Framework
Decision Intelligence Framework
How the ChainAlign assessment works
What the Assessment Measures
The ChainAlign Decision Intelligence Assessment measures two distinct capabilities and the gap between them.
Capability 1
AI Maturity
How effectively your organisation uses AI across operations, modelling, coordination, and integration.
Capability 2
Decision Maturity
How your organisation captures, connects, learns from, and retains the reasoning behind its decisions.
The Gap
Decision Yield
The ratio between them. How much of your AI capability is actually reaching your decisions.
AI Maturity is mapped to the MIT CISR Enterprise AI Maturity Model (Weill, Woerner, Sebastian, 2024), based on a survey of 721 companies. Decision Maturity is an original ChainAlign framework measuring what no AI maturity model measures: whether the human judgment layer is instrumented.
MIT CISR research: Building Enterprise AI Maturity · MIT Sloan summary
AI Maturity Stages
Four stages, assessed across four dimensions: Operations, Financial Decision Support, Cross-functional Coordination, and Decision-Aware AI.
Experimenting
Piloting
Scaling
Future-Ready
Stage 1
Experimenting
The organisation is exploring AI but has not embedded it in workflows. AI tools exist as standalone experiments. Reporting is mostly manual. Financial modelling relies on single-point estimates built in spreadsheets. Each department uses its own data sources with no shared truth. AI tools work in silos, and the human bridges every gap between systems.
Stage 2
Piloting
AI automates specific tasks but humans still assemble the overall picture. Some dashboards exist but key reports still need manual work. Scenario modelling is possible but manual and low-confidence. Teams use the same systems but interpret data differently. Alignment takes time.
Stage 3
Scaling
AI is embedded in key workflows and influences how work gets prioritised. Dashboards are live and exceptions are flagged automatically. Statistical models produce ranges and confidence intervals. Key systems are connected and AI can pull cross-functional data, though the logic for combining it may still be hard-coded.
Stage 4
Future-Ready
AI runs significant parts of operations autonomously within defined guardrails. The system surfaces what needs attention before anyone asks. Uncertainty is modelled probabilistically and decision options are weighted automatically. A single live source of truth exists. AI assembles context across systems autonomously and presents the full cross-functional trade-off.
Decision Maturity Levels
Original ChainAlign framework. Five dimensions, each assessed on a four-level scale: Reactive, Documented, Coherent, Compounding.
Dimension 1
Coherence
Does every tactical decision connect to strategic intent?
Each decision is evaluated on its own merits. Strategy exists as an annual document but is loose enough to justify anything. Nobody tracks cumulative drift.
Strategy is referenced in decision memos and alignment is reviewed quarterly. The check is periodic, not continuous. Alignment depends on whether the right person raises the concern in the right meeting.
Guardrails flag when decisions conflict with stated priorities. A shared framework connects tactical choices to strategic objectives, though applying it still requires human facilitation.
Every significant decision is automatically scored against strategic objectives before commitment. When individual decisions are locally rational but collectively incoherent, the system surfaces that pattern.
Dimension 2
Trace Capture
Does the reasoning behind a decision survive the meeting where it was made?
Only outputs survive: the board paper, the approved budget, the signed contract. The reasoning lives in the heads of the people in the room.
Decision memos capture the what but not the why. Meeting minutes note that alignment was "discussed" but not that three people had serious reservations. The documentation is a label, not knowledge retention.
Structured decision logs exist for major decisions, including how conclusions were reached and what constraints were active. A future team member can understand the reasoning without starting from scratch.
Reasoning, assumptions, and constraints are captured in instrumented workflows as the decision forms. The reasoning is a first-class data asset, queryable and linked to outcomes.
Dimension 3
Decision Learning
Does the organisation get smarter from its decisions?
When a decision proves wrong, the conversation focuses on the outcome, not the reasoning. Assumption drift goes undetected until the P&L shows it.
Post-mortems happen but findings rarely change how the next decision gets made. The same class of mistake recurs because the insight was documented but not embedded in the workflow.
Lessons are documented and actively fed back into planning. KPIs are tracked and linked to the decision logic that produced them.
Automated monitoring flags when assumptions behind a previous decision are drifting from reality, before the outcome degrades.
Dimension 4
Decision Execution
Does analysis actually reach a commitment, and is it stress-tested before it becomes binding?
The output of analytical work is a presentation that gets discussed in a meeting. Nobody runs a simulation to compare options.
Dashboards exist that leaders can explore. Basic what-if scenarios are built in spreadsheets. The actual decision happens through unstructured debate that is not captured.
Recommendations come with supporting data and multiple options. Monte Carlo simulations or departmental digital twins are available for specific decisions.
A ranked set of options, each with quantified trade-offs. The meeting is about committing, not interpreting. Full value chain simulation tests decisions before commitment.
Dimension 5
Judgement Retention
Does the organisation's decision quality survive the departure of the people who built it?
If two or three key people left, decision-making quality would degrade significantly. The organisation solves the same class of problem every twelve to eighteen months.
Documentation exists but does not capture the real reasoning. A new team member can see what was decided but cannot reconstruct the judgment that produced it.
For major decisions, the reasoning is accessible. For routine decisions, institutional knowledge still lives in people. New hires take months to reach full context.
Organisational judgment lives in systems, not people. Past decisions and their reasoning are queryable. New talent steps into high-context roles without a knowledge cliff.
Decision Yield
Decision Maturity Score ÷ AI Maturity Score
Under-Powered Wisdom
Precision Alignment
Absorption Friction
Systemic Waste
Decision processes are strong but not yet amplified by AI. Ready for more aggressive AI investment.
AI investment and decision infrastructure are in balance. The opportunity is acceleration.
AI is generating capability your decision infrastructure cannot fully absorb. Value is leaking between insight and action.
AI investment is decoupled from decision logic. High risk of unauditable errors and wasted investment.
Yield is always interpreted alongside the absolute stage and level. A yield of 0.9 at Stage 4 means something very different from 0.9 at Stage 1.
This framework measures infrastructure, not instinct. An organisation with Level 1 Trace Capture may still make excellent decisions because it has excellent people. But it is fragile: those decisions do not survive the people who made them. The framework measures what is repeatable and compounding, not what is brilliant and ephemeral.
The Quadrants
The Compounding Leader
Both AI capability and decision infrastructure are strong. Decisions compound. This is the target state.
The Manual Artisan
Strong decision instincts and processes, but limited AI amplification. AI investment would compound what is already built.
The Expensive Engine
Significant AI investment is outpacing the organisation's ability to act on it. The AI works. The decisions do not capture it.
The Legacy Observer
Both capabilities need investment. Vulnerable to disruption from competitors who compound faster.
Find out where you stand
The assessment takes 10-15 minutes. You receive your AI Maturity stage, Decision Maturity level, and Decision Yield score immediately.
The quick assessment is 20 questions with instant results. The AI interview is a personalised 15-minute conversation adapted to your industry.
Attribution
AI Maturity is mapped to the MIT CISR Enterprise AI Maturity Model (Weill, Woerner, Sebastian, 2024). Based on a survey of 721 companies.
Decision Maturity is an original ChainAlign framework. Related reading: Context Graphs are Judgement Graphs.
Decision Yield is an original ChainAlign metric.
Questions about the framework: framework@chainalign.com
Version 1.0