• Home
  • BVSSH
  • C4E
  • Playbooks
  • Frameworks
  • Good Reads
Search

What are you looking for?

Standard : Model iteration cycles are measured and continuously shortened

Purpose and Strategic Importance

This standard requires that the end-to-end cycle time for a model iteration — from identifying a need to improve to deploying a validated improvement to production — is measured, visible to the team, and subject to continuous improvement. It supports the policy of shipping AI incrementally to learn faster by treating iteration speed as a strategic metric rather than an incidental outcome of project planning. Teams that can iterate on models quickly compound their learning faster and respond to production signals before they become serious problems.

Strategic Impact

  • Creates a competitive advantage in AI-driven products by enabling more learning cycles per time period than slower-moving competitors
  • Shortens the window between identifying a performance problem and deploying a fix, reducing the duration of user impact
  • Builds engineering discipline around the full model iteration lifecycle rather than treating it as unpredictable project work
  • Encourages incremental improvement culture that favours many small validated improvements over rare large-batch releases
  • Provides the organisation with data to make evidence-based investments in tooling and process improvements that accelerate delivery

Risks of Not Having This Standard

  • Model iteration cycles grow progressively longer as technical debt, process overhead, and approval friction accumulate unnoticed
  • Teams cannot predict delivery timelines for model improvements, undermining product planning and stakeholder confidence
  • Slow iteration cycles prevent the organisation from responding to model degradation before it significantly impacts users
  • Engineers become demoralised by the gap between identifying an improvement and seeing it in production
  • Competitors with faster iteration cycles learn and adapt faster, eroding AI-driven product advantages

CMMI Maturity Model

Level 1 – Initial

Category Description
People & Culture - Model iteration cycle time is unknown; teams have no sense of how long a typical improvement takes from start to deployment
Process & Governance - No tracking of model iteration lifecycle; each improvement is treated as a unique project with ad hoc planning
Technology & Tools - No tooling measures cycle time across the model development lifecycle stages
Measurement & Metrics - Cycle time is not measured; the team cannot identify where time is being lost in the iteration process

Level 2 – Managed

Category Description
People & Culture - Teams begin tracking the start and end dates of model improvement work; awareness of cycle time is growing
Process & Governance - Model iteration lifecycle stages are defined (identify, experiment, train, evaluate, deploy); time spent in each stage is recorded
Technology & Tools - A simple tracking board or project management tool captures model improvement items with stage transitions and timestamps
Measurement & Metrics - Mean cycle time per model iteration is calculated; the team discusses outliers in retrospectives

Level 3 – Defined

Category Description
People & Culture - Reducing cycle time is an explicit team goal; engineers are aware of where friction and delays occur in the iteration pipeline
Process & Governance - Cycle time targets are set per model type; improvements to cycle time are a standing item in engineering reviews
Technology & Tools - MLOps pipeline instrumentation captures time spent at each lifecycle stage automatically; flow metrics are visible on a live dashboard
Measurement & Metrics - Cycle time is broken down by stage (data prep, training, evaluation, deployment approval, deployment); bottlenecks are identified and addressed

Level 4 – Quantitatively Managed

Category Description
People & Culture - Teams are accountable for cycle time targets; iteration speed is reviewed alongside quality and reliability in engineering governance
Process & Governance - Process improvement initiatives are prioritised based on data showing where the most cycle time is being lost
Technology & Tools - Automated pipeline optimisation tools identify parallelisation opportunities and resource allocation improvements
Measurement & Metrics - Cycle time trend, throughput (iterations per month), and lead time distribution are tracked and compared to targets quarterly

Level 5 – Optimising

Category Description
People & Culture - Cycle time improvement is a shared organisational discipline; teams publish their improvement stories and tooling contributions
Process & Governance - Cycle time standards are continuously updated based on advances in tooling and the organisation's growing operational maturity
Technology & Tools - Automated retraining pipelines trigger, execute, and validate model iterations without manual intervention for defined change classes
Measurement & Metrics - Cycle time data drives strategic roadmap decisions about automation investment and pipeline architecture

Key Measures

  • Mean end-to-end model iteration cycle time (from improvement identification to production deployment) per model
  • Cycle time broken down by stage: data preparation, training, evaluation, deployment approval, and deployment execution
  • Cycle time trend per model over rolling 12-month period
  • Percentage of model iterations completed within the target cycle time for their model risk tier
  • Number of process improvement actions taken in the last quarter specifically to reduce model iteration cycle time
Associated Policies
Associated Practices
  • MLOps Pipeline Design
  • Experiment Tracking and Management

Technical debt is like junk food - easy now, painful later.

Awesome Blogs
  • LinkedIn Engineering
  • Github Engineering
  • Uber Engineering
  • Code as Craft
  • Medium.engineering