Catalog
concept#Architecture#Governance#Product#Reliability

Uncertainty

A conceptual framework describing knowledge gaps, variability and ambiguity that influence decisions in engineering, architecture and product management.

Uncertainty describes limited or incomplete knowledge about future states, outcomes, or system behaviour that affects decisions in engineering, architecture, and product management.
Established
Medium

Classification

  • Medium
  • Organizational
  • Architectural
  • Intermediate

Technical context

Monitoring and observability tools (e.g. Prometheus)Experimentation platforms and feature flagsProject and product management tools

Principles & goals

Explore uncertainty through small, rapid experiments.Evaluate decisions based on observable metrics and hypotheses.Prefer adaptive plans and incremental investments over large irreversible decisions.
Discovery
Enterprise, Domain, Team

Use cases & scenarios

Compromises

  • Wrong metrics lead to misleading decisions.
  • Excessive hesitation can prevent market opportunities.
  • Incorrect mitigations tie up resources without benefit.
  • Record assumptions explicitly and review regularly.
  • Small, fast experiments before large investments.
  • Use metrics and SLOs to observe uncertainties.

I/O & resources

  • List of hypotheses and assumptions
  • Available user data and telemetry
  • Project and time constraints
  • Prioritised experiments and tests
  • Decision documentation with uncertainty assumptions
  • Monitoring and alerting strategy

Description

Uncertainty describes limited or incomplete knowledge about future states, outcomes, or system behaviour that affects decisions in engineering, architecture, and product management. It encompasses variability, ambiguity and unknowns, and frames how teams prioritise experiments, hedging strategies, and adaptive plans. Understanding uncertainty informs trade-offs and monitoring to reduce decision fragility.

✔

  • Better risk control and fewer costly wrong decisions.
  • Improved learning cycles through targeted experiments.
  • Transparency about assumptions and planned mitigations.

✖

  • May require time-consuming experiments and measurement effort.
  • Not all uncertainties can be fully eliminated.
  • Requires disciplined hypothesis formulation and measurement culture.

  • Decision latency

    Time between identifying an uncertainty and taking action.

  • Uncertainty reduction

    Measurable reduction in estimate range after validation steps.

  • Post-decision error rate

    Number and severity of wrong decisions observed in operation.

Feature experiment in a SaaS product

A/B tests used to clarify user preferences and adjust the roadmap.

Incremental architecture validation

Piloting a microservice variant to evaluate scalability and cost.

Monitoring adjustment after load spikes

Introduced dynamic thresholds to avoid false alarms from unknown load patterns.

1

Identify and document key uncertainties and assumptions.

2

Prioritise uncertainties by impact and remediation effort.

3

Define measurable hypotheses and run small experiments.

4

Integrate results into architecture decisions and monitoring.

⚠️ Technical debt & bottlenecks

  • Missing telemetry hinders future uncertainty analyses.
  • Monolithic structures prevent rapid iterations.
  • Unstructured decision documentation impedes traceability.
Insufficient telemetrySlow feedback cyclesTight budget or time constraints
  • Delaying all decisions indefinitely waiting for more data.
  • Misconfiguring experiments and drawing incorrect conclusions.
  • Using uncertainty as an excuse for lack of planning.
  • Relying solely on historical data when context changes.
  • Overestimating estimate accuracy without validation.
  • Unclear ownership of experiment results prevents learning.
Fundamentals of hypothesis formation and statisticsExperience with metrics, monitoring and A/B testingAbility for interdisciplinary communication (product, engineering, ops)
Availability and fault-tolerance requirementsUnpredictable load and usage profilesBusiness uncertainties and changing market conditions
  • • Limited measurability of certain assumptions
  • • Regulatory or safety-related restrictions
  • • Budget and time constraints for experiments