Andreas Ramos | Professor of Digital Marketing

Business Models, Metrics, and KPIs

By Andreas Ramos, July 7th, 2026

Summary: Metrics function as reductionist abstractions to model complex organizational realities. The core thesis establishes that unlinked quantitative measurement systems cause systemic degradation unless leadership applies the relevant metrics. Misaligned incentives produce systemic collapse and unanchored compensation structures generate catastrophic collapse. Wells Fargo paid US$3 billion dollars in fines and terminated 5,300 employees in 2020 after staff generated 4 million fraudulent accounts to satisfy isolated quotas.

The analysis classifies business systems into specific measurement models:

  • Digital Recurring systems (Salesforce, Microsoft, Adobe) mandate continuous event driven measurement of Monthly Recurring Revenue (MRR), Customer Acquisition Cost (CAC), and Net Revenue Retention (NRR).
  • High Volume Omnichannel operations (Walmart, Target, Costco) deploy mixed velocity tracking to target Net Promoter Score (NPS) and Marketing Mix Modeling (MMM).
  • Low Velocity Enterprise frameworks (Lockheed Martin, Boeing, McKinsey) monitor pipeline velocity and Total Contract Value (TCV) across extended sales cycles.
  • Capital Intensive Infrastructures (AT&T, Equinix) track Capacity Utilization Rates and Overall Equipment Effectiveness (OEE) against rigid reporting intervals.

Let's start by looking at the types of metrics and their application, followed by frameworks for metrics. We then look at the manipulation of metrics and the inherent problems in metrics. The article ends by offering a solution.

The Types of Metrics

Metrics and Key Performance Indicators (KPIs) are the bedrock of modern organizational management:

  • "If you can’t measure it, you can’t manage it" (Peter Drucker, The Practice of Management, 1954).
  • "When you can measure what you are speaking about, and express it in numbers, you know something about it... otherwise your knowledge is of a meagre and unsatisfactory kind." (Lord Kelvin, Popular Lectures and Addresses, 1889).

Metrics must be designed around the business model, its data asset velocity, and its channel composition. This can be broken down into four types of organizations.

Digital Recurring & Product-Led Growth (PLG)

  • Business Models: Software-as-a-Service (SaaS), digital infrastructure subscriptions, and high-margin recurring utility contracts. Examples include Salesforce, Adobe, Oracle, Microsoft, HubSpot.
  • Data Asset Velocity: High-velocity telemetry requires consumption-based metrics. Real-time and continuous. Event-driven telemetry measures active user behavior, compute usage, and in-app triggers into databases.
  • Channel Composition: Pure digital performance networks. Customer acquisition is driven through paid search, social media, and email.
  • KPIs include: Revenue Retention (NRR); Net Revenue Retention (NRR), Monthly Recurring Revenue (MRR) Churn; Target Cost per Action (tCPA); Customer Acquisition Cost (CAC), Customer Lifetime Value to Customer Acquisition Cost ratio (LTV:CAC); Product Usage Velocity, Sales Qualified Leads (SQL), SQL-to-orders, and similar.

High-Volume Fragmented Omnichannel Models

  • Business Models: High-volume retail, transactional consumer products, and hybrid digital-physical commerce operations. Examples include Walmart, Target, Costco, Home Depot, Best Buy, Nike.
  • Data Asset Velocity: Mixed velocity. Front-end point-of-sale transactions are immediate. Back-end inventory tracking, supply chain metrics, and logistics adjustments operate on batch-delayed schedules.
  • Channel Composition: Decentralized and unlinked. Heavy reliance on integration of offline mass media (television, radio, print, major events) with localized digital performance networks. Direct click-to-sale attribution is impossible.
  • KPIs include: Customer Satisfaction (NPS), Brand Recall, and Brand Sentiment via Marketing Mix Modeling (MMM); Inventory Shrinkage rates; Return on Invested Capital (ROIC) per physical footprint.

