iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Entity propensity models (EPMs) estimate how likely a particular person, organization, asset, or other entity is to experience an outcome or take an action. Better predictions could help organizations target interventions, allocate resources, and manage risk more efficiently—but the March 11, 2025 article that popularizes this framing does not report measured EPM-specific savings, revenue gains, or return on investment. The economic case is plausible; it still has to be demonstrated against a credible baseline.
What is an entity propensity model?
Bill Schmarzo defines an EPM as “a predictive analytic profile that quantifies an entity’s likelihood of a specific outcome or behavior.” In practical terms, a model might estimate a patient’s risk of a hospital-acquired infection, a student’s likelihood of dropping out, a technician’s chance of resolving an issue on the first visit, or a machine’s chance of breaking down. These are examples in Schmarzo’s article, not independently validated deployments. Schmarzo’s article
The word “entity” does not mean that an EPM matches duplicate records. That task is entity resolution: determining whether separate records refer to the same real-world person, business, or establishment, as the U.S. Census Bureau’s 2024 Federal CASIC Workshops archive explains.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Scan for outdated or missing drivers - takes under a minute3Clear out junk files and repair common Windows errorsAn EPM also does not, by itself, estimate the causal effect of an intervention. A prediction that someone is likely to buy or a machine is likely to fail is not proof that a marketing offer will cause a purchase or that preventive maintenance will avert a breakdown. That distinction matters economically: a high-propensity entity may act without intervention, while a lower-propensity one may be more persuadable or more preventable.
#1 Best Overall
How could EPMs create economic value?
The proposed value comes from using more specific predictions to make a decision differently from the existing process. An organization might prioritize outreach to entities at higher risk, direct scarce staff or equipment to where they are most useful, or intervene before a costly failure. Schmarzo also suggests reusing and updating profiles as new data arrives. These are potential mechanisms, not quantified results from EPM deployments.
For a prediction to create value, it must connect to an available action whose benefits exceed its costs and harms. A model can be accurate yet economically useless if its output arrives too late, does not change a decision, or identifies cases for which no effective intervention exists. It can also increase costs if false alarms prompt unnecessary treatment, maintenance, or customer contact.
Rank #2
- Use scikit-learn to track an example ML project end to end
- Explore several models, including support vector machines, decision trees, random forests, and ensemble methods
- Exploit unsupervised learning techniques such as dimensionality reduction, clustering, and anomaly detection
- Dive into neural net architectures, including convolutional nets, recurrent nets, generative adversarial networks, autoencoders, diffusion models, and transformers
- Use TensorFlow and Keras to build and train neural nets for computer vision, natural language processing, generative models, and deep reinforcement learning
- Targeted action: Focus an intervention where it is expected to produce more benefit than a broad or untargeted approach.
- Risk and prevention: Identify possible adverse outcomes early enough to take a useful, proportionate action.
- Resource allocation: Match staff, inventory, equipment, or attention to forecast need.
- Operational planning: Use predicted demand, resolution likelihood, or equipment condition to inform scheduling and workflow.
These mechanisms should be treated as hypotheses to test, not automatic savings. The UK Government Digital Service and Department for Science, Innovation and Technology’s Digital and Data Benefits framework, published April 7, 2026, offers a general context for quantifying benefits of digital and data programs; it is not an EPM outcome study.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →What does the available economic evidence show?
The March 11, 2025 article argues that individualized prediction could improve targeting, risk assessment, resource use, and operations, but it does not provide a measured EPM-specific dollar benefit, percentage gain, controlled deployment result, or ROI. The adjacent figures sometimes used to support data-driven decision-making are not evidence of EPM returns:
Rank #3
| Evidence | What it reports | What it does not establish |
|---|---|---|
| Brynjolfsson and McElheran, 2016 | Adoption of data-driven decision-making among U.S. manufacturing plants rose from 11% to 30% between 2005 and 2010; the authors said the diffusion and companion evidence were consistent with productivity benefits. American Economic Association paper | It is an adoption statistic, not an EPM adoption rate or an EPM ROI estimate. |
| FICO-reported Forrester Consulting study, 2021 | FICO reported a modeled 356% three-year ROI for a composite $10 billion financial-services organization using FICO Decision Modeler. FICO report | This is vendor-commissioned, product-specific modeled evidence, not an independent finding about EPMs generally. |
| Ludwig, Mullainathan, and Rambachan, 2024 | The paper estimates high social returns for selected algorithmic interventions in regulation, criminal justice, medicine, and education, while cautioning that those estimates do not imply the interventions should necessarily be scaled. American Economic Association paper | It evaluates selected interventions, not EPMs as a category. |
Together, these sources can inform a broader discussion of data-driven decisions and algorithmic interventions, but none demonstrates that EPMs as a class produce a particular financial return.
How to test an EPM business case
A sound evaluation begins with the decision, not the algorithm. Define what action the model will change, who or what the entity is, what outcome is being predicted, and the time horizon in which the prediction can still be acted on. Then specify the existing process as the baseline and identify an intervention that could plausibly improve outcomes.
Rank #4
- Set the decision and target. State the prediction target, decision owner, population, forecast horizon, and action the score may trigger.
- Document the baseline. Measure what currently happens without the EPM: outcomes, resource use, delays, costs, and how decisions are made.
- Compare like with like. Where feasible, evaluate the EPM-guided intervention against a credible comparator, such as the existing process or an appropriate controlled design. Separate improved prediction from the effect of acting on predictions.
- Measure both benefits and error costs. Track relevant outcomes and operating costs, including false positives and false negatives. Include the cost of unnecessary intervention and the cost of missed cases.
- Account for implementation and operation. Include data preparation, integration, staff time, infrastructure, monitoring, maintenance, and the cost of updating the model.
- Check harms and distribution. Assess whether performance or intervention effects differ across affected groups, and examine privacy, security, explainability, and the ability to challenge decisions.
- Monitor after launch. Recheck performance and net value as data, behavior, operating conditions, and the intervention change.
Algorithmic decision systems can create risks as well as benefits. A 2019 European Parliament Research Service review discusses broader concerns including discrimination, unfair practices, reduced autonomy, and constrained access to markets. Those are general risks of algorithmic decision systems, not findings specific to EPMs.
What do the baseball examples show?
Schmarzo opens with Strat-O-Matic, a baseball simulation board game whose player cards describe tendencies in hitting, fielding, pitching, and running. The analogy helps explain the idea of a player-specific profile; the cards are not machine-learning models, and the analogy is not evidence of an MLB deployment.
Best Value
The article proposes hypothetical uses for player-level predictions, including pitcher selection, batting matchups, lineup construction, defensive positioning, scouting, development, workload management, injury prevention, and return-to-play decisions. Examples such as choosing a reliever for a matchup or adjusting positioning based on a predicted ball direction illustrate possible decisions. The article does not identify a team, dataset, implemented system, or measured effect on wins, injuries, or finances. They should therefore be read as scenarios, not case-study results.
When is an EPM worth using?
An EPM is worth considering when a consequential, repeatable decision can be improved by a timely prediction and there is a realistic intervention to test. Before deployment, an organization should be able to explain what the score changes, how success will be measured against the current process, and how it will account for costs, errors, and affected people. If the score merely describes risk without changing an effective action, its predictive quality alone does not establish economic value.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →
