The Automation Opportunity Portfolio: A Value, Feasibility, Risk and Learning Playbook

You may have more automation ideas than the team can fund or deliver. Compare them against the same outcomes, evidence, dependencies, capacity and risks so the strongest demonstration or loudest sponsor does not decide the order.

This guide helps a COO, CIO and CFO decide which accepted automation opportunities should receive a next evidence gate, which should be validated or piloted, and which should be held, deferred or stopped. It does not choose rules, platforms, agents or custom software; The Automation Architecture Decision: Rules, Platforms, Agents, Custom Software or No Build owns that architecture decision. It does not design a production pilot; The Proof-of-Value Pilot Standard: Designing Automation Tests That Can Earn Production owns that standard.

1. The Market Connection Context

The recognisable leadership condition is a spreadsheet of automation or AI ideas with incomparable claims. One entry promises time saved, another promises service improvement, another responds to a control concern, and another exists because a platform vendor demonstrated a feature. Baselines are missing, dependencies appear only after selection, and every sponsor calls their idea urgent.

The primary value path is:

Commitments and work
→ evidence-bounded opportunities
→ coherent portfolio decisions
→ usable capacity and reliable service
→ customer, employee or operating outcome
→ evidence-led rebalance, stop or scale decision

The bottom-line risk lies in portfolio conversion. If weak ideas consume scarce integration, process-owner, data, security and change capacity, stronger opportunities wait. If feasibility is treated as value, the organisation automates what is easy rather than what matters. If projected labour time is treated as cash, a capacity claim becomes a saving without an operating decision to release, redeploy or absorb that capacity. If risks are averaged into a score, one unacceptable harm can disappear inside an attractive total.

Potential effects may involve revenue, margin, cost, capacity, retention, risk, investment efficiency and service value. Every quantified effect is REQUIRES VALIDATION until its population, period, baseline, definition, source, counterfactual, owner and attribution limits are accepted.

The portfolio decision contract

For every opportunity, leadership assigns one primary disposition:

Funding is not proof of value and a high rank is not authority. Each decision must state what is funded: validation, architecture, pilot design, implementation planning or another bounded gate. Detailed architecture remains The Automation Architecture Decision: Rules, Platforms, Agents, Custom Software or No Build; governance design The AI and Automation Governance Standard: Proportionate Control from Assistance to Autonomous Action; pilot acceptance The Proof-of-Value Pilot Standard: Designing Automation Tests That Can Earn Production; production operations The Automation Production Operating Standard: Integration, Monitoring, Incidents and Change Control; partner diligence The Automation Partner and Commercial Due-Diligence Playbook: Scope, Cost, Risk and Exitability; and enterprise scale The Enterprise Automation Scale Standard: Shared Capabilities, Portfolio Governance and Capability Transfer.

A common comparison model without a magic score

Compare each opportunity through five lenses:

1. Value: the affected commitment, market or service outcome, current baseline, potential exposure and capacity conversion decision.
2. Feasibility: workflow stability, data fitness, integration, exception handling, supportability, capability and operating ownership.
3. Risk: consequence of error, customer and employee harm, privacy, security, legal, financial, service and third-party exposure, plus reversibility and human authority.
4. Dependency: shared systems, data, people, controls, vendors, decisions and sequencing constraints.
5. Learning: which decision-critical uncertainty a bounded validation or pilot could resolve and whether that learning transfers to other opportunities.

The lenses support judgement; they do not create an automatic rank. A mandatory control failure cannot be offset by attractive value. A missing owner cannot be offset by technical ease. A large speculative benefit cannot outrank a smaller evidenced outcome merely because both were converted to one number.

HM Treasury's Green Book describes appraisal as comparing costs, benefits and risks of options against objectives, retaining business-as-usual and communicating uncertainty (The Green Book 2026, VERIFIED official public-sector guidance at the source cut-off). Its rules are mandatory in their stated UK public-sector context, not automatically for private organisations; this guide applies only the transparent options-and-evidence discipline.

NIST's AI Risk Management Framework organises voluntary AI risk-management activity through Govern, Map, Measure and Manage (NIST AI RMF 1.0, VERIFIED official source at the source cut-off). This guide uses that source only when an opportunity genuinely involves AI. Conventional workflow automation still requires proportionate business, operational, privacy, security and control review.

Scope, exclusions and dependencies

Use the method when leadership has multiple accepted opportunities competing for limited evidence, architecture, integration, control, process-owner, delivery or change capacity. Each opportunity must have an outcome owner and enough The Automation and Technology Readiness Assessment: An Executive Standard for Choosing What Should Change/The Workflow Constraint and Root-Cause Audit: Simplify the Work Before Automating It evidence to state the problem without prescribing technology.