Low-Velocity High-Contract Enterprise Models

  • Business Models: Consultative B2B services, aerospace and defense procurement, specialized medical hardware, and multi-million-dollar annual contract values (ACV). Examples include Lockheed Martin, Boeing, Northrop Grumman, McKinsey, Accenture, Siemens.
  • Data Asset Velocity: Low velocity. Data points are milestone driven. Operational metrics are entered in the CRM over 6-to-18-month sales cycles.
  • Channel Composition: Account-Based Marketing (ABM). Hyper-targeted executive outreach, direct enterprise sales pipelines, and industry conferences.
  • KPIs include: Pipeline Velocity; Sales Qualified Lead (SQL) to Opportunity conversion efficiency; Account Engagement Scores; Total Contract Value (TCV) Expansion Rate; Economic Profit per account.

Capital-Intensive Infrastructure Operations

  • Business Models: Datacenters, telecommunications grids, utility networks, and automated fulfillment hubs. Examples include AT&T, Equinix, Duke Energy, Digital Realty, American Tower.
  • Data Asset Velocity: Continuous operational telemetry coupled with low-frequency financial accounting. Millions of sensor packets monitor power loads and cooling efficiency. Financial performance metrics operate on rigid quarterly and annual reporting cycles.
  • Channel Composition: Institutional procurement contracts, sovereign joint ventures, and multi-year service level agreements (SLAs).
  • KPIs include: Return on Invested Capital (ROIC) relative to the cost of capital; Capacity Utilization Rate; Opex-to-Revenue Ratio; Equipment Depreciation vs. Realized Margin; Overall Equipment Effectiveness (OEE).

However, many organizations are a blend of models. For example, an automotive manufacturer (Infrastructure Operations) that also offers autonomous driving via over-the-air updates is in both the Capital-Intensive quadrant and the Digital Recurring/PLG quadrant.

Frameworks for Measurement

The Balanced Scorecard Architecture

Developed by Robert S. Kaplan and David P. Norton in the Harvard Business Review (1992), the Balanced Scorecard is a system for large corporations to track strategic elements:

  • Financial Perspective: Evaluates historical economic health to ensure operational improvements translate into financial returns and avoid the trap of metrics that fail to impact the bottom line.
  • Customer Perspective: Measures of how the market perceives the organization with metrics such as customer satisfaction, net promoter score (NPS), and service quality.
  • Internal Business Processes: Tracks the efficiency of internal operations, ensuring they support customer satisfaction and financial objectives.
  • Innovation and Learning: Serves as the foundation for long-term survival, assessing the corporation's capacity to develop new competencies, train staff, and innovate.

McKinsey's Performance and Health Matrix

McKinsey looks at the historical performance and future health of the organization:

  • Long-Term Value Drivers: The true drivers of enterprise value are not short-term earnings per share, but rather long-term revenue growth and Return on Invested Capital (ROIC) relative to the cost of capital.
  • Historical Performance Metrics: Corporations assess the economic value they have already created by using financial statements and metrics such as economic profit and ROIC.
  • Future Health Indicators: To gauge the ability to generate future economic value, enterprises use productivity metrics covering sales, operating costs, and capital efficiency to track capital utilization.

KPI Systemic Integration

Organizations must use daily metrics that support their strategy:

  • Cause-and-Effect Linkage: As outlined in Harvard Business School execution strategies, every KPI must map upward; learning points to process, process to customer, and customer to financial.
  • Critical Performance Variables: Metrics must be restricted to those variables that a failure to deliver on them would cause the entire corporate strategy to collapse.
  • Limitation of Variables: Prevent information overload by limiting the number of metrics.

Enterprise Value Measurement Hierarchy

A layered pyramid framework for metrics starts with the base of "Innovation and Learning KPIs" feeding upward into "Internal Process Metrics," which subsequently drive "Customer Perspective Data." The apex represents "Financial Returns," specifically Return on Invested Capital (ROIC) and long-term economic profit.

A similar approach is Return on Invested Capital (ROIC) driver tree methodology. ROIC links top-level strategic value drivers to specific, P&L-relevant operational KPIs. This directly solves the KPI Overload problem by ensuring every metric has a traceable connection to enterprise value.

