14. Digital Procurement, AI and the Future of SRM

Introduction and Context

Procurement has moved from being primarily a transactional purchasing function to becoming a digitally enabled source of value, resilience and strategic intelligence. The change is visible in the technologies used to identify suppliers, negotiate agreements, manage contracts, process orders, monitor performance and anticipate disruption. E-procurement suites, supplier networks, contract lifecycle management systems, robotic process automation, advanced analytics, digital twins, blockchain and artificial intelligence are reshaping how organisations interact with markets and suppliers.

Digital procurement, however, is not simply the replacement of paper documents with electronic forms. It involves redesigning processes, data structures, decision rights and relationships around digital information. A poorly designed digital process can automate waste, reinforce biased decisions or expose sensitive commercial information. Conversely, a well-governed digital procurement system can provide visibility across expenditure, shorten cycle times, improve compliance, strengthen supplier collaboration and support more informed strategic choices.

Supplier relationship management (SRM) is particularly affected by this transformation. Traditional SRM relied heavily on periodic meetings, spreadsheets, personal knowledge and retrospective performance reports. Digital SRM supplements these activities with shared platforms, continuous performance data, collaborative workflows and predictive insight. The supplier relationship therefore becomes both more measurable and more complex. Managers must understand what should be automated, what requires human judgement and how technology can support trust rather than merely increase surveillance.

This chapter examines the architecture, applications and managerial implications of digital procurement. It also considers the limits of technology, including data quality, cybersecurity, explainability, ethical responsibility and the need for human oversight.

Learning Objectives

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

1. Explain the role and architecture of contemporary digital procurement systems.

2. Distinguish among e-procurement suites, marketplaces, supplier platforms and contract lifecycle management tools.

3. Evaluate robotic automation, analytics, digital twins, blockchain and artificial intelligence in procurement contexts.

4. Design a technology-enabled SRM approach that integrates supplier collaboration and performance management.

5. Assess risks involving data quality, cybersecurity, algorithmic bias, privacy and explainability.

6. Recommend governance, implementation and change-management practices for procurement transformation.

7. Evaluate how emerging technologies may contribute to autonomous sourcing and new supply ecosystems.

From Electronic Purchasing to Digital Procurement Ecosystems

E-procurement refers broadly to the use of electronic systems to support purchasing activities. Early applications focused on electronic requisitions, purchase orders and invoice processing. Contemporary digital procurement extends across the entire source-to-pay and, increasingly, source-to-contract lifecycle. It connects internal users, buyers, suppliers, finance, legal, logistics and risk functions through common data and workflows.

A source-to-pay process begins with demand identification and sourcing, continues through supplier selection and contracting, and ends with ordering, receipt, invoicing and payment. A source-to-contract process concentrates on category strategy, supplier discovery, tendering, negotiation and contract creation. SRM operates across these processes by managing supplier performance, development, risk, communication and strategic collaboration.

A useful way to understand a digital procurement ecosystem is through four layers:

• Experience layer: portals, dashboards, mobile applications, catalogues and collaboration interfaces used by employees and suppliers.
• Process layer: workflows for requisitioning, tendering, approval, contracting, ordering, invoicing, performance management and issue resolution.
• Data and intelligence layer: supplier master data, contract data, transaction records, risk indicators, analytics and artificial intelligence models.
• Integration and control layer: application programming interfaces, identity management, cybersecurity controls, audit trails and links to enterprise resource planning, finance, logistics and external information sources.

The value of this architecture depends on integration. A supplier risk score that is not connected to sourcing decisions has limited practical value. Similarly, an electronic contract repository that cannot alert users to renewal obligations may function as storage rather than contract management. Digital maturity is therefore determined not by the number of applications purchased, but by the quality of process integration, information and decision-making.

Core Digital Platforms in Procurement

E-procurement suites and procure-to-pay systems

An e-procurement suite generally supports requisitioning, approvals, catalogue management, purchase orders, goods receipt and invoice matching. It can enforce delegated authority, preferred suppliers and negotiated prices. Three-way matching among the purchase order, receipt and invoice can reduce payment errors and improve control.

The principal managerial benefit is process visibility. Procurement leaders can identify maverick spending, monitor approval delays and compare actual purchases with negotiated agreements. Yet a suite may create frustration when catalogues are incomplete, approval paths are excessive or users must enter the same information repeatedly. Adoption depends on usability and on whether the system reflects real purchasing needs.