Do not use the full portfolio method for one urgent, already-authorised containment action; one bounded opportunity with no portfolio trade-off; or a statutory or control conclusion that belongs to a qualified authority. Stop or narrow the work if evidence is unauthorised, sponsors will not expose assumptions, opportunity owners cannot act, or a predetermined technology purchase makes comparison performative.

Scope includes portfolio charter, eligible opportunity intake, baseline and claim normalisation, value ranges, feasibility conditions, risk and control gates, dependencies, learning value, capacity, sequence, executive dispositions, handover and review. It excludes new root-cause diagnosis, solution architecture, detailed cost model, procurement, implementation design, production release, workforce determination, legal advice and guaranteed return.

Expected results are a governed opportunity inventory, comparable evidence contracts, dependency map, capacity envelope, viable sequence, executive decision register and review mechanism. They are not a forecast, business case approval, architecture selection or delivery promise.

2. The Leadership vs. Delegation Boundary

The six value paths cross functional budgets, evidence and ownership; portfolio diagnosis therefore needs cross-functional authority, access and independent challenge, whether led internally or externally.

When an internal team can lead

A capable internal team can lead when the COO can resolve cross-functional outcome and capacity conflicts, finance can challenge benefit and cost definitions, the CIO can expose shared technical dependencies without steering architecture, qualified owners can apply control gates, and sponsors accept that their opportunity may be held or stopped. The team needs an independent chair or challenger who is not measured by the number of initiatives funded.

When independent leadership is justified

A senior independent partner is proportionate when opportunities span competing budgets and functions; platform, supplier or programme interests shape the list; evidence maturity varies materially; leadership lacks one comparison model; or the same scarce people and systems appear in many proposals. The partner may reconcile evidence and facilitate trade-offs but cannot accept risk, invent baselines or fund the portfolio.

When a narrower specialist is sufficient

Use a process specialist to complete a missing The Workflow Constraint and Root-Cause Audit: Simplify the Work Before Automating It diagnosis, an analyst or finance specialist to repair a defined baseline, an architect after an opportunity passes into The Automation Architecture Decision: Rules, Platforms, Agents, Custom Software or No Build, or a qualified risk specialist for a bounded control question. If one opportunity is already isolated and no cross-portfolio trade-off exists, apply its next-gate standard directly rather than commissioning a portfolio exercise.

Stop and escalation routes

Pause an opportunity for unauthorised data, unsafe participation, an uncontained incident, unmanageable conflict, missing owner, false saving claim or unacceptable mandatory control. Escalate outcome and capacity conflicts to the COO or economic buyer; financial definitions to the CFO and benefit owner; technical dependencies to the CIO; privacy, security, legal, employment and regulatory matters to qualified authorities; and supplier interests to procurement and the independent challenger.

Questions for evaluating a provider

1. How will you reject opportunities that have attractive technology but weak workflow evidence?
2. How will you distinguish released, redeployed and absorbed capacity from cash savings?
3. Which non-negotiable gates cannot be averaged into a score?
4. How will you preserve source, baseline, uncertainty and counterfactual for each value claim?
5. How will shared data, integration, control and operating-owner dependencies affect sequencing?
6. How will learning value be separated from production value?
7. Which do-nothing, direct improvement and stop alternatives will remain visible?
8. How will sponsor, supplier and sunk-cost interests be recorded?
9. Where will this portfolio method stop before architecture, pilot or procurement?
10. Which reusable records and capability will remain with our team?

Reject any proposal that promises a universal automation score, converts employee time directly into savings, or ranks opportunities before outcome, evidence and mandatory controls are visible.

3. The Step-by-Step Execution Standard

Use the Portfolio Charter and Capacity Register, Opportunity Evidence and Baseline Register, Value Feasibility Risk and Learning Matrix, and Portfolio Sequence Decision and Review Register. Retain stable IDs, approved access and version history.

Step 01: Charter the portfolio decision

Purpose and decision served Define which opportunities compete, what outcomes matter and which dated resource decision leadership will make.

Accountable owner COO as portfolio sponsor, with the economic buyer accepting resource authority.

Delegated lead and contributors Portfolio lead, CFO, CIO, operating owners, benefit owners, data, risk, people, procurement and customer/service contributors.

Preconditions, inputs and access Candidate opportunity list, decision forum, known capacity constraints and authority to stop or hold sponsored ideas.

