Assigning Key Roles and Solo-Work Safeguards

Good to see you again. In the previous lesson, you assigned a run to a lane and set its time and money caps. Those caps make the bet bounded; roles make its decision gates credible. A six-week Lane M cap is not meaningful if the person who proposed the idea can also decide that its weak evidence is “good enough.”

This lesson establishes the governance roles used throughout a Product Factory run: Owner, Challenger, Factory Keeper, and gate-decider. You will also see how these roles relate to the familiar DACI model, and how to preserve the essential safeguards when one person is doing most of the work.

---

Why a gate needs more than a checklist

A product gate is a decision made against precommitted thresholds. It is not a meeting in which the team asks whether it “feels ready.” That distinction creates four separate kinds of work:

1. Someone must move the run forward and assemble the artifacts.
2. Someone must make the strongest honest case for stopping.
3. Someone must have authority to decide whether the written threshold was met.
4. Someone must protect the integrity of the process across many runs.

If one person informally performs all four jobs, predictable biases appear:

• The person who invested effort in the idea is motivated to interpret ambiguous evidence generously.
• A senior sponsor’s enthusiasm can make dissent socially costly.
• The team can quietly change the threshold after seeing disappointing results.
• No one notices recurring process failures across different product runs.

The Factory’s answer is not bureaucracy for its own sake. It is a small separation of duties designed to make STOP a genuine possible outcome.

---

The four Product Factory roles

Owner: advances the run, but cannot clear it alone

The Owner is the named person responsible for proposing the idea and doing, coordinating, or commissioning the work. The Owner maintains the Idea Record, gathers the required evidence, keeps the work within the lane cap, and makes a recommendation at each gate.

The Owner should be able to answer:

• What problem are we investigating?
• What threshold did we commit to before collecting evidence?
• What evidence did we actually collect?
• What is the strongest case to continue?
• What is the strongest case to stop?
• What action follows this gate decision?

The Owner is responsible for the quality and traceability of the work, but not entitled to a favourable outcome. Most importantly, the Owner cannot clear a gate alone. An Owner who is also the only person able to authorize progress has an incentive structure that defeats the gate.

A useful distinction is this:

The Owner does	The Owner does not do
Frames the run and coordinates work	Treat personal confidence as evidence
Collects and tags evidence	Rewrite thresholds after seeing results
Makes a recommendation	Clear their own gate without independent challenge or authority
Records what happened	Hide missed caps, failed tests, or contrary evidence
Reduces scope when necessary	Extend the lane informally to preserve the original idea

Challenger: protects the right to stop

The Challenger is named at Phase 0 and is never the Owner. Their task is not to be generally negative, nor to obstruct delivery. Their job is specific:

At every gate, write the strongest honest case for STOP before anyone writes the case for PROCEED.

The Challenger needs independence more than seniority or specialist expertise. A useful Challenger is someone who loses little if the idea ends: they are not rewarded for launch, have not made the product proposal their reputation, and are able to ask inconvenient questions without career risk.

The Challenger should test claims such as:

• “Are these really twelve relevant customer conversations, or twelve convenient ones?”
• “Did eight people describe the problem unprompted, or did we introduce the language first?”
• “Is this price supported by evidence, or is it an assumption?”
• “Does this feature test a named hypothesis, or are we adding it because it seems standard?”
• “Does the estimate fit the lane, or have we converted an overrun into optimism?”

The Challenger does not fabricate counter-evidence, impose an arbitrary personal standard, or veto work because they dislike the concept. Their case must be honest, grounded in the agreed threshold and available evidence.

The Challenger’s written STOP case is a practical defense against escalation of commitment. Near a decision, organizations often reward confidence and treat dissent as disloyalty. A premortem changes the social rule: looking for failure modes becomes the assigned work.

A short gate premortem can be run in about ten minutes:

1. Assume the team passed this gate and the product failed six months later.
2. Each participant writes a short explanation of why it failed.
3. Group the explanations into evidence gaps, untested assumptions, scope risks, and operational risks.
4. Compare those risks with the actual gate criteria.
5. Record material findings in the STOP case.

The point is not to predict the future perfectly. It is to surface concerns that optimism, hierarchy, or sunk cost might otherwise suppress.

Gate-decider: applies the threshold and chooses the prescribed action

The gate-decider is the person with standing to make the decision recorded at a particular gate. Their role is not to generate evidence or advocate for the proposal. It is to determine whether the prewritten threshold has been met and to select the action permitted by the standard.