Digital marketplaces and supplier networks

Marketplaces provide searchable access to products and services from multiple suppliers. They can increase supplier reach, simplify comparison and support standardised purchasing. Business-to-business supplier networks may also facilitate electronic orders, shipment notices, invoices and payment information.

Marketplaces are not neutral environments. Search rankings, platform fees, supplier qualification rules and data ownership influence commercial outcomes. Procurement managers must examine whether a marketplace encourages genuine competition, concentrates dependence on a dominant platform or excludes smaller suppliers through technical and financial barriers.

Supplier relationship platforms

Supplier platforms provide a shared environment for onboarding, qualification, document exchange, performance reporting, corrective actions, sustainability information and communication. A supplier can maintain organisational details, certificates and banking information, while the buying organisation can apply approval rules and monitor expiry dates.

The strongest platforms support two-way collaboration. Suppliers should not experience the system merely as an additional compliance burden. Shared forecasts, innovation proposals, capacity information and structured issue resolution can make a platform a relationship instrument rather than a monitoring device.

Contract lifecycle management

Contract lifecycle management (CLM) covers the stages from request and drafting through negotiation, approval, execution, obligation management, amendment, renewal and expiry. CLM systems may include clause libraries, version control, electronic signatures, approval workflows and alerts.

Contract data is particularly important because commercial value is often lost after negotiation. A price adjustment clause, service-level commitment, rebate condition or insurance requirement has little value if it is not translated into operational tasks and monitored. Effective CLM connects legal language with procurement, finance and operational processes. It also preserves an auditable record of who approved changes and when.

Automation: Rules, Robots and Intelligent Workflows

Robotic process automation (RPA) uses software robots to perform repetitive, rule-based activities through user interfaces or system connections. In procurement, RPA may transfer supplier data between applications, validate invoice fields, send reminders, reconcile records or prepare routine reports. It is most suitable when processes are stable, inputs are structured and exceptions are limited.

RPA should not be confused with artificial intelligence. A robot may follow a defined rule—such as routing an invoice above a threshold for approval—without learning or making an open-ended judgement. Intelligent automation combines workflow, RPA, machine learning, natural language processing or decision rules.

Automation should begin with process analysis. If a process contains unnecessary approvals, unclear ownership or poor master data, automating it may accelerate defects. A practical assessment asks:

1. Is the activity repetitive and sufficiently standardised?

2. Are input data and business rules reliable?

3. What happens when an exception occurs?

4. Who owns the automated decision?

5. Can the result be audited and reversed?

Human intervention remains necessary for supplier disputes, unusual contractual language, strategic negotiations, ethical concerns and decisions involving material uncertainty.

Analytics and the Data Foundation of SRM

Procurement analytics ranges from descriptive reporting to prescriptive decision support. Descriptive analytics explains what happened, such as expenditure by category or supplier. Diagnostic analytics examines why it happened, such as price variance caused by fragmented demand. Predictive analytics estimates what may happen, such as the probability of late delivery or supplier financial stress. Prescriptive analytics recommends actions, such as reallocating volume or initiating a corrective action plan.

Supplier performance dashboards commonly combine indicators relating to quality, delivery, cost, responsiveness, innovation, compliance, safety and sustainability. A single composite score can be useful for prioritisation but may conceal important differences. A supplier with excellent delivery but serious labour-practice concerns should not necessarily be classified as equivalent to a supplier with balanced performance. Dashboards should therefore show underlying measures, weights, trends and confidence levels.

Data quality is a strategic issue. Common defects include duplicate suppliers, inconsistent names, outdated bank details, missing category classifications, incompatible units of measure and contracts that cannot be linked to transactions. Data governance should establish ownership, definitions, validation rules, cleansing routines, access rights and retention standards. The principle of “garbage in, garbage out” is especially important when analytics and AI are used to make recommendations.

Procurement teams should distinguish data relevance from data availability. More data does not automatically produce better decisions. A small number of reliable indicators connected to a clear decision may be more valuable than a large dashboard of unvalidated measures.

Digital Twins and Visibility of Supply Networks

A digital twin is a digital representation of a physical asset, process, facility or network that is updated with relevant data. In procurement and supply management, a digital twin may represent supplier capacity, inventory positions, transport routes, production dependencies or the flow of materials across tiers.