Actions

1. State the portfolio decision without naming a preferred technology.
2. Define affected markets, services, workflows and value paths.
3. Set intake close, evidence close, decision date and review boundary.
4. Approve the six dispositions, minimum gates and stop conditions.

Evidence to retain Charter, original list, forum mandate, conflicts, boundaries and approval.

Decision or scoring rule Proceed only when leadership can fund, validate, pilot, hold, defer or stop; change an advisory-only exercise.

Output and template Charter rows in the Portfolio Charter and Capacity Register with A2P-PORT-* and A2P-DEC-* IDs.

Acceptance gate PROCEED if outcome, boundary, decision date, authority and stop route are accepted; otherwise CHANGE, DEFER or STOP.

Leadership checkpoint The sponsor confirms portfolio selection is not a target count or technology adoption quota.

Human control point Only the authorised forum sets capital, capacity and risk boundaries.

Failure modes and safeguards An idea contest and enterprise-wide scope are controlled by eligibility and explicit boundaries.

Change, stop or escalation condition Stop if leadership cannot remove or defer opportunities; escalate disputed resource authority.

Handoff to the next step Pass the accepted charter to Step 02.

Step 02: Establish authority, conflicts and capacity

Purpose and decision served Make decision rights, scarce capacity, control mandates and sponsor interests visible before comparison.

Accountable owner COO and CFO jointly for portfolio capacity and funding boundaries.

Delegated lead and contributors Portfolio lead, finance, CIO, enterprise delivery, information authority, risk, procurement and people owners.

Preconditions, inputs and access Charter, organisational capacity sources, current commitments, supplier relationships and control mandates.

Actions

1. Name decision, evidence, funding, operating and control owners.
2. Record available process-owner, data, integration, architecture, security, change and support capacity by authoritative source.
3. Declare sponsor, supplier, platform and sunk-cost interests.
4. Set recusal, escalation, evidence handling and capacity-reservation rules.

Evidence to retain Authority matrix, capacity baselines, commitment register, conflicts and approvals.

Decision or scoring rule Capacity is available only when its owner accepts the unit, period, existing load and allocation authority.

Output and template Authority and capacity rows in the Portfolio Charter and Capacity Register.

Acceptance gate PROCEED when material authorities and scarce capacity have named owners and evidence; otherwise VALIDATE, DEFER or STOP.

Leadership checkpoint Leadership accepts that portfolio capacity is a constraint, not an assumption created by prioritisation.

Human control point Capacity and control owners approve what may be committed.

Failure modes and safeguards Double-booked specialists and hidden supplier incentives are controlled by one capacity and conflict record.

Change, stop or escalation condition Escalate overlapping commitments; stop unauthorised evidence or unmanaged conflicts.

Handoff to the next step Pass authority and capacity bounds to Step 03.

Step 03: Gate opportunity eligibility

Purpose and decision served Keep unresolved complaints and technology ideas out of portfolio comparison until they have an owned problem and evidence path.

Accountable owner Operating owner for each opportunity.

Delegated lead and contributors Portfolio analyst, The Automation and Technology Readiness Assessment: An Executive Standard for Choosing What Should Change readiness lead, The Workflow Constraint and Root-Cause Audit: Simplify the Work Before Automating It workflow lead, finance, technology and control contributors.

Preconditions, inputs and access Candidate list, stable IDs and supporting readiness or root-cause records.

Actions

1. Require a bounded workflow, affected population, outcome owner and current condition.
2. Require evidence that simplification, policy clarification and direct repair were considered.
3. Identify readiness and root-cause gaps without repeating those methods.
4. Admit, return for validation, defer or stop each candidate.

Evidence to retain Eligibility result, supporting record links, gaps, owner and rationale.

Decision or scoring rule An idea is eligible only when its problem and outcome can be compared without assuming architecture.

Output and template Eligibility rows in the Opportunity Evidence and Baseline Register.

Acceptance gate ADMIT, RETURN TO READINESS/WORKFLOW DIAGNOSIS, DEFER or STOP.

Leadership checkpoint The sponsor accepts that popularity and executive sponsorship do not override eligibility.

Human control point The operating owner accepts the problem and the portfolio chair accepts admission.

Failure modes and safeguards Tool-led ideas and duplicated opportunities are controlled by problem evidence and stable IDs.

Change, stop or escalation condition Merge duplicates; stop ownerless or immaterial cases.

Handoff to the next step Pass admitted opportunities to Step 04.