Improve Metrics with SMART Framework

The SMART framework is a guide to developing metrics. SMART = Specific: Define the goal. Measurable: Use quantitative metrics. Attainable: The goal is possible. Relevant: The goal is relevant to the business goals. Time-bound: Set a deadline for the goal.

Marketing Mix Modeling (MMM)

Major corporations often have tens of thousands of products and hundreds of marketing teams across dozens of countries and languages. Teams far from the Central Office often ignore corporate standards, use local methods, or make mistakes in implementation of tracking systems. Large corporations also use offline channels such as TV, radio, print, and events that make digital tracking impossible. These organizations use Marketing Mix Modeling (MMM) to measure their marketing. They hire research firms to survey thousands of consumers on a regular schedule to measure brand awareness, recall, sentiment, and similar criteria. MMM questions include:

  • Where have you recently seen our brand? (radio, TV, billboard, social media?)
  • Which brand comes to mind when thinking of (examples include hamburgers, drinks, soap, cars)?
  • How would you describe our brand personality?
  • How do you feel our brand compares to our top competitor?
  • How likely are you to purchase in the next 30 days?
  • What is the main reason you choose a competitor over us?
  • Which features matter most when selecting this type of product?
  • How likely are you to recommend our product to a friend?

Note: People may offer wrong replies. For example, 16% may say they saw the ads on TV. There were no TV ads.

Inherent Problems with Metrics

Epistemological Nature of Measurement

Epistemology is the study of the nature and limits of knowledge. How do we know what we know? What is the nature of what we know? What are the limits to what we know?

Metrics and KPIs are reductionist abstractions to simplify complex realities. In plain language, metrics try to describe what we know. Since reality is complex, metrics create a simplified model of the organization. If we can manage the model, we can manage the complex reality.

Not understanding the epistemic limitations of metrics leads to problems. Here is a list of metrics failures: Operational degradation, behavioral distortion, deliberate fraud, degradation of data, failure in recursive compliance, and epistemic blindness.

Operational Breakdown and Systemic Degradation

  • The Shift to Localized Focus: In the crush of day-to-day operations, humans often lose sight of long-term goals. Instead, they focus on managing and optimizing the metrics within their area of responsibility.
  • Incentivized Manipulation: Some actors realize they can improve their compensation and bonuses by manipulating the metrics. This results in double booking, channel stuffing, and more.
  • Bureaucratic Distance: Distance between offices will degrade mission loyalty in large corporations.
  • Organizational Silo: The sense of departmental survival also leads to fraud.

Behavioral Distortions

  • Goodhart's Law: When a measure becomes a target, it ceases to be a good measure. Humans optimize their workflow to satisfy the metric. This ignores the objective the metric was designed to create. (Goodhart, C.A.E., Problems of Monetary Management, Papers in Monetary Economics, 1975.) (Note: Marilyn Strathern summarized Goodhart’s Law in 1997, based on his 1975 paper.)
  • Campbell's Law: Quantitative indicators for social decision-making are vulnerable to corruption. This alters and damages the social processes they are intended to monitor. (Campbell, D.T., Assessing the Impact of Planned Social Change, Evaluation Studies Review Annual, 1976.)
  • The Cobra Effect (Perverse Incentives): A metric may actively worsen the problem. This happens because the metric rewards outputs, not the goals. (Siebert, H., Der Kobra-Effekt, Deutsche Verlags-Anstalt, 2001.)

If a sales person gets a bonus for setting appointments, it’s a short step to creating fake appointments to get the bonus. When the client doesn’t show up, well, that was the client’s decision. For example, starting in 2002, Wells-Fargo gave staff a bonus for every new bank account. Staff created 3,500,000 bank accounts, 500,000 credit card accounts, and many other types of accounts. Wells-Fargo fired 5,300 staffers and paid nearly US$3 billion in fines and lawsuits in 2020.