For example, at the Phase 3 problem-validation gate, the decider checks the actual figures against the committed Lane M thresholds:

Required check	Threshold	Example actual figure
Relevant conversations	At least 12	12
Unprompted problem recognition	At least 8	7
Revealed cost	At least 3	4
Abandoned alternative	At least 1	2
Quantified value	At least 3	5

Despite four favourable figures, this run fails because the unprompted-recognition threshold was not met. The gate-decider must record STOP, or the explicitly permitted simplification to Lane S where applicable. They cannot silently turn “at least 8” into “7 seems close enough.”

A capable decider:

• has authority appropriate to the spend, risk, and commitment involved;
• sees sourced evidence, not only a polished summary;
• reads the STOP case before the PROCEED case;
• checks actual results against the stated threshold;
• records the decision and its rationale;
• applies the prescribed action when the run fails.

A decider should ordinarily be someone other than the Owner. Separating Challenger and decider is also a strong default: the Challenger argues the STOP case, while the decider judges both cases against the standard.

The decider’s authority is especially important for committed ideas. Where a contract or prior promise means the idea cannot simply be cancelled, the gates change from pure go-or-stop decisions to scope-and-risk controls. Even then, the decider can stop unsafe work, reject excessive scope, or withhold launch until quality requirements are met.

Factory Keeper: owns the system, not the outcome of one run

The Factory Keeper owns the Product Factory standard itself. This person maintains the current version of the process, ensures that Decision and Deviation Logs are being captured, and runs Phase 17 calibration.

The Factory Keeper asks a different question from the gate-decider:

Is our process producing decisions that predict real outcomes?

Their responsibilities include:

• maintaining the standard, thresholds, lane definitions, and changelog;
• ensuring completed and stopped runs are logged, rather than selectively remembered;
• reviewing recurring deviations from the process;
• calculating factory metrics, such as kill rate and lane-overrun rate;
• facilitating quarterly calibration of gates against live-product outcomes;
• versioning changes when repeated evidence shows that the standard is wrong.

The key conflict-of-interest rule is firm:

A Factory Keeper must not calibrate a run they owned.

Otherwise, a person could defend their own earlier judgement by declaring its gate predictive. They may still contribute context or records, but another qualified person should assess that run during calibration.

---

DACI is useful, but it is not the Factory’s complete governance model

DACI is a widely used decision framework: Driver, Approver, Contributors, and Informed. It is useful for making communication and decision authority visible, particularly when many stakeholders are involved.

The Product Factory can borrow DACI’s clarity, but its roles are not identical:

DACI role	Closest Factory equivalent	Important difference
Driver	Owner	The Owner drives the work and artifacts, but may not clear their own gate.
Approver	Gate-decider	The gate-decider makes a threshold-based decision, rather than approving a general plan.
Contributors	Interview participants, subject-matter experts, technical reviewers, legal or security specialists	Contributors supply input or evidence; they do not automatically receive a vote or veto.
Informed	Sponsors, affected teams, customer-facing teams, leadership, support	They receive the decision and rationale, but do not take part in making it.
No direct DACI equivalent	Challenger	The Challenger has a distinct obligation to develop the strongest STOP case.
No direct DACI equivalent	Factory Keeper	The Factory Keeper protects and improves the decision system across runs.

The most important addition is the Challenger. In ordinary DACI use, Contributors may offer concerns, but nothing requires anyone to make the best possible case against the preferred proposal. The Factory makes that work explicit and sequenced: STOP first, then PROCEED.

For a multi-stakeholder discovery effort, a simple stakeholder map helps distinguish people who actively work on the run, people who must be consulted, and people who simply need an update.

---

A workable role assignment for one run

A role is not assigned when a document says “Product” or “Leadership.” Assign a named person, a decision boundary, and an availability commitment.

At Phase 0, add a governance section like this to the Idea Record:

Run: [name and identifier]
Lane: [S / M / L]
Commitment status: [uncommitted / committed]

Owner
- Name:
- Accountable for:
- May recommend decisions: yes
- May clear own gate alone: no

Challenger
- Name:
- Independence basis:
- Must write STOP case first at every gate: yes

Gate-deciders
- Phase 0 and Phase 1:
- Phase 3 to Phase 6:
- Phase 9:
- Phase 11:
- Phase 12:
- Decision authority and escalation route:

Factory Keeper
- Name:
- Owns process version and Phase 17:
- Calibration conflict check:

Contributors
- Named expertise required:

Informed
- People or groups:
- Update cadence:

A Lane M run may use one decider for several gates, provided they have the required authority and were not the Owner. A Lane L programme may require different deciders for risk, investment, security, or launch decisions. The purpose is not to multiply approvals. It is to make the decision right visible before pressure builds.