Step 04: Normalise outcome, baseline and counterfactual

Purpose and decision served Give unlike opportunities a common evidence contract without pretending their outcomes use one metric.

Accountable owner Benefit owner, with CFO authority for financial definitions.

Delegated lead and contributors Analyst, finance, operating owner, customer/service owner, data owner and source custodians.

Preconditions, inputs and access Eligible opportunities, source access, baseline periods and existing commitments.

Actions

1. Define outcome, population, period, measure, denominator and exclusions.
2. Record business-as-usual and do-minimum counterfactuals.
3. Separate observed baseline, inferred consequence, proposed change and evidence gap.
4. State source quality, lag, attribution and expiry.

Evidence to retain Metric contract, baseline extract, counterfactual, source lineage and limitations.

Decision or scoring rule No quantified value enters comparison without an accepted baseline or explicit REQUIRES VALIDATION status.

Output and template Baseline and claim rows in the Opportunity Evidence and Baseline Register.

Acceptance gate COMPARABLE, VALIDATE, DEFER or STOP.

Leadership checkpoint The CFO confirms that capacity, cost avoidance and cash effects remain distinct.

Human control point Source and benefit owners accept definitions and counterfactuals.

Failure modes and safeguards False equivalence is controlled by common fields rather than a forced common unit.

Change, stop or escalation condition Reduce claims that exceed the source boundary; escalate contested finance definitions.

Handoff to the next step Pass accepted evidence contracts to Step 05.

Step 05: Bound potential value and capacity conversion

Purpose and decision served Estimate decision-useful exposure without presenting speculative opportunity as realised benefit.

Accountable owner Benefit owner and CFO.

Delegated lead and contributors Finance analyst, operating owner, workforce/capacity owner, customer/service owner and evidence lead.

Preconditions, inputs and access Accepted baselines, counterfactuals, demand evidence, workload definitions and constraints.

Actions

1. Trace the workflow condition to customer, service, operating and potential economic consequence.
2. Define low/base/high or bounded scenarios only from explicit assumptions; avoid point certainty.
3. Separate released, redeployed and absorbed capacity and name the operating decision required for each.
4. Record benefit dependencies, disbenefits, lag and confidence.

Evidence to retain Value chain, scenario inputs, assumptions, capacity conversion route and owner acceptance.

Decision or scoring rule Use potential value only within its evidence confidence; no owner or conversion decision means no claimed saving.

Output and template Value rows in the Value Feasibility Risk and Learning Matrix.

Acceptance gate ACCEPT RANGE, VALIDATE, EXCLUDE CLAIM or STOP.

Leadership checkpoint The economic buyer sees what must happen after automation for value to become usable.

Human control point Finance and benefit owners approve economic interpretation.

Failure modes and safeguards Time-to-cash conversion and double-counted benefits are controlled by conversion routes and dependency IDs.

Change, stop or escalation condition Remove unsupported values; escalate overlap between opportunities.

Handoff to the next step Pass bounded value claims to Step 06.

Step 06: Assess feasibility without choosing architecture

Purpose and decision served Test whether conditions exist for a supportable intervention while preserving The Automation Architecture Decision: Rules, Platforms, Agents, Custom Software or No Build as a later decision.

Accountable owner CIO and operating owner jointly.

Delegated lead and contributors Workflow, data, integration, architecture, security, support, change and supplier contributors.

Preconditions, inputs and access Workflow evidence, rule and exception conditions, data lineage, system dependencies and capability evidence.

Actions

1. Test workflow stability, exception ownership, input/output clarity and human decisions.
2. Test data fitness, access, integration, identity, logging, fallback and support conditions.
3. Test operating ownership, skills, adoption burden and change capacity.
4. Record unknowns without selecting rules, platforms, agents or custom software.

Evidence to retain Feasibility conditions, dependencies, source links, gaps and qualified interpretations.

Decision or scoring rule Feasible means no known critical condition blocks a bounded next gate; it does not mean production-ready.

Output and template Feasibility rows in the Value Feasibility Risk and Learning Matrix.

Acceptance gate PASS FOR NEXT GATE, VALIDATE, DEFER or STOP.

Leadership checkpoint The CIO confirms that feasibility has not become architecture selection.

Human control point Technical and operating owners accept conditions; vendors cannot self-certify feasibility.

Failure modes and safeguards Demo bias is controlled by workflow, integration, operating and fallback evidence.

Change, stop or escalation condition Escalate material resilience or security exposure; defer missing prerequisites.

Handoff to the next step Pass feasibility and dependencies to Step 07.

