Business Unit Goals

10 min read

Chapter 7 established the minimum goals required to operate a software dark factory: deliver verified software, improve the factory, and preserve continued operation. An autonomous business unit keeps those goals, but changes their place in the hierarchy.

The factory goals describe how a software-production capability should operate. Business-unit goals describe which outcomes the organisation wants that capability, and its other capabilities, to produce. This is why goals are reintroduced here rather than merely carried forward unchanged.

A composable goal model

The autonomous business unit adds a governed portfolio above the factory’s minimum goals. That portfolio separates five kinds of goals so they can be combined without confusing an outcome with the means used to achieve it:

Continuity goals preserve the unit’s ability to exist and act: identity, authority, solvency, observability, recovery, and dependable access to world signals.
Business outcome goals define a measurable change in revenue, return, cost, customer value, risk, or market position within a time horizon.
Capability goals acquire, build, or improve something the unit needs in order to pursue a business outcome. The software factory’s delivery and improvement goals live here.
Operating goals direct authorised interaction with customers, markets, suppliers, financial systems, and other parts of the real world while preserving security and compliance.
Optimisation goals make successful outcomes cheaper, faster, safer, or more reliable and help the unit adapt to relevant external developments.

Business goals also require broader evidence. A merged change is a factory outcome, not proof of a business outcome. The unit must connect delivery evidence to authorised world signals and measures that show whether the intended external condition changed.

Goals are therefore part of the system’s operational identity. They tell agents which changes in the world count as progress, give the trigger and planning layers a durable basis for creating work, and provide measures against which outcomes can be evaluated. Axioms constrain how those goals may be pursued.

Each goal should be expressible as a small contract:

Outcome: what condition should change.
Measure: how the unit will recognise progress and success.
Target and horizon: how much change is required, and by when.
Constraints: risk appetite, budget, axioms, and actions that remain prohibited.
Authority: what the unit may decide or execute without escalation.
Signals: which observations can start work or cause the goal to be reconsidered.
Dependencies: the capabilities required to pursue and measure the outcome.

The unit can then compose goals rather than flatten them. For example, it can pursue a revenue target as a business outcome, create a billing integration as a capability goal, operate a sales channel through an operating goal, and require a minimum cash runway through a continuity goal. Every delegated task should trace back to at least one goal, while actions that affect several goals must record the trade-off being made.

Founder intent and the initial constitution

An autonomous business unit does not determine for itself what winning means. Its founders establish an initial constitution that defines the outcomes it should pursue, the constraints it must preserve, and the authority it may exercise. One unit may prioritise near-term profit. Another may accept lower current returns to build a stronger long-term position. Neither preference follows automatically from autonomy; each must be made explicit in the unit’s governing artifacts.

The initial constitution should distinguish four kinds of instruction:

Axioms are durable constraints that actions and goals must preserve.
Goals define the outcomes the unit should produce and the measures used to evaluate them.
Risk appetite defines which uncertainty, loss, and exposure the unit may accept.
Capital-allocation policy defines how it trades current returns against investment in future advantage.

This separation prevents a temporary target from becoming an immutable principle. “Earn a specified profit this year” is a goal. “Preserve solvency” may be an axiom. The acceptable probability and scale of loss belong to risk appetite, while the choice between distributing current profit and reinvesting it belongs to capital-allocation policy.

Founder intent should also define the unit’s initial authority and amendment rules. Some founders may make particular instructions effectively immutable. Others may define conditional policy transitions: when specified evidence satisfies an unlock condition, a new budget, authority, goal, or risk limit becomes eligible. These transitions resemble smart contracts because the permitted state change and its preconditions are defined in advance. They do not need to be implemented on a blockchain.

Each transition rule should record:

the policy currently in force
the proposed successor policy
the evidence required for the transition
the authority permitted to attest that the evidence is sufficient
the authority permitted to ratify or reject the change
the appeal path when authorities disagree
any axioms that cannot be amended through the mechanism

The execution ledger should preserve the proposal, evidence, attestations, decision, and resulting authority. This makes a constitutional change inspectable, but does not allow the unit to legitimise the change through self-attestation. The enforcement and independent-assurance requirements are developed in Defensibility and External Enforcement.

Five stages of autonomy

The stages describe how the goal portfolio develops. They are cumulative: reaching a later stage does not retire the earlier goals. Continuity remains protected while the unit discovers its business direction, builds capabilities, operates in the world, and improves how it produces outcomes.

Stage 1: Bootstrap continuity

The first purpose of the autonomous unit is to survive. Before it can pursue an external business outcome, it must establish a dependable operational identity and preserve the conditions that allow it to continue acting.