Deliberate and Fraudulent Manipulations

  • Cooking the Books (Earnings Management): The falsification or manipulation of data creates a false reality. (Healy, P.M. and Wahlen, J.M., A Review of the Earnings Management Literature, Accounting Horizons, 1999.)
  • Creaming: Creaming is the exclusion of difficult cases to artificially inflate success rates. A surgeon has a 98% success rate because he only accepts easy cases. This is difficult to uncover because the bad metrics aren’t counted.
  • Double Booking: The fraudulent practice of recording the same revenue transaction or physical asset across multiple periods, subsidiaries, or financial accounts to artificially multiply the performance.
  • Channel Stuffing: The inflation of short-term sales metrics by forcing excess inventory down the distribution channel faster than downstream markets can absorb it. This pulls future revenue forward while manufacturing a looming inventory crisis. (SEC Enforcement Action, In the Matter of Bristol-Myers Squibb Company, 2004.)
  • Inventory Shrinkage: Loss, theft, or damage is hidden from the data. This creates a false expectation of revenues and profits while masking asset destruction.
  • Competition: Competition between people or teams may result in sabotage of projects (shredding documents, erasing data, stealing computers), blocking projects by other teams, harassment (in many forms), and other negative activities.
  • Manipulation by Management: Managers can override internal controls. The metrics are useless if the system is manipulated by executives.
  • Top-Down Sabotage (TDS): Along with bottom-up fraud (fraud by lower-level staff), there is downward fraud by upper-level staff. They may sabotage or suppress high-ability subordinates to protect their status, position, compensation, and prevent future competition.

Data Integrity and Structural Failures

  • Survivorship Bias: The logical error of focusing exclusively on successful outcomes within a dataset while ignoring failures. This skews performance metrics and creates false precedents for scalability. (Wald, A., A Method of Estimating Plane Vulnerability, Center for Naval Analyses, 1943.)
  • Selection Bias: The distortion of metrics by data collection methods that favor or ignore demographics or behaviors. The performance indicators will not represent operational reality.
  • Bad Data: Caused by broken telemetry, corrupt tracking, or human errors. Agents may game a metric without triggering an anomaly alert. This inputs noise into the metrics system.
  • The 10/80/10 Framework: This assigns (for example) 10% for strategic alignment, 80% for execution, and 10% for review. However, there is widespread execution failure in corporate governance. Analysis of 20,500 strategic plans indicates that 67% of KPIs lacked an assigned owner, rendering them useless or accelerating behavioral distortion and fraud. (Ron Carucci, Executives Fail to Execute Strategy Because They’re Too Internally Focused, Harvard Business Review, November 13, 2017).

The Recursive Compliance Failure

  • Higher-Level Metrics Corruption: Implementation of advanced metrics to detect or prevent corrupted metrics will lead staff to develop a higher layer of corrupt metrics. (Cressey, D. R., Other People's Money, 1953).
  • Gaming the Metrics: The staff will shift from gaming the primary operational metrics to actively gaming the compliance metrics and auditing metrics. The manipulation becomes extremely difficult to detect, especially if upper management people are involved.
  • Take Over the Metrics System: The staff take over the teams that apply ethics and compliance. Instead of gaming the metrics, they game the system. This happens in police departments when corruption switches from low-level bribery to manipulation of Internal Affairs.

Epistemic Blindness

Metrics is a form of reductionist knowledge. We create a simplified version that we can understand in order to understand and manage the original version.

Not understanding the epistemic limitations of metrics leads to blindness about internal manipulation or external reality. For example, relying on historical data and metrics that worked well in the past leaves the organization vulnerable to sudden technological or macroeconomic changes that the metrics were not designed to track. The organization may see good metrics but doesn’t see external conditions have changed until it's too late.

The conversion of information into data also strips context, history, and meaning from the data, which leads to precisely wrong tunnel vision.

There is also the problem of not understanding the metrics. The sample size, how avoid regression to the mean, how to tell if there is meaningful difference betwee data points, and other issues may cause mistakes to be made, because people don't understand the data.

Metrics Beyond Business

The problem of fraud isn’t just in business. Yuen Yuen Ang’s China’s Gilded Age (Cambridge University Press, 2020) shows as countries develop, they go through stages of corruption. She classifies these as Speed Money (low-level payments to accelerate services, such as installation of city services or speeding tickets), Petty Theft (stealing office supplies for resale), Access Money (paying bribes for access), and Grand Theft (paying millions of dollars for lucrative state projects).

