learning Objectives

By the end of this chapter, students should be able to:

1. Define emerging technology and distinguish it from merely new technology.
2. Explain technology evolution, diffusion, and adoption cycles.
3. Analyze platform economics, network effects, switching costs, and ecosystem advantage.
4. Distinguish automation, analytics, machine learning, and generative AI.
5. Evaluate business value using financial, operational, customer, innovation, and risk perspectives.
6. Identify implementation, ethical, governance, and regulatory challenges.
7. Apply a structured framework to an organizational technology decision.

1. What Is an Emerging Technology?

An emerging technology is a technology whose use, capabilities, economic significance, or institutional consequences are still developing. The term refers less to chronological age than to uncertainty and transition. A technology may be decades old but remain emerging in a particular industry. For example, robotics is mature in automobile manufacturing but may be emerging in small professional-service firms.

Four characteristics commonly indicate emergence:

• Technical development: capabilities, reliability, performance, or interoperability are still improving.
• Market uncertainty: demand, pricing, competitors, and viable business models are not yet settled.
• Institutional adjustment: standards, regulation, skills, professional practices, and social expectations are still forming.
• Potential for systemic impact: the technology may change relationships among firms, customers, workers, and public institutions.

An emerging technology should therefore be assessed on at least three dimensions: technical maturity, adoption maturity, and governance maturity. A technically impressive system may still be unsuitable for a business if its data are poor, its costs are unpredictable, or accountability is unclear.

Technology, innovation, and digital transformation

These terms are related but not interchangeable.

• Technology is a method, tool, system, or body of knowledge used to achieve practical purposes.
• Innovation is the implementation of a new or significantly improved product, process, marketing method, or organizational method that creates or captures value.
• Digital transformation is a broader organizational change in which digital technologies alter processes, capabilities, business models, and relationships with stakeholders.

Purchasing software is not necessarily innovation, and introducing an AI model is not necessarily transformation. Transformation requires changes in work, decision rights, processes, skills, and performance measurement.

2. Technology Evolution and Adoption

Technology evolution is often cumulative. New systems combine earlier components, reduce costs, improve usability, or make previously impractical applications feasible. The evolution of business computing illustrates this pattern:

1. Transaction processing digitized repetitive record keeping.
2. Management information systems organized data into periodic reports.
3. Decision-support systems enabled interactive analysis and modelling.
4. Enterprise systems and cloud computing connected functions and provided scalable infrastructure.
5. Machine learning identified patterns and generated predictions from data.
6. Generative AI produced text, code, images, audio, and other content in response to instructions.

These stages overlap. A firm can use transaction processing, analytics, machine learning, and generative AI simultaneously. New technology normally complements some older technologies rather than replacing all of them.

The technology life cycle

A common life-cycle model contains five stages:

1. Research and invention: scientific or engineering feasibility is explored.
2. Experimentation: early users test prototypes and uncertain applications.
3. Early commercialization: suppliers offer products, but standards and business models remain unstable.
4. Mainstream adoption: costs, skills, infrastructure, and use cases become more predictable.
5. Maturity or displacement: growth slows, incremental improvement dominates, or another technology replaces the system.

The life cycle is not deterministic. Some technologies remain niche; others experience a revival after complementary infrastructure becomes available. Smartphones, for example, depended on advances in mobile networks, processors, batteries, application stores, and digital payment systems.

Diffusion of innovation

Everett Rogers’ diffusion framework describes adoption through social groups: innovators, early adopters, early majority, late majority, and laggards. These groups differ in tolerance for uncertainty, access to resources, perceived risk, and social influence. The framework helps explain why a technology can be celebrated by experts yet remain uncommon among ordinary organizations.

Adoption is influenced by five perceived attributes:

• Relative advantage: perceived improvement over the existing approach.
• Compatibility: fit with existing values, processes, systems, and skills.
• Complexity: difficulty of understanding or using the technology.
• Trialability: ability to experiment before making a large commitment.
• Observability: visibility of results to potential adopters.

For business analytics, a small pilot improves trialability, while a clearly measured performance improvement increases observability.

The hype cycle and the productivity gap

Technology markets often pass through enthusiasm, disappointment, adjustment, and practical adoption. Early publicity can exaggerate capability, especially when demonstrations are mistaken for dependable performance. A productivity gap may arise between technical possibility and organizational benefit because implementation requires data preparation, process redesign, employee training, integration, and governance.

The appropriate managerial response is neither unconditional enthusiasm nor automatic rejection. Organizations should separate:

• what the technology can do in a controlled demonstration;
• what it can do reliably in the organization’s operating environment; and
• what the organization can govern, afford, and scale.

