Stock and Flow Models in Technology Service Planning
Stock and flow modeling applies formal system dynamics principles to the planning and management of technology services, treating resources, capacity, and demand as interconnected accumulations and rates of change. This page describes the structure of stock and flow models, how they operate within technology service contexts, the professional and organizational scenarios where they appear, and the criteria that determine when this modeling approach is appropriate. The framework originates in the work of Jay W. Forrester at MIT and was formalized as a discipline through what is now the System Dynamics Society.
Definition and scope
A stock and flow model represents a system as a set of stocks — quantities that accumulate over time — and flows — rates that increase or decrease those stocks. In technology service planning, stocks include measurable accumulations such as server capacity, software licenses, unresolved incident tickets, trained personnel headcount, and allocated budget. Flows are the rates at which those stocks change: provisioning rate, attrition rate, ticket resolution rate, hiring velocity, and expenditure rate.
The scope of stock and flow modeling within technology services encompasses infrastructure planning, IT service management (ITSM), workforce capacity forecasting, and software delivery pipeline analysis. The MIT System Dynamics Group, which produced foundational texts including Forrester's Industrial Dynamics (1961) and John Sterman's Business Dynamics (2000), established the mathematical conventions that practitioners apply in these domains. Sterman's Business Dynamics remains the primary professional reference for stock and flow model construction and validation.
Unlike causal loop diagrams, which map qualitative feedback relationships, stock and flow models are quantitative: each stock has defined units of measurement, each flow has a rate equation, and the model produces time-series outputs that can be compared against observed data. This distinction positions stock and flow modeling as an intermediate step between conceptual feedback loop analysis and full simulation.
The broader methodology situates within system dynamics, one of the major formalisms covered across systems modeling methods.
How it works
A stock and flow model operates through four structural elements:
- Stocks — state variables represented as accumulations (e.g., number of active service requests, gigabytes of consumed storage, full-time equivalent staff in a role).
- Inflows — rates that increase a stock (e.g., new request arrival rate measured in tickets per hour).
- Outflows — rates that decrease a stock (e.g., ticket closure rate, server decommissioning rate).
- Auxiliary variables and parameters — constants and intermediate calculations that determine flow rates (e.g., average handling time, mean time between failures).
The fundamental equation governing every stock is an integral over time:
Stock(t) = Stock(t₀) + ∫[Inflow(s) − Outflow(s)] ds from t₀ to t
Simulation tools such as Vensim (referenced in Sterman's Business Dynamics) and STELLA implement this integral numerically using Euler or Runge-Kutta methods. The NIST Cybersecurity Framework does not prescribe modeling tools but references capacity and resource planning as components of the Identify and Protect functions, within which stock and flow simulation supports quantitative gap analysis.
Model construction follows a structured sequence:
Common scenarios
Infrastructure capacity planning — Organizations model server inventory as a stock depleted by decommissioning and replenished by procurement. A 12-month simulation can expose capacity shortfalls before they affect service-level agreements, particularly when demand growth is nonlinear. The ITIL 4 framework (published by Axelos, now part of PeopleCert) identifies capacity and performance management as a core practice, and stock and flow models are a recognized quantitative instrument within that practice.
Incident and problem management — Unresolved incidents constitute a stock. When inflow (new incident rate) exceeds outflow (resolution rate) for sustained periods, the backlog grows nonlinearly due to aging effects, prioritization shifts, and staff fatigue — dynamics that linear projections miss entirely.
Workforce planning in technology teams — Headcount in a given skill category is a stock governed by hiring rate, voluntary attrition, involuntary separation, and internal transfers. The U.S. Bureau of Labor Statistics Occupational Outlook Handbook, which publishes workforce data for technology occupations, provides empirical flow rate inputs (e.g., median annual turnover rates by occupation) that anchor model parameters.
Software delivery pipelines — Work items in progress represent a stock. Cycle time, throughput, and queue depth are all derivable from stock and flow relationships, making this approach compatible with the quantitative measurements advocated in the DORA (DevOps Research and Assessment) State of DevOps reports published annually by Google Cloud.
Decision boundaries
Stock and flow modeling is appropriate when:
Stock and flow models are less appropriate — and often replaced by agent-based modeling — when individual actor heterogeneity drives outcomes, or when spatial distribution of entities is analytically central. They are also distinct from soft systems methodology, which addresses problem structuring in ill-defined organizational situations rather than quantitative simulation.
The systems theory reference index provides orientation across the full landscape of modeling formalisms and their application domains, situating stock and flow models within the broader taxonomy of systems analysis techniques.