Step 07: Apply risk, harm and control gates

Purpose and decision served Identify non-negotiable conditions and downside that cannot be averaged into value.

Accountable owner Risk executive and relevant qualified control owners.

Delegated lead and contributors Risk analyst, privacy, security, legal, employment, finance, service, safety, accessibility and operational owners.

Preconditions, inputs and access Use context, affected populations, data and action boundaries, consequence of error and existing controls.

Actions

1. Map customer, employee, service, financial, privacy, security, legal, regulatory and third-party consequences.
2. Define human authority, permissions, verification, fallback, containment and incident routes.
3. Separate mandatory gates from risk that the authorised owner may accept.
4. Record reversibility, residual uncertainty and qualified referrals.

Evidence to retain Risk scenarios, control decisions, referrals, approvals, limits and stop routes.

Decision or scoring rule Failure of a mandatory gate blocks funding regardless of aggregate score.

Output and template Risk and guardrail rows in the Value Feasibility Risk and Learning Matrix.

Acceptance gate PASS WITHIN BOUNDARY, VALIDATE, CONTAIN, DEFER or STOP.

Leadership checkpoint Leaders see distribution of benefit and harm, not only portfolio totals.

Human control point Qualified organisational authorities retain control and residual-risk decisions.

Failure modes and safeguards Risk averaging and generic “human oversight” are controlled by mandatory gates and named authority.

Change, stop or escalation condition Contain and escalate material incidents immediately; stop unacceptable harm.

Handoff to the next step Pass accepted risk boundaries to Step 08.

Step 08: Map dependencies and portfolio collisions

Purpose and decision served Expose shared prerequisites and contention that individual opportunity cases hide.

Accountable owner Portfolio sponsor.

Delegated lead and contributors Portfolio analyst, enterprise delivery, architecture, data, security, process, people, procurement and operating owners.

Preconditions, inputs and access Opportunity conditions, capacity register, current initiatives and dependency owners.

Actions

1. Assign stable dependency IDs across data, systems, integrations, people, policy, controls and decisions.
2. Identify enabling, blocking, shared and conflicting relationships.
3. Test whether a prerequisite should become a separate platform or capability decision.
4. Record contention, acceptance conditions and sequencing logic.

Evidence to retain Dependency map, owner acceptance, capacity demand, conflicts and source evidence.

Decision or scoring rule No opportunity may assume an unowned or double-booked critical dependency.

Output and template Dependency rows in the Portfolio Charter and Capacity Register and portfolio matrix.

Acceptance gate SEQUENCEABLE, VALIDATE, DEFER, REMOVE DEPENDENCY or STOP.

Leadership checkpoint Leadership distinguishes reusable enablement from hidden portfolio overhead.

Human control point Dependency owners accept availability and conditions.

Failure modes and safeguards Invisible shared work is controlled by one cross-opportunity dependency register.

Change, stop or escalation condition Escalate capacity collisions; defer opportunities whose prerequisite lacks an owner.

Handoff to the next step Pass the dependency graph to Step 09.

Step 09: Value decision learning

Purpose and decision served Prioritise bounded evidence that can resolve material uncertainty or unlock several decisions.

Accountable owner Decision owner for the uncertainty.

Delegated lead and contributors Research, analytics, architecture, pilot, operating and risk contributors.

Preconditions, inputs and access Uncertainty register, decision consequences, validation methods, dependencies and safe test boundaries.

Actions

1. State the decision that new evidence will change.
2. Define the smallest validation or pilot-design question, evidence, burden and expiry.
3. Identify learning reusable across opportunities without double-counting value.
4. Specify negative results and stop conditions before testing.

Evidence to retain Learning question, method, owner, reuse boundary, decision date and result criteria.

Decision or scoring rule Learning value exists only when possible results cause different decisions.

Output and template Learning rows in the Value Feasibility Risk and Learning Matrix.

Acceptance gate VALIDATE, ROUTE TO PILOT DESIGN ONLY AFTER AUTOMATION ARCHITECTURE AND APPLICABLE GOVERNANCE CONTROLS, NO TEST NEEDED or STOP.

Leadership checkpoint The sponsor confirms learning is not disguised implementation.

Human control point The decision owner accepts the question and qualified owners accept test safety.

Failure modes and safeguards Open-ended experimentation is controlled by a decision, expiry, burden and stop rule.

Change, stop or escalation condition Stop tests that cannot change a decision or operate within safeguards.

Handoff to the next step Pass viable opportunities and learning actions to Step 10.