Corruption starts as the lowest form and evolves as the country grows. Paying a small bribe to a cop to avoid a traffic ticket is one thing. Paying US$30 million to lobbyists and public relations (PR) firms who make campaign donations to secure favorable legislation (or prevent unfavorable legislation) is the respectable work of politicians and governments worldwide. They also create fake boards of experts or committees of concerned moms, along with reports and data, to show cigarettes are healthy or coal is clean. That’s more fake metrics.

Incentives and Metrics

The problem with metrics is the incentives. There are two kinds of incentives: extrinsic versus intrinsic:

  • Extrinsic Incentives: The attachment of external consequences (such as rewards, compensation, bonus, job security, etc.) to metrics.
  • Intrinsic Incentives: The cultivation of internal motivation through autonomy, peer recognition, and professional ethics.

Extrinsic incentives (rewards) introduces Principal-Agent Theory (better said, Principal versus Agent Theory). This is the conflict of interest between principals (the management, owners, investors, etc.) versus the agents (the staff, workers, etc.). Principals optimize for the organization’s long-term goals and survival, but agents optimize for short-term personal compensation and day-to-day survival in their jobs. Metrics manipulation becomes inevitable due to this mismatch in incentives (per Goodhart's Law and Campbell's Law). (Jensen, M. C., & Meckling, W. H., Theory of the firm: Managerial Behavior, Agency Costs and Ownership Structure, Journal of Financial Economics, 1976).

Intrinsic incentives motivate actors per Stewardship Theory, where they behave as stewards of the organization. Their incentives are autonomy, peer recognition, and professional values to collaborate to uncover failures and improve the system. This aligns their objectives with the principal's long-term goals. Instead of monitoring and punitive compliance, the principals create high-autonomy, diagnostic peer-review environments. Stewardship Theory is the opposite of Principal-Agent Theory.

For example, the US Federal Aviation Administration's Aviation Safety Reporting System (ASRS) is based on diagnostic peer review. Agents are encouraged to actively expose critical failures in order to improve the system. For example, aircraft pilots may experience near-misses, cognitive overload, or mechanical anomalies. If a report leads to penalties, license suspension, or termination, the pilot will not report the problem. Flaws remain hidden until catastrophic failure occurs. ASRS guarantees legal immunity to pilots who self-report non-criminal errors within 24 hours. Because their jobs are secured, pilots aggressively document their failures. The reports are anonymized, aggregated, and distributed to peer committees, which analyze the reports and data to redesign cockpit ergonomics and rewrite landing protocols, which reduces fatalities. (ASRS at https://asrs.arc.nasa.gov/)

Organizations that use extrinsic incentives can consider a switch to intrinsic stewardship. The organization can look into how to apply the FAA’s ASRS model.

If the organization must use extrinsic incentives, it can implement processes to reduce fraud:

  • Verification by independent third-party forensic auditors. The forensic auditing report should be available to all.
  • Design counter-metrics that automatically degrade if a primary metric is artificially inflated (e.g., pairing high-velocity sales targets with delayed churn-rate penalties).
  • Apply forensic auditing to all levels: vendors, service suppliers, contractors, staff, managers, directors, VPs, C-level, and the board.
  • All metrics, scores, rewards, and so on should be open for review by all.

Don’t assume third-party forensic auditing will prevent fraud. Multi-million dollar auditing fees can incite the auditors to fraud, such as phantom assets, round-tripping, executive embezzlement, or stock manipulation. See examples by searching for “major instances of fraud or collusion by independent third-party forensic auditors" and you'll find Arthur Andersen & Enron (2001), PricewaterhouseCoopers (PwC) & Satyam (2009), Ernst & Young (EY) & Wirecard (2020), or KPMG & Gupta State Capture (2017).

Your Comments and Ideas

Let me know what you think of this. Send me an email: andreas@andreas.com.