For example, a network model can simulate the effect of a plant shutdown, a port delay, a sudden demand increase or a change in supplier allocation. It can help managers compare scenarios before committing to a sourcing decision. Digital twins are most valuable where the system is complex, interdependent and exposed to disruption.

Their limitations must be recognised. A model is not the supply network itself. It depends on assumptions, data frequency, boundary definitions and representations of human behaviour. Tier-two and tier-three supplier data may be incomplete, and a model may create false confidence if uncertainty is hidden. Scenario results should therefore be presented with assumptions and ranges rather than as precise forecasts.

Blockchain and Trusted Records

Blockchain is a distributed ledger technology in which transactions are recorded across a network according to agreed validation rules. Its potential procurement applications include provenance, certification, chain-of-custody records, product authentication and automated execution through smart contracts.

The technology can reduce disputes where multiple parties need a common record and no single participant is fully trusted to control it. A traceability application, for instance, could record custody events for a sensitive product. However, blockchain does not guarantee that the original input was truthful. A fraudulent or inaccurate event can still be entered into an immutable ledger. Governance of participants, identity, data standards and verification at the point of capture remains essential.

Blockchain is therefore not a universal solution for procurement transparency. A conventional shared database may be cheaper and more appropriate where participants already trust a central administrator. The relevant question is not whether blockchain is fashionable, but whether distributed validation creates benefits that justify its complexity.

Artificial Intelligence in Sourcing and SRM

Artificial intelligence includes computational methods that perform tasks associated with perception, language, prediction, classification or decision support. In procurement, AI may classify expenditure, extract clauses from contracts, identify similar suppliers, forecast demand, detect anomalous invoices, predict delivery risk, recommend negotiation priorities or generate draft communications.

Predictive AI estimates likely outcomes from historical or real-time data. A model might identify suppliers with an increased probability of late delivery. Its output should be treated as a risk signal, not an unquestionable conclusion. Managers must investigate the drivers, consider current circumstances and document the response.

Generative AI produces text, summaries, scenarios, code or other content in response to prompts. It can assist with market research summaries, draft requests for information, compare contract clauses, prepare meeting agendas and produce first versions of supplier performance reports. It may also generate inaccurate statements, omit relevant context, reproduce bias or expose confidential information if used through an uncontrolled service.

AI applications should be evaluated against decision criticality. A low-risk application such as formatting a report requires less control than an application that recommends supplier exclusion or changes payment terms. Appropriate controls include approved tools, restricted data inputs, human review, source verification, logging, testing and clear accountability.

Explainability and human oversight

Explainability concerns the ability to understand how an AI system reached or supported an output. In procurement, an explanation may need to identify relevant variables, data sources, confidence, limitations and the reason a recommendation differs from alternatives. Full technical transparency is not always possible, but operational users must receive enough information to challenge and responsibly use the output.

Human oversight is not satisfied by a person clicking “approve.” The reviewer must have authority, competence, time and access to evidence. High-impact decisions should include escalation routes, independent review and the possibility of correcting the data or overriding the recommendation. Organisations should also monitor outcomes for disparate effects on small suppliers, suppliers from particular regions or other protected groups.

A Framework for Procurement Technology Decisions

Technology selection should begin with a business problem rather than a product demonstration. A useful evaluation framework considers five dimensions.

Strategic fit asks whether the technology supports category strategy, resilience, sustainability, supplier development and organisational objectives. Process fit assesses whether it improves a clearly defined workflow without creating unnecessary complexity. Information fit examines data quality, interoperability, ownership and analytical usefulness. Control fit considers security, auditability, compliance, privacy and decision rights. Adoption fit evaluates usability, supplier capability, training, incentives and change readiness.

A total-cost assessment should include licences, implementation, integration, data cleansing, cybersecurity, training, support and supplier enablement. Benefits should be measured through appropriate indicators such as cycle time, contract compliance, invoice exception rates, supplier response time, cost avoidance, risk visibility and user adoption. Savings attributed to technology should not be claimed without distinguishing price effects, demand changes, specification changes and accounting timing.

Implementation is generally more reliable when conducted incrementally. Organisations can establish a data and process baseline, pilot a meaningful category or supplier group, test controls, capture user feedback and expand after evidence of value. Procurement transformation is an organisational change programme, not merely an information technology project.