Step 10: Compare opportunities with gates and ranges

Purpose and decision served Produce a transparent shortlist without reducing unlike evidence to one authoritative score.

Accountable owner Portfolio sponsor and CFO.

Delegated lead and contributors Portfolio lead, analyst, benefit, operating, technology, risk and independent challenge contributors.

Preconditions, inputs and access Value ranges, feasibility, controls, dependencies, learning actions and evidence status.

Actions

1. Remove opportunities that fail eligibility or mandatory gates.
2. Compare remaining cases through common fields and explicit evidence confidence.
3. Use ranges and sensitivity where assumptions are material.
4. Retain business-as-usual, direct improvement, hold and stop alternatives.

Evidence to retain Comparison view, evidence links, assumptions, sensitivity, exclusions and dissent.

Decision or scoring rule A score may organise discussion but cannot override mandatory gates, ownership, evidence or capacity.

Output and template Comparison rows in the Value Feasibility Risk and Learning Matrix.

Acceptance gate SHORTLIST, VALIDATE, HOLD, DEFER or STOP with rationale.

Leadership checkpoint The CFO and COO challenge speculative value, false precision and local optimisation.

Human control point The authorised portfolio forum accepts the shortlist.

Failure modes and safeguards Weighted-score theatre is controlled by gates, raw fields and visible dissent.

Change, stop or escalation condition Return cases with material evidence gaps; stop manipulated or ownerless cases.

Handoff to the next step Pass the shortlist to Step 11.

Step 11: Sequence within real capacity

Purpose and decision served Build the smallest coherent sequence that respects dependencies, learning and operating absorption.

Accountable owner COO, CFO and CIO through the mandated portfolio forum.

Delegated lead and contributors Portfolio planner, capacity owners, dependency owners, operating leads, finance, change and risk contributors.

Preconditions, inputs and access Shortlist, capacity envelope, dependency graph, existing commitments and learning actions.

Actions

1. Reserve capacity for mandatory operations, controls and existing commitments.
2. Sequence validations and enabling dependencies before dependent opportunities.
3. Limit concurrent change to accepted operating and support capacity.
4. Define entry, exit, hold and rebalance gates without invented delivery timing.

Evidence to retain Sequence rationale, capacity allocations, dependencies, conflicts and gate definitions.

Decision or scoring rule An opportunity enters the sequence only when its next gate has capacity, owner and prerequisites.

Output and template Sequence rows in the Portfolio Sequence Decision and Review Register.

Acceptance gate SEQUENCE, HOLD, DEFER, VALIDATE or STOP.

Leadership checkpoint Leadership confirms the sequence protects normal service and required control work.

Human control point Capacity owners accept allocations; the ranking tool cannot book people.

Failure modes and safeguards Overcommitment and attractive parallelism are controlled by owner-accepted capacity and work-in-progress limits.

Change, stop or escalation condition Rebalance when a dependency, guardrail or capacity condition changes materially.

Handoff to the next step Pass the feasible sequence to Step 12.

Step 12: Authorise portfolio dispositions

Purpose and decision served Make explicit resource and evidence decisions for every eligible opportunity.

Accountable owner Economic buyer through the authorised forum.

Delegated lead and contributors Decision secretary, sponsor, CFO, CIO, COO, operating and benefit owners, control owners and challenger.

Preconditions, inputs and access Decision pack, evidence links, capacity sequence, conflicts, dissent and forum authority.

Actions

1. Assign fund next gate, validate, route to architecture, hold, defer or stop. Route to The Proof-of-Value Pilot Standard: Designing Automation Tests That Can Earn Production pilot design only when accepted The Automation Architecture Decision: Rules, Platforms, Agents, Custom Software or No Build architecture evidence and applicable The AI and Automation Governance Standard: Proportionate Control from Assistance to Autonomous Action controls are linked.
2. Record resource commitment separately from disposition.
3. Name owner, output, acceptance, guardrail, expiry and stop condition.
4. Record dissent, assumptions and rejected alternatives.

Evidence to retain Signed decisions, attendance, authority, rationale, conditions and evidence links.

Decision or scoring rule Every eligible opportunity receives one primary disposition; ranking alone authorises nothing.

Output and template Decision rows in the Portfolio Sequence Decision and Review Register.

Acceptance gate ACCEPTED only with complete disposition, owner, resource boundary and next gate.

Leadership checkpoint The economic buyer confirms aggregate commitments remain within portfolio capacity and risk appetite.

Human control point Only the mandated forum funds, accepts risk or stops work.

Failure modes and safeguards Conditional ambiguity is controlled by one enum and separate decision, resource, measure and gate rows.