Figure: An illustrative adoption cycle. Actual technologies may skip, repeat, or remain in particular stages.

3. Platform Economics and Ecosystems

A platform is a foundation that enables interactions, transactions, or complementary products among different participant groups. Examples include digital marketplaces, payment networks, operating systems, cloud platforms, and application programming interface (API) ecosystems.

A conventional product creates value mainly through production and sale. A platform creates value by enabling interactions. Its success depends on governance, trust, standards, search, pricing, and the quality of complements.

Network effects

A network effect occurs when the value of a service changes as participation changes. A direct network effect arises when users benefit from more users of the same type, as in a communication network. An indirect network effect arises when one group attracts complementary participants, such as developers creating applications for an operating system.

Network effects can produce rapid growth and market concentration, but they are not automatically positive. Poor-quality participants, fraud, congestion, privacy concerns, or incompatible standards can reduce value. Platforms must actively manage participation rather than simply maximize user numbers.

A simplified representation of potential interaction opportunities in a single-sided network is:

Here, n is the number of participants and I is the number of possible undirected participant pairs. This is only an illustrative upper bound: real users do not interact with everyone, and interaction quality matters more than the number of theoretical pairs. If n = 5, then I = 5 × 4 ÷ 2 = 10.

Multi-sided platforms

A multi-sided platform serves two or more interdependent groups. A payment platform connects buyers, merchants, banks, and sometimes advertisers. A ride-hailing platform connects riders and drivers. One group may be subsidized to attract the other. Pricing therefore cannot be evaluated only by the cost of serving each group; it must consider cross-group effects.

Important platform mechanisms include:

• Cross-side network effects: more participants on one side increase value for another side.
• Switching costs: financial, technical, contractual, or learning costs of changing providers.
• Data advantages: usage data can improve matching, personalization, prediction, or quality control.
• Complementors: independent developers, suppliers, partners, and creators who increase platform usefulness.
• Governance: rules concerning access, ranking, safety, privacy, pricing, dispute resolution, and data use.

Platform power can create economies of scale and innovation, but it can also create dependency. A business may become vulnerable if a platform changes fees, restricts access, uses partner data competitively, or experiences an outage.

4. From Technology to Business Value

Business value is the net contribution of a technology to organizational objectives. It includes financial returns but is not limited to immediate profit. Value may arise through revenue growth, cost reduction, risk reduction, service quality, resilience, employee capability, or strategic flexibility.

A useful value chain is:

Technology capability → organizational use → process change → measurable outcome → stakeholder value

A capability produces no value unless it is embedded in a process and connected to a decision. For example, a predictive model has limited value if employees cannot interpret it, do not trust it, or cannot act before the predicted event occurs.

A value-assessment framework

Before implementation, managers should examine five questions:

1. Problem: What important business problem is being addressed?
2. Users and decisions: Who will use the output, and what decision will change?
3. Economics: What benefits, costs, and dependencies are expected?
4. Feasibility: Are data, skills, infrastructure, integration, and change capacity available?
5. Responsibility: Can the system be operated fairly, securely, transparently, and legally?

The basic return-on-investment measure is:

B represents measurable benefits over the evaluation period, and C represents total relevant costs over that period. Both must use the same currency and time period. Costs may include acquisition, implementation, training, integration, maintenance, cybersecurity, monitoring, and opportunity cost. ROI is useful for comparison but ignores timing, risk, and benefits that are difficult to monetize.

For example, a customer-service automation project produces annual measurable benefits of $180,000 and annualized total costs of $120,000. Its ROI is:

ROI = (180,000 − 120,000) ÷ 120,000 × 100% = 50%.

This calculation does not prove that the project should proceed. Management must still examine service quality, employee effects, security, legal duties, and whether the benefits are reliable.

Categories of business value

Strong evaluation combines leading indicators, such as adoption and task completion, with lagging indicators, such as profit and customer retention.

5. Distinguishing Four Business Capabilities

These capabilities are often confused because they can appear in the same system. They differ in purpose and method.

Automation

Automation executes a defined task or workflow with limited human intervention. Rule-based automation follows explicit conditions, such as routing an invoice when its amount is below a threshold. Robotic process automation can imitate user actions across existing applications. Automation is valuable when processes are stable, rules are clear, inputs are structured, and exceptions can be managed.

Automation does not necessarily learn. A rule that remains unchanged is automation even if it runs on a sophisticated cloud platform.

Analytics

Analytics uses data to describe, explain, predict, or prescribe. Four common forms are:

• Descriptive analytics: What happened?
• Diagnostic analytics: Why did it happen?
• Predictive analytics: What is likely to happen?
• Prescriptive analytics: What action should be considered?

