Theory 01
Digitalization: A Strategic Transformation
Origin: Iansiti & Lakhani (2020); IEB Session 1 theme
Iansiti & Lakhani (2020) differentiate between digitization (converting analog info to digital) and digitalization (re-architecting the firm around digital logic). Digitalization isn't about adding IT to a traditional firm; it's about ending the reliance on human labor at the core of the operating model.
Digitization
Converting analog information into digital bits. (e.g., scanning a paper invoice). This is a technical process, not a strategic one.
Digitalization
The transformation of business processes and organizational structures. Redesigning value creation around digital scalability.
Operating Model
The set of processes, systems, and people that deliver value. Digital firms replace human bottlenecks with algorithmic "factories".
Datafication
Turning activities into data. If you can't measure it, you can't optimize it. Every interaction becomes a learning signal.
Theory 02
The Strategic Collision
Origin: Iansiti & Lakhani (2020)
A Collision occurs when a firm with a digital operating model enters a space traditionally served by conventional firms. Because digital models exhibit increasing returns to scale, scope, and learning, they eventually overwhelm traditional players.
Why Digital Firms Win
According to Iansiti & Lakhani, the advantage comes from three interconnected dynamics:
- Scale: Software can be replicated at near-zero marginal cost.
- Scope: Digital data can be reused to enter new categories (e.g., Uber moving from rides to food).
- Learning: Algorithms improve automatically with more usage data.
Theory 03
AI as Prediction
Origin: Agrawal, Gans & Goldfarb (2018)
The "Economics of AI" (Agrawal et al., 2018) argues that AI is essentially a drop in the cost of prediction. When prediction becomes cheap, it is used more frequently, and the value of human judgment and data increases.
Prediction
Filling in missing information. AI uses data to predict outcomes, lowering the cost of uncertainty.
Judgment
Human role of assigning value to outcomes. AI says what will happen; humans decide if we want it.
Automation
When the cost of prediction drops enough, it replaces manual decision-making tasks entirely.
Feedback
The output of one decision becomes the input (data) for the next prediction, creating a learning loop.
Theory 04
Data as a Strategic Asset
Origin: Iansiti & Lakhani (2020); Session 1 dashboard theme
Data becomes strategically valuable when it is connected to decisions, feedback, and learning. A firm does not gain advantage simply by storing data. It gains advantage when data is proprietary, relevant, high quality, difficult to imitate, and embedded in processes that repeatedly improve products or decisions.
| Data Concept | Meaning | Strategic Implication |
|---|---|---|
| Volume | Amount of observations available for training, testing, or optimization. | Large datasets can improve prediction, but only if relevant and clean. |
| Variety | Different forms of data: text, images, transactions, sensor data, clinical records. | Broader data can create scope economies across products and functions. |
| Velocity | How quickly data is generated and fed back into decisions. | Fast feedback loops let digital firms learn faster than traditional rivals. |
| Validity | Whether data measures the phenomenon that matters. | Bad proxies can optimize the wrong objective and destroy value. |
| Exclusivity | Whether competitors can access equivalent data. | Proprietary data can become a durable moat if linked to learning effects. |
Data Is Not Automatically a Resource-Based Advantage
Data must be valuable, rare, difficult to imitate, and organizationally embedded to produce advantage. Public data can support AI, but it rarely creates a defensible moat on its own. Firm-specific operational, customer, or scientific data is more strategically important.
Theory 05
The Digital Operating Model and the AI Factory
Origin: Iansiti & Lakhani (2020)
The central distinction in Iansiti & Lakhani is between a traditional operating model and a digital operating model. Traditional firms scale by adding people, assets, managerial layers, and coordination routines. Digital firms scale by using software, data, algorithms, APIs, and automated decision systems.
Data Pipeline
Captures, cleans, stores, and makes data available for decision systems.
Algorithms
Turn data into predictions, classifications, recommendations, or automated actions.
Experimentation
A/B tests, pilots, and feedback loops show which digital interventions improve outcomes.
Software Scaling
Once encoded, the process can be replicated across users and contexts at low marginal cost.
Why This Changes Management
The bottleneck shifts from human execution to system design. Managers must ask: what decisions should be automated, what data feeds them, where human judgment remains necessary, and how the organization learns from outcomes.
Theory 06
The Digital Moat: Scale, Scope, and Learning
Origin: Iansiti & Lakhani (2020)
A digital moat is created when scale, scope, and learning reinforce one another. Scale produces more interactions. More interactions produce more data. More data improves learning. Better learning improves products, which increases scale again. The resulting loop can make digital leaders difficult to displace.
| Moat Driver | Definition | Risk for Incumbents |
|---|---|---|
| Scale | Serving more users with relatively low marginal cost. | Traditional firms may be constrained by human-heavy processes. |
| Scope | Reusing data, software, and digital infrastructure across adjacent markets. | Digital entrants can cross industry boundaries faster than expected. |
| Learning | Improving through feedback, experimentation, and algorithmic refinement. | Firms without feedback loops learn more slowly even if they have strong assets. |
Case Application
Novo Nordisk — From Pharma to AI Factory?
Applying Session 01 logic to your synopsis on Novo Nordisk and OpenAI:
| Concept | Application |
|---|---|
| Collision | Novo is preempting a collision by partnering with OpenAI. They are moving before an "AI-first" biotech disintermediates them. |
| Prediction | Drug discovery is a massive prediction task. AI lowers the cost of identifying which molecules will be successful. |
| Learning Loop | Clinical trial data is Novo's "proprietary moat". By feeding this into OpenAI's models, they create a faster learning loop than rivals. |
Exam Preparation
Likely Oral Exam Questions
-
Core What is the difference between digitization and digitalization according to the course? ▶
- Digitization: Purely technical. Converting analog to digital. Does not change the business logic.
- Digitalization: Strategic. Re-architecting the organization. Changing how value is created and captured using digital logic.
- Use the Airbnb vs. Marriott example: Marriott digitized (online booking); Airbnb digitalized (platform-based operating model).
-
Core Explain the "Collision" framework and why digital firms eventually win. ▶
- Collision = Digital model vs. Traditional model.
- Traditional firms have diminishing returns (it gets harder to scale human-heavy ops).
- Digital firms have increasing returns (scale, scope, and learning effects).
- The digital firm might have lower performance initially, but its growth trajectory is exponential.
-
Synthesis How do scale, scope, and learning create a digital moat? ▶
- Scale creates more interactions at low marginal cost.
- Scope lets the firm reuse data and infrastructure across adjacent activities.
- Learning improves predictions and processes through feedback loops.
- Together, they form a self-reinforcing loop that traditional firms struggle to match.
-
Apply Why is data not automatically a source of competitive advantage? ▶
- Data must be relevant, high quality, and connected to decisions.
- If competitors can buy or copy similar data, it is not rare.
- Advantage comes when data is proprietary and embedded in a learning loop that improves products or operations.
-
Synthesis How does the "AI as Prediction" theory relate to Novo Nordisk's strategy? ▶
- Drug discovery is essentially predicting biological success.
- By using OpenAI, Novo lowers the cost of these predictions.
- This shifts the value to Novo's Judgment (deciding which diseases to target) and their Data (proprietary clinical trial results).