Change, stop or escalation condition Defer absent quorum; stop unresolved mandatory-control failure or unmanaged conflict.

Handoff to the next step Pass accepted next gates and held cases to Step 13.

Step 13: Handover, review and rebalance

Purpose and decision served Keep opportunity evidence, capacity and portfolio sequence current as learning arrives.

Accountable owner Portfolio sponsor and benefit owners.

Delegated lead and contributors Portfolio operator, downstream owners, measurement owners, finance, technology, operating and control owners.

Preconditions, inputs and access Accepted decisions, next-gate briefs, measure contracts, source access, event triggers and review authority.

Actions

1. Obtain explicit handover acceptance from each next-gate owner.
2. Review learning, delivery, operating, outcome and guardrail evidence separately.
3. Reconcile changed dependencies, capacity, baselines and opportunity costs.
4. Continue, change, hold, validate, defer, stop, scale or re-charter through named decisions.

Evidence to retain Handover receipts, current measures, review decisions, changes, closures and retired access.

Decision or scoring rule Rebalance on material evidence or capacity change, not to protect sunk cost or an original rank.

Output and template Handover, measure and review rows in the Portfolio Sequence Decision and Review Register.

Acceptance gate Every review ends with an owned disposition and next date; closed items retire access and claims.

Leadership checkpoint The portfolio forum reviews total burden, learning and outcome evidence rather than delivery activity alone.

Human control point Named owners interpret evidence, accept changes and retain qualified authority.

Failure modes and safeguards Stale rankings and zombie pilots are controlled by expiry, evidence refresh and stop decisions.

Change, stop or escalation condition Escalate guardrail breaches immediately; re-charter when outcome or portfolio boundary changes materially.

Handoff to the next step Close the cycle or return materially changed portfolio questions to Step 01.

4. The Evidence & Measurement Checklist

Preflight and authority

• The portfolio has one dated resource-and-evidence decision rather than an idea-ranking exercise. Evidence: accepted portfolio charter. Owner: COO.
• Eligible practices, workflows, populations and exclusions are bounded. Evidence: scope register. Owner: portfolio sponsor.
• Funding, operating, evidence, control and stop authorities are named. Evidence: authority matrix. Owner: economic buyer.
• Scarce capacity has authoritative units, periods, owners and existing commitments. Evidence: capacity register. Owner: CFO and CIO.
• Sponsor, supplier, platform and sunk-cost interests are disclosed. Evidence: conflict register. Owner: independent challenger.

Evidence access and quality

• Every opportunity has a stable ID, outcome owner and supporting The Automation and Technology Readiness Assessment: An Executive Standard for Choosing What Should Change or The Workflow Constraint and Root-Cause Audit: Simplify the Work Before Automating It evidence. Evidence: opportunity intake. Owner: operating owner.
• Baseline population, period, unit, denominator and exclusions are accepted. Evidence: metric contract. Owner: source owner.
• Business-as-usual and do-minimum counterfactuals are recorded. Evidence: baseline register. Owner: benefit owner.
• Value claims retain source, assumption, confidence, lag and attribution limits. Evidence: evidence rows. Owner: proof owner.
• Missing evidence is marked REQUIRES VALIDATION rather than zero. Evidence: evidence-gap register. Owner: decision owner.

Method execution

• Eligibility rejects tool-led, ownerless and undiagnosed ideas. Evidence: eligibility decisions. Owner: portfolio chair.
• Value, feasibility, risk, dependency and learning use common fields without forced common units. Evidence: portfolio matrix. Owner: portfolio lead.
• Released, redeployed and absorbed capacity remain distinct from cash savings. Evidence: value records. Owner: CFO.
• Mandatory gates cannot be offset by an aggregate score. Evidence: control-gate rows. Owner: risk executive.
• Business-as-usual, direct improvement, hold and stop alternatives remain visible. Evidence: comparison record. Owner: independent challenger.

Governance, privacy, security and risk

• Evidence access, minimisation, retention and custody are approved. Evidence: information approval. Owner: information authority.
• Affected customer and employee groups and distributional harms are considered. Evidence: impact review. Owner: qualified control owner.
• Privacy, security, legal, employment and regulatory conclusions have qualified owners. Evidence: referral log. Owner: executive sponsor.
• Consequential exceptions have named human authority and fallback. Evidence: control design record. Owner: operating owner.
• Material incidents bypass portfolio cadence for immediate containment. Evidence: incident route. Owner: risk executive.