Cybersecurity, Privacy and Governance

Procurement systems contain commercially sensitive information, including prices, bids, contracts, bank details, forecasts, intellectual property and supplier credentials. Threats include phishing, ransomware, credential theft, fraudulent bank-account changes, unauthorised access, malicious insiders and third-party vulnerabilities.

Security controls should include strong identity management, multi-factor authentication, least-privilege access, encryption, segregation of duties, secure interfaces, continuous monitoring, tested backups and incident-response procedures. Supplier access must be governed because a platform can become an entry point into wider enterprise systems.

Privacy obligations arise when systems process personal information relating to supplier employees, contacts or beneficial owners. Data should be collected for legitimate purposes, retained appropriately and accessed only by authorised users. Contractual arrangements should define responsibilities among the buying organisation, technology provider and suppliers.

Governance should assign accountability for technology decisions. A procurement AI governance process may include a use-case register, risk classification, model approval, data impact assessment, testing requirements, performance monitoring and periodic review. Procurement, legal, information security, finance, data specialists and business users should participate where decisions have significant consequences.

Illustrative Case: Northstar Components’ Digital SRM Transformation

Background

Northstar Components, a fictional mid-sized manufacturer, sourced electronic and precision-engineered parts from approximately 420 active suppliers. Procurement used an enterprise resource planning system for orders and invoices, while supplier certificates, performance spreadsheets and contracts were stored in separate locations.

Situation and problem

The company experienced recurring late deliveries and inconsistent quality. Buyers had difficulty identifying which suppliers supported critical products, and contract renewal dates were occasionally missed. Senior management proposed purchasing an AI platform that promised autonomous supplier selection.

Analysis

A diagnostic review found that the principal problem was not an absence of advanced algorithms. Supplier names were duplicated, part numbers were inconsistent and delivery data used different definitions across plants. Forty per cent of supplier records lacked current certificates, and performance ratings were based on subjective monthly comments. The proposed AI system would have trained on unreliable and incomplete data.

The transformation team therefore classified decisions by risk. Routine onboarding and certificate reminders were considered suitable for workflow automation. Invoice anomaly detection was approved as a decision-support use case. Supplier exclusion and allocation decisions required human review and documented evidence.

Approach and solution

Northstar first created a supplier data governance group involving procurement, finance, quality, operations and information security. It established a single supplier identifier, standard performance definitions and ownership for critical fields. A supplier portal was introduced for document updates and corrective-action communication. CLM software linked contracts to suppliers, categories and renewal alerts.

An analytics dashboard then combined delivery, quality and responsiveness measures. Rather than relying on one score, it displayed trends and critical exceptions. A predictive model identified suppliers requiring investigation, but buyers had to record the evidence supporting any intervention. RPA handled certificate reminders and matched standard invoices; exceptions were routed to staff.

For a strategic group of suppliers, Northstar created quarterly collaboration reviews using shared forecasts and capacity information. Cybersecurity requirements were incorporated into supplier agreements, and access to the portal was protected through multi-factor authentication.

Outcome

The organisation achieved better visibility of contract obligations and reduced manual follow-up. More importantly, supplier discussions became evidence-based: performance reviews could distinguish isolated incidents from persistent patterns. The company decided not to pursue autonomous supplier selection until data quality, model validation and governance had reached a higher level.

Learning points

The case illustrates that digital transformation should begin with process and data readiness. AI is more valuable when embedded in a governed decision process than when introduced as a replacement for procurement judgement. It also shows that supplier platforms create value when they enable collaboration, not only when they collect compliance information.

Autonomous Sourcing and Emerging Supply Ecosystems

Autonomous sourcing refers to systems that can execute parts of the sourcing process with limited human intervention. In a mature form, a system might monitor demand, scan approved markets, identify qualified suppliers, compare total-cost scenarios, conduct rule-based negotiations and issue recommendations or orders. Fully autonomous sourcing remains constrained by uncertain information, legal accountability, strategic relationships, ethical considerations and the consequences of error.

The more plausible near-term model is supervised autonomy. Machines handle high-volume pattern recognition and routine transactions, while people set objectives, constraints, risk tolerances and escalation rules. Procurement professionals will increasingly act as orchestrators of data, markets, technology and relationships.