Analytics may use dashboards, statistical analysis, optimization, or machine learning. Its distinguishing feature is the production of insight to support decisions, not necessarily automatic execution.

Machine learning

Machine learning (ML) is a computational approach in which a system learns patterns or parameters from data rather than relying only on hand-coded rules. In supervised learning, a model learns from examples containing inputs and known outcomes. In unsupervised learning, it identifies structure without labelled outcomes. Reinforcement learning uses feedback from actions and consequences.

A supervised prediction can be represented as:

Here, x is an input vector, θ represents learned parameters, f is the model, and ŷ is the prediction. The notation does not imply that the prediction is correct; performance depends on data quality, assumptions, evaluation design, and deployment conditions.

ML is appropriate when useful patterns exist, sufficient representative data are available, prediction quality can be measured, and errors can be managed. It may support automation, but prediction and action should remain conceptually separate.

Generative AI

Generative AI produces new content based on patterns learned from data. Outputs may include text, images, audio, video, software code, or structured data. Large language models generate sequences of tokens in response to prompts and contextual information. They are powerful for drafting, summarizing, classification, translation, coding assistance, and conversational interfaces.

Generative AI differs from conventional predictive analytics because its output is often open-ended content rather than a fixed numerical estimate or category. It can also be probabilistic and may produce fluent but inaccurate statements, commonly called hallucinations. Human review, grounding in trusted information, access controls, and output testing are therefore essential in high-impact settings.

The capabilities can be combined. For example, ML may predict which invoices are suspicious, analytics may display the risk pattern, generative AI may draft an explanation, and automation may route the case for review. Clear role separation is necessary because a generated explanation is not evidence that a prediction is accurate.

6. Applied Case: Generative AI in a Retail Service Centre

The following is an illustrative case, not a report of a specific company.

Background

A regional retailer receives approximately 40,000 customer-service contacts each month through email and chat. Agents spend substantial time searching policy documents, rewriting routine responses, and transferring complex cases. Management considers a generative-AI assistant that retrieves approved policy content, proposes a response, and leaves the agent responsible for approval and sending.

Problem

The retailer expects shorter response times and more consistent service, but it faces risks involving inaccurate answers, disclosure of personal information, unequal service, and overreliance by inexperienced staff.

Analysis

The proposed system is not simply “an AI chatbot.” It combines several capabilities:

• document retrieval and search provide relevant policy passages;
• a language model drafts a response;
• analytics measures response time, resolution, escalation, and customer satisfaction;
• workflow automation records approval and routes exceptions;
• human agents make the final decision for sensitive cases.

The business case must compare total benefits with total costs. Benefits may include reduced handling time, increased first-contact resolution, and lower training costs. Costs include software, integration, document maintenance, evaluation, security, monitoring, and employee training.

Approach and solution

The retailer conducts a controlled pilot for one product category. It creates an approved knowledge base, removes unnecessary personal data, labels AI-generated drafts, and requires human approval. High-risk topics—refund disputes, legal threats, vulnerable customers, and safety complaints—are automatically escalated. Evaluation uses a test set of representative cases and live monitoring.

Success criteria are defined before deployment:

• average handling time decreases without reducing satisfaction;
• factual accuracy meets a predetermined threshold;
• escalation decisions are reliable;
• no confidential data are sent to unauthorized services;
• agents understand when to reject or revise a draft.

Outcome

Suppose the pilot reduces average handling time by 12 percent and improves consistency, but evaluation finds that the system occasionally cites an outdated warranty rule. The organization therefore delays full deployment, establishes document ownership and expiry dates, and adds a retrieval check showing the source policy beside every draft. The pilot creates value only after the governance problem is corrected.

Learning points

1. The business value came from redesigning the service process, not from model novelty alone.
2. Human oversight was designed into the workflow rather than added after failure.
3. Accuracy had to be evaluated on realistic cases, including exceptions.
4. Knowledge management was as important as model selection.
5. A limited pilot reduced technical, financial, and reputational exposure.

7. Challenges, Ethics, and Governance

Emerging technologies create risks that are technical, organizational, social, and legal.

Data and model risks

Poor, incomplete, outdated, or unrepresentative data can produce unreliable results. ML systems may reproduce historical discrimination. Generative systems may produce fabricated content or expose sensitive information through prompts, logs, or outputs. Model drift occurs when the relationship between inputs and outcomes changes over time.