Consider a hypothetical Lane M idea: a service that helps independent consultants convert client-call notes into a follow-up plan.

Responsibility	Named person	Why this assignment works
Owner	Product manager leading discovery	Coordinates interviews, evidence, prototype, and records
Challenger	Operations lead from another product area	Has no launch target tied to this idea and can inspect claims independently
Gate-decider	Head of Product	Can authorize the next investment stage or stop the run
Factory Keeper	Product operations lead	Maintains the process and later calibrates the gates across runs
Contributors	Security specialist, finance partner, design researcher	Supply expert input; none gains automatic approval rights
Informed	Sales lead, support manager, executive sponsor	Receive decisions and implications without turning every gate into a committee

Notice that the Challenger is not chosen because they are “the most critical person in the room.” They are chosen because they can be candid without being penalised when the answer is STOP.

---

Working solo: preserve independence where it matters most

A solo maker can perform most of the Owner’s work, but cannot honestly become an independent Challenger merely by changing labels. The Factory therefore changes the form of the safeguard, while keeping its purpose.

When working solo:

1. Name yourself as Owner.
   You remain responsible for the work, artifacts, sourcing, scope discipline, and recommendation.

2. State every threshold before collecting evidence.
   Do not decide after interviews that “seven out of twelve was enough after all.” Write the numeric threshold, lane cap, decision date, and falsifier first.

3. Use one outside human at Phases 3 and 11.
   These are the points at which weak problem evidence or weak real-user validation can cause the most expensive error. Give this person the actual evidence, interview notes or recordings where appropriate, actual figures, and your written STOP case.

4. Write the STOP case first at every other gate.
   Do this before drafting the PROCEED case and before revisiting the original concept. Time-stamp the two sections if possible, so the sequence is visible.

5. Preassign a human decision authority.
   The decision may be a manager, client sponsor, advisor, partner, or another person with legitimate authority over the budget or commitment. The Owner should not declare an independent pass alone.

6. Keep AI in an assisting role.
   An AI may help structure a STOP-case template, identify inconsistent figures, or summarise human-collected material with links. It cannot be the Challenger, provide invented evidence, score the gate, or sign the decision. It has no stake to lose if the idea proceeds.

A solo protocol can look like this:

Gate stage	Solo safeguard
Phase 0–1	Thresholds and falsifier written before work; STOP case written first
Phase 3	Outside human reviews evidence quality, participant fit, actual figures, and STOP case
Phase 5–6	STOP case first; designated decider checks the spike, economics, reachability, and concept decision
Phase 9	Independent review of scope, security or data requirements, analytics, and lane fit
Phase 11	Outside human reviews real-user evidence and the two-strike decision
Phase 12	Named decider verifies baselines, tested rollback, support ownership, kill criteria, and review date
Phase 17	Another person calibrates the run if the Factory Keeper also owned it

The outside person at Phases 3 and 11 does not need to be a product expert. They do need enough independence and care to ask whether the evidence actually supports the claimed result. A trusted peer who will read the material is more useful than a prestigious advisor who only hears a five-minute summary.

---

The minimum governance rule set

Before moving beyond Phase 0, confirm these conditions:

• One Owner is named for the run.
• One Challenger is named and is not the Owner.
• A gate-decider is named for each material gate, with authority appropriate to the lane.
• A Factory Keeper is identified for the standard and later calibration.
• The Owner cannot clear their own gate alone.
• The Challenger writes the STOP case before the PROCEED case.
• The Factory Keeper does not calibrate a run they owned.
• Contributors provide evidence or expertise; Informed stakeholders receive updates; neither status silently creates decision rights.
• For solo work, thresholds are precommitted, STOP cases are written first, and independent human review is scheduled at Phases 3 and 11.
• AI assistance is separated from evidence, challenge, scoring, and signatures.

These rules create productive friction at precisely the moments where enthusiasm, hierarchy, and sunk cost would otherwise make the process unreliable.

---

The Product Factory separates progress, dissent, authorization, and process stewardship. The Owner moves the work forward; the Challenger makes the honest case to stop; the gate-decider applies the precommitted threshold and records the action; and the Factory Keeper improves the system across runs. DACI can clarify who drives, approves, contributes, and is informed, but the Factory adds the Challenger and Factory Keeper because both are necessary for disciplined learning.

Next, you will turn this governance structure into a concrete Phase 1 Idea Record: a solution-free problem statement, a named sufferer, a falsifier, and credible alternatives, including doing nothing.