Emerging supply ecosystems may include manufacturers, logistics providers, platforms, financial institutions, technology firms, regulators and customers exchanging data through interoperable networks. This can improve responsiveness but may also create concentration risk. Dependence on one platform, cloud provider or data intermediary can reduce strategic flexibility. Ecosystem strategy must therefore address portability, common standards, access rights and exit arrangements.

Sustainability is also becoming more data-intensive. Digital tools can support emissions estimation, traceability, circular-economy monitoring and supplier due diligence. Yet environmental and social claims require credible measurement and verification. Technology should support accountability, not provide a superficial appearance of responsible sourcing.

Common Failure Modes and Best Practices

Several failures recur in procurement technology programmes. Organisations may select software before defining the problem, automate unstable processes, underestimate supplier enablement, treat dashboards as performance management, ignore integration costs or assume that a vendor’s model is inherently objective. Another mistake is to define success solely as licence adoption rather than improved decisions and outcomes.

Effective practice includes the following:

• Establish a clear business case and measurable baseline before implementation.
• Treat supplier and contract master data as governed organisational assets.
• Design around user and supplier journeys, not only internal control requirements.
• Use risk-based automation and reserve human judgement for ambiguous or high-impact decisions.
• Require explainability, audit trails, override procedures and model monitoring for AI applications.
• Integrate cybersecurity and privacy from the design stage.
• Pilot with representative users and suppliers, including smaller suppliers with limited digital capability.
• Measure adoption, data quality, process performance, supplier outcomes and unintended effects.
• Review technology periodically because models, threats, regulations and supply conditions change.

Chapter Summary

Digital procurement connects sourcing, contracting, purchasing, payment and supplier collaboration through integrated platforms, data and workflows. E-procurement suites improve control and visibility; marketplaces expand access to suppliers; supplier platforms support shared information and performance management; and CLM systems connect negotiated commitments with operational execution.

RPA is useful for repetitive, rule-based work, while analytics progresses from describing past events to predicting risks and recommending actions. Digital twins can support supply-network scenario analysis, and blockchain may provide trusted records where distributed validation is justified. Predictive and generative AI can assist classification, forecasting, contract analysis, communication and decision support, but their value depends on reliable data, explainability, cybersecurity and human accountability.

Procurement transformation is therefore socio-technical. It requires process redesign, data governance, change leadership, supplier participation and risk-based controls. The future is more likely to involve supervised autonomy than the complete removal of human judgement. Procurement leaders must ensure that digital systems strengthen resilience, fairness, collaboration and strategic value rather than merely increasing transaction speed.

Key Terms

Digital procurement: The use of connected digital technologies, data and workflows to manage procurement activities and decisions.

E-procurement suite: An integrated system supporting electronic purchasing processes such as requisitions, orders, receipts and invoices.

Source-to-pay: The complete process from sourcing and contracting through purchasing, receipt and payment.

Source-to-contract: Activities involving category strategy, supplier discovery, tendering, negotiation and contract creation.

Supplier network: A digital platform connecting buying organisations and suppliers for transactions, information exchange and collaboration.

Contract lifecycle management: The digital management of contracts from request and drafting through execution, obligations, amendment and expiry.

Robotic process automation: Software-based automation of repetitive, structured and rule-governed tasks.

Predictive analytics: Analysis that estimates the likelihood of future events or outcomes.

Prescriptive analytics: Analysis that recommends possible actions based on data and models.

Digital twin: A data-enabled digital representation of a physical asset, process or network used for monitoring and simulation.

Blockchain: A distributed ledger that records validated transactions across a network.

Generative AI: Artificial intelligence that produces new content such as text, summaries, scenarios or code.

Explainability: The extent to which users can understand the basis, inputs and limitations of an AI output.

Human oversight: Meaningful human authority to review, challenge, correct or override an automated result.

Master data: Core organisational data, such as supplier, item, category and contract records, used across systems.

Maverick spending: Purchases made outside approved suppliers, contracts or procurement procedures.

Interoperability: The ability of different systems and organisations to exchange and use information effectively.

Supervised autonomy: A model in which automated systems perform defined tasks while humans establish boundaries and oversee consequential decisions.