Human and organizational risks

Automation can eliminate tasks, redesign jobs, or shift responsibility without adequate training. Employees may overtrust system recommendations, a phenomenon known as automation bias. Conversely, excessive alerts or poor interfaces may lead to rejection of useful systems. Effective implementation includes participation, training, appeal mechanisms, and clear accountability.

Security and privacy

Controls should include least-privilege access, encryption, secure interfaces, audit logs, data minimization, vendor assessment, incident response, and testing against prompt injection or other attacks. Personal data should be collected and retained only for legitimate purposes and handled according to applicable privacy law.

Explainability and accountability

Not every system requires the same level of explanation. A low-risk recommendation may need a simple rationale, while a system affecting credit, employment, healthcare, or access to essential services may require stronger documentation, human review, impact assessment, and a method for contesting decisions.

Regulatory and professional responsibility

Organizations must monitor applicable laws and standards concerning privacy, consumer protection, intellectual property, discrimination, cybersecurity, records, and sector-specific duties. Regulation differs across jurisdictions and changes over time. Compliance is a minimum requirement, not a substitute for ethical judgment.

A practical governance structure assigns:

• a business owner accountable for intended outcomes;
• a technical owner accountable for performance and maintenance;
• a data steward accountable for data quality and permissions;
• a risk or compliance function accountable for controls;
• end users accountable for appropriate use and escalation.

8. Best Practices for Evaluating Emerging Technologies

1. Start with a clearly defined business problem rather than a fashionable tool.
2. Establish a baseline process and measurable success criteria.
3. Test technical feasibility with representative data and realistic exceptions.
4. Calculate total cost of ownership, not only license price.
5. Run a bounded pilot with an explicit stop or scale decision.
6. Design human oversight according to impact and risk.
7. Document data sources, assumptions, limitations, versions, and decisions.
8. Monitor performance, fairness, security, adoption, and unintended effects after launch.
9. Prefer interoperable architectures and avoid unnecessary dependence on one vendor.
10. Treat change management, training, and process redesign as core implementation work.

Contemporary Developments

Current technology development is increasingly characterized by convergence. Cloud platforms provide scalable computing; APIs connect systems; IoT devices produce real-time data; ML identifies patterns; and generative AI provides natural-language interfaces. Edge computing can process data near its source when latency, connectivity, or privacy matters. Digital twins can combine sensor data, models, and simulation to represent physical assets or processes.

These developments increase opportunity but also increase complexity. A system composed of many services may have more failure points, suppliers, data transfers, and governance obligations. Business analysts therefore need both technical literacy and institutional awareness: they must understand what a system does, how it is evaluated, who is affected, and who is accountable.

Chapter Summary

Emerging technologies are developing technologies whose technical capabilities, markets, uses, and governance arrangements remain uncertain. Their significance depends not only on invention but also on complementary infrastructure, organizational adoption, skills, standards, and regulation. Technology life cycles and diffusion models explain why adoption is uneven, while hype cycles warn that demonstrations may overstate practical value.

Platforms create value by enabling interactions among participant groups. Network effects, switching costs, data advantages, complementors, and governance can generate scale and competitive power, but may also create dependency and concentration. Business value should be traced from capability through organizational use and process change to measurable outcomes. ROI is helpful but must be supplemented by operational, customer, employee, strategic, risk, and social measures.

Automation executes defined rules; analytics produces insight; machine learning learns patterns for prediction or classification; and generative AI produces new content. These capabilities can be combined, but their outputs, risks, and controls differ. Responsible adoption requires pilots, representative evaluation, data governance, human oversight, security, privacy protection, monitoring, and clear accountability.

Key Terms

Emerging technology: A technology whose capabilities, adoption, economic significance, or governance are still developing.

Innovation: Implementation of a new or significantly improved product, process, marketing method, or organizational method that creates value.

Digital transformation: Organizational change in which digital technologies alter processes, capabilities, and business models.

Technology life cycle: The progression from research and experimentation through commercialization, adoption, and maturity or displacement.

Diffusion of innovation: The process through which an innovation spreads among members of a social system.

Relative advantage: The perceived improvement of an innovation over an existing alternative.

Platform: A foundation that enables interactions among users, producers, or complementary services.

Network effect: A change in the value of a service associated with changes in participation.

Switching cost: The cost or difficulty of changing from one provider, system, or technology to another.

Ecosystem: A network of organizations, technologies, users, and institutions that jointly create value.

Automation: Execution of a task or workflow with limited human intervention.

Analytics: Use of data to describe, explain, predict, or recommend actions.

Machine learning: Computational methods that learn patterns or parameters from data.

Generative AI: AI that produces new content such as text, images, audio, code, or structured data.