Bootstrap goals should establish:

durable mission, founder intent, axioms, goals, risk appetite, capital-allocation policy, authority, budgets, amendment rules, and human recovery paths
persistent state, knowledge, execution history, and auditable decision records
observability of the unit’s health, resources, obligations, and failures
reliable interfaces for authorised world signals, with provenance and access controls
a software factory capable of maintaining the unit’s own software and integrations
redundancy or recovery for dependencies whose loss would stop the unit

The primary success condition is continuity, not revenue. The stage is complete when the unit can detect loss of capability, recover or escalate, maintain its required interfaces, and resume authorised work without losing its identity or history.

Stage 2: Identify business goals

Once continuity exists, the unit can turn its mission and observed environment into explicit business goals. These goals may include a revenue target, a required return within a defined period, a customer or market outcome, a cost boundary, or a risk appetite.

The unit should discover candidate goals from authorised directives, financial plans, commitments, customer evidence, and other governed signals. It may clarify conflicts and propose targets, but high-impact goals and changes to risk appetite remain subject to their human approval rules.

The output of this stage is a prioritised goal portfolio in which every goal has a measure, target, time horizon, constraints, authority, and evidence source. This prevents broad instructions such as “grow revenue” or “reduce risk” from entering execution without a testable meaning.

Stage 3: Build capability

For each ratified business goal, the unit identifies the gap between what it can do now and what the goal requires. It can then acquire, provision, integrate, or build the missing capabilities.

The software dark factory is central to this stage. It can create software, adapters, automations, evaluation surfaces, and operational tools, then improve them through the governed learning loop. Software is still a means rather than the business outcome: a deployed service only counts as progress when it enables a required action or changes the measure attached to the goal.

Capability goals should state which business goals they support, how readiness will be verified, what they cost to create and operate, and when they should be improved, replaced, or retired. This keeps the unit from accumulating capabilities that are technically impressive but strategically irrelevant.

Stage 4: Operate in the world

The final stage activates the unit’s capabilities against real business goals. The unit interacts with customers and counterparties, delivers products or services, incurs costs, collects revenue, fulfils obligations, and responds to changes in its environment.

Operating goals govern recurring activities such as demand generation, sales, delivery, support, billing, treasury, supplier management, and incident response. Security, privacy, compliance, fraud prevention, and contractual limits are part of the operating definition of success rather than separate concerns added after revenue appears.

The unit must close the loop between action and outcome: observe world signals, decide under delegated authority, act through an approved capability, measure the external result, and update its plans. Revenue is real only when it is attributable, collectible, compliant, and durable enough to contribute to the unit’s continuity.

Stage 5: Optimise and continue

Once the unit can repeatedly produce real outcomes, it should improve their economics and preserve its relevance as the outside world changes. Continuance is more than keeping the system online. It means refining successful processes, reducing the resources required per outcome, and recognising developments that create new opportunities or invalidate existing assumptions.

Optimisation goals can target cost per acquired customer, cost per transaction, delivery time, model or infrastructure spend, error and recovery rates, capital efficiency, or the proportion of work completed without intervention. These measures remain subordinate to the business outcome and continuity goals: making an activity cheaper is not progress if it reduces customer value, increases unacceptable risk, or optimises work the unit should no longer perform.

The unit also needs dedicated capabilities for learning about its environment. A researcher agent can monitor relevant research, tools, techniques, and vendor capabilities. A current-events agent can track market, customer, competitor, regulatory, economic, and security developments. Their findings enter through the same governed signal and knowledge systems as other evidence, with provenance, freshness, access controls, and confidence attached.

External developments should create hypotheses rather than silently rewrite the unit’s behaviour. The unit can rank ideas by expected value, cost, strategic fit, and risk; test promising changes in a controlled evaluation or limited rollout; compare them with an established baseline; and promote or reject them through the governed learning loop. This prevents continuous improvement from becoming continuous novelty chasing.

The stage succeeds when the unit can repeatedly lower the cost or improve the quality of valued outcomes, detect material changes in its environment, and incorporate proven advances without sacrificing continuity, authority, security, or accountability.

How the stages combine

A mature business unit usually carries goals from all five stages at once:

Remain viable while preserving recovery, authority, and signal access.
Reach a business target within an explicit horizon and risk appetite.
Create the missing capability required to pursue that target.
Operate the capability in the real world and measure whether it produces the intended outcome.
Improve and adapt by reducing outcome costs and testing relevant external developments.

This composition is the core control structure of the autonomous business unit. It ties survival to strategy, strategy to capability, capability to evidence from the world, and evidence to continuous adaptation.