Deliverables and acceptance

• All four portfolio working files are complete and linked by stable IDs. Evidence: working-pack index. Owner: portfolio lead.
• Each shortlisted opportunity has accepted value, feasibility, risk, dependency and learning fields. Evidence: portfolio matrix. Owner: decision owner.
• Shared dependencies have owners, capacity and acceptance conditions. Evidence: dependency register. Owner: CIO.
• The sequence protects normal operations and mandatory control work. Evidence: accepted capacity plan. Owner: COO.
• Every next gate states output, owner, acceptance, exclusion and stop condition. Evidence: decision register. Owner: downstream owner.

Measurement

• Learning, delivery, operating, commercial outcome and guardrail measures remain separate. Evidence: measurement register. Owner: benefit owner.
• Every measure has definition, baseline, source, owner, cadence and response threshold. Evidence: metric contracts. Owner: measurement owner.
• Portfolio totals do not combine unlike outcomes or proof states. Evidence: aggregation rules. Owner: CFO.
• Capacity utilisation uses owner-accepted available capacity rather than theoretical hours. Evidence: capacity baseline. Owner: capacity owner.
• Attribution limits and concurrent changes are visible at review. Evidence: review pack. Owner: benefit owner.

Decision and handover

• Every eligible opportunity receives one controlled disposition. Evidence: signed portfolio decision. Owner: decision secretary.
• Resource commitments are recorded separately from dispositions and ranking. Evidence: resource rows. Owner: economic buyer.
• Conditions, dissent, conflicts and rejected alternatives accompany each decision. Evidence: decision pack. Owner: portfolio chair.
• Each receiving owner accepts the next-gate brief and evidence limits. Evidence: handover receipt. Owner: downstream owner.
• Architecture, pilot, production and scale work route to their separate standards. Evidence: handoff register. Owner: portfolio lead.

Review, change and stop controls

• Evidence, baselines and decisions have expiry or event triggers. Evidence: review schedule. Owner: evidence custodian.
• Portfolio capacity and dependencies are reconciled at each material change. Evidence: rebalance record. Owner: COO.
• Held and deferred opportunities have review conditions rather than silent backlog status. Evidence: hold register. Owner: opportunity owner.
• Stopped opportunities close access, claims and supplier activity. Evidence: closure record. Owner: decision owner.
• Material outcome or boundary changes require a new charter. Evidence: re-charter decision. Owner: executive sponsor.

Measurement contracts

Measure A: Eligible opportunity evidence coverage

Measure B: Next-gate acceptance

Measure C: Portfolio capacity commitment

Measure D: Accepted portfolio outcome movement

Measure E: Decision learning resolved

Measure F: Material portfolio harm or control exception

Evidence classifications used in this guide

• VERIFIED: supported by retained decision-fit organisational evidence or an authoritative source within its stated boundary.
• INFERRED: a reasoned interpretation linked to evidence with alternatives and limitations visible.
• PROPOSED: an opportunity, option, range, sequence or method awaiting authorised acceptance.
• REQUIRES VALIDATION: a material claim or condition that cannot yet support funding or sequencing.

Engagement records distinguish observed fact, corroborated interpretation, single-source statement, assumption, evidence gap and constraint. Portfolio consensus does not change proof status.

Sources and evidence notes

1. HM Treasury: The Green Book 2026 supports objective-led options, business-as-usual comparison, costs, benefits, risks and visible uncertainty. VERIFIED official public-sector guidance; local applicability requires judgement.
2. HM Treasury: Multi-Criteria Decision Analysis in options appraisal supports transparent evidence and assumptions when criteria assist option comparison. VERIFIED official source; this guide does not treat MCDA as automatic authority.
3. NIST: Artificial Intelligence Risk Management Framework 1.0 provides voluntary Govern, Map, Measure and Manage risk outcomes for AI systems. VERIFIED official source; it applies only where AI is intrinsic.
4. NIST: AI RMF Core provides official companion detail on AI risk-management outcomes. VERIFIED official source at the source cut-off.
5. OECD: AI Principles supports accountability, transparency, robustness, security and safety for AI actors. VERIFIED intergovernmental source; it does not determine a specific organisation's portfolio decision.
6. U.S. GAO: Technology Readiness Assessment Guide supports evidence-based technology readiness assessment in acquisition contexts. VERIFIED official source; this portfolio does not substitute that method for local feasibility evidence.

The five-lens model, six portfolio dispositions, common evidence contract, steps, checklists and visual structures are Abrasim-created working methods. They are PROPOSED, not proof of a return or externally certified standard.