Review Questions

Short Answer Questions

1. What is the difference between source-to-pay and source-to-contract?

2. Why is supplier master data important for procurement analytics?

3. Identify two suitable applications of robotic process automation in procurement.

4. What is meant by explainability in an AI-enabled procurement decision?

5. Why may a digital twin produce misleading results if its assumptions are hidden?

Long Answer Questions

1. Explain the architecture of a digital procurement ecosystem and discuss the importance of integration among its layers.

2. Compare e-procurement suites, supplier platforms and contract lifecycle management systems in terms of their procurement and SRM contributions.

3. Evaluate the opportunities and limitations of predictive and generative AI in sourcing and supplier relationship management.

4. Discuss a governance framework for managing cybersecurity, privacy, data quality and human oversight in digital procurement.

5. Explain why procurement transformation should be treated as an organisational change programme rather than simply an information technology project.

Analytical / Application Questions

1. An organisation has accurate invoice data but inconsistent supplier names, missing certificates and disconnected contract records. Recommend a sequence of digital transformation initiatives and justify the order.

2. An AI model identifies a small regional supplier as high risk because it has limited historical data. How should the procurement team investigate, govern and respond to the recommendation?

3. A marketplace promises rapid savings but requires suppliers to pay substantial platform fees and stores all transaction data under the platform owner’s control. Analyse the strategic, ethical and operational issues before adoption.

22. Multiple Choice Questions with Answers

Question 1

Which feature most directly connects negotiated contractual commitments with operational follow-up?

A. A digital marketplace search ranking

B. Contract lifecycle management

C. Robotic invoice data entry

D. A supplier risk heat map

Correct Answer: B. Contract lifecycle management

Question 2

A procurement team wants to automate certificate reminders and standard invoice matching while routing unusual cases to employees. This is best described as:

A. Risk-based intelligent automation

B. Fully autonomous sourcing

C. Blockchain governance

D. Manual supplier development

Correct Answer: A. Risk-based intelligent automation

Question 3

Why can a supplier dashboard based on a single composite score be misleading?

A. Composite scores cannot be calculated digitally

B. It may conceal important differences among performance dimensions

C. Suppliers cannot access dashboards

D. It eliminates the need for performance data

Correct Answer: B. It may conceal important differences among performance dimensions

Question 4

Which condition most strongly supports the use of a digital twin in supply management?

A. A simple, stable process with no interdependencies

B. A complex network requiring scenario analysis

C. A purchase requiring only one approval

D. A contract containing no operational obligations

Correct Answer: B. A complex network requiring scenario analysis

Question 5

What is a fundamental limitation of blockchain-based traceability?

A. It cannot record transactions permanently

B. It prevents all participants from viewing records

C. It does not guarantee that an original input was truthful

D. It can only be used for financial payments

Correct Answer: C. It does not guarantee that an original input was truthful

Question 6

Which practice best demonstrates meaningful human oversight of an AI supplier recommendation?

A. Automatically accepting every model output

B. Allowing a reviewer to approve without seeing evidence

C. Providing authority, evidence and procedures to challenge or override the output

D. Removing all historical data from the model

Correct Answer: C. Providing authority, evidence and procedures to challenge or override the output

Question 7

Northstar’s fictional case delayed autonomous supplier selection primarily because:

A. Its suppliers refused to use electronic communication

B. Its data was incomplete and inconsistent, weakening model reliability

C. Its contracts had no renewal dates

D. Its ERP system could not process invoices

Correct Answer: B. Its data was incomplete and inconsistent, weakening model reliability

Question 8

Which security control most directly reduces the risk of unauthorised supplier-platform access?

A. A larger product catalogue

B. Multi-factor authentication and least-privilege access

C. A higher composite supplier score

D. A longer purchase order description

Correct Answer: B. Multi-factor authentication and least-privilege access

Question 9

A model recommends reallocating volume after predicting late delivery. This output is an example of:

A. Descriptive reporting only

B. Prescriptive decision support informed by predictive analytics

C. A blockchain transaction

D. Contract execution by electronic signature

Correct Answer: B. Prescriptive decision support informed by predictive analytics

Question 10

What is the most appropriate starting point for selecting procurement technology?

A. Choosing the platform with the most advanced features

B. Defining the business problem, baseline and desired outcomes

C. Automating every existing approval immediately

D. Replacing supplier meetings with dashboards

Correct Answer: B. Defining the business problem, baseline and desired outcomes