Model drift: Decline in model performance caused by changes in data or relationships over time.

Total cost of ownership: All relevant costs of acquiring, implementing, operating, maintaining, and retiring a technology.

Human oversight: Review, intervention, or accountability by people in the operation of a technology.

Automation bias: Excessive reliance on an automated recommendation or decision.

Data governance: Policies, roles, processes, and controls for responsible data management.

Return on investment: Benefits minus costs, divided by costs, usually expressed as a percentage.

Review Questions

5 Short Answer Questions

1. What makes a technology emerging rather than simply new?
2. Name and briefly explain the five perceived attributes that influence diffusion of innovation.
3. What is the difference between a direct and an indirect network effect?
4. Distinguish descriptive analytics from predictive analytics.
5. Why is human oversight important when using generative AI in customer service?

5 Long Answer Questions

1. Explain the technology life cycle and discuss why technologies may move through its stages unevenly.
2. Analyze how platforms create value through network effects, switching costs, data, and complementors.
3. Compare automation, analytics, machine learning, and generative AI, using a business example for each.
4. Explain why technical capability does not automatically produce business value.
5. Discuss the main ethical, privacy, security, and accountability issues associated with emerging technologies.

3 Analytical / Application Questions

1. A bank is considering an AI system that predicts loan default and automatically rejects applications. Apply the five-question value-assessment framework and recommend safeguards.
2. A manufacturer has sensor data, a predictive-maintenance model, and a generative-AI assistant for technicians. Classify the role of each capability and design a suitable human-workflow arrangement.
3. A digital marketplace has many buyers but few sellers. Analyze the platform’s network-effect problem and recommend pricing, governance, or partnership actions to improve participation on both sides.

Multiple Choice Questions with Answers

Question 1

Which characteristic most strongly indicates that a technology is emerging?

A. It has a high purchase price.

B. Its capabilities, uses, and governance are still developing.

C. It is used by at least one large company.

D. It replaces every previous technology.

Correct Answer: B. Its capabilities, uses, and governance are still developing.

Question 2

Which diffusion attribute refers to how well an innovation fits existing processes and values?

A. Observability

B. Trialability

C. Compatibility

D. Relative advantage

Correct Answer: C. Compatibility

Question 3

What is an indirect network effect?

A. Users benefit from more users of the same type.

B. One participant group increases value for another participant group.

C. A platform operates without any participants.

D. A firm reduces its switching costs through training.

Correct Answer: B. One participant group increases value for another participant group.

Question 4

A system follows the rule “if an invoice is below $1,000, route it for standard approval.” Which capability is illustrated primarily?

A. Generative AI

B. Unsupervised learning

C. Rule-based automation

D. Diagnostic analytics

Correct Answer: C. Rule-based automation

Question 5

Which capability primarily produces new text, images, code, or audio?

A. Generative AI

B. Transaction processing

C. Descriptive analytics

D. Rule-based workflow

Correct Answer: A. Generative AI

Question 6

A project has benefits of $180,000 and total costs of $120,000. Using ROI = (B − C) ÷ C × 100%, what is the ROI?

A. 33.3%

B. 40%

C. 50%

D. 60%

Correct Answer: C. 50%

Question 7

What is a major limitation of ROI as a technology-evaluation measure?

A. It cannot use monetary values.

B. It ignores all project costs.

C. It does not by itself capture timing, risk, or difficult-to-monetize benefits.

D. It can only be calculated for manufacturing projects.

Correct Answer: C. It does not by itself capture timing, risk, or difficult-to-monetize benefits.

Question 8

Which statement best distinguishes machine learning from conventional rule-based automation?

A. Machine learning always operates without data.

B. Machine learning learns patterns or parameters from data.

C. Machine learning cannot support prediction.

D. Machine learning is identical to a dashboard.

Correct Answer: B. Machine learning learns patterns or parameters from data.

Question 9

Why should a generative-AI draft be reviewed in a high-impact business process?

A. Generative AI cannot produce fluent text.

B. A fluent output may still be inaccurate, biased, or inappropriate.

C. Review eliminates the need for data governance.

D. Human review guarantees perfect accuracy.

Correct Answer: B. A fluent output may still be inaccurate, biased, or inappropriate.

Question 10

Which practice best supports responsible scaling of an emerging technology?

A. Deploying it everywhere before defining success criteria.

B. Ignoring exceptions because average performance is strong.

C. Using a bounded pilot, representative testing, monitoring, and assigned accountability.

D. Selecting the most expensive vendor available.

Correct Answer: C. Using a bounded pilot, representative testing, monitoring, and assigned accountability.