You know you can delegate. You do not know what to delegate.
L-0522 established that delegation targets include systems, tools, and automated processes — not just people. That lesson expanded who or what can receive your work. But it left the harder question unanswered: which work should you actually hand off?
Most people default to one of two broken heuristics. The first: delegate whatever you dislike. This feels rational but selects for comfort, not strategic value. You end up keeping pleasant busywork while offloading challenging tasks that would develop your most important capabilities. The second: delegate nothing until you are drowning. This is not a strategy. It is triage under duress, which produces panicked handoffs with no specification, no verification, and predictable failure.
A delegation decision framework replaces both heuristics with explicit criteria. It gives you a repeatable method to evaluate any task against defined dimensions and produce a clear routing decision: do it yourself, delegate it to a person, delegate it to a system, automate it, or eliminate it entirely. The framework does not remove judgment. It structures judgment so that you apply it consistently rather than reactively.
The economic logic: comparative advantage applied to your attention
The theoretical foundation for delegation decisions is older than management science. David Ricardo formalized the principle of comparative advantage in 1817, demonstrating that even when one party is more productive at everything, both parties benefit when each specializes in the activity where their relative advantage is greatest. Ricardo's original illustration used two workers who could both make shoes and hats — even if one worker was faster at both, total output increased when each specialized in the product where their efficiency gap was largest.
Applied to your cognitive work, the principle is precise: you should not do a task simply because you are good at it. You should do a task only when your relative advantage over all available alternatives — other people, systems, AI tools — is highest for that task compared to other tasks competing for the same hours. A CEO who is also the best typist in the company should not type memos. The opportunity cost — strategic decisions not made while typing — exceeds the quality gain of slightly better memos.
This is where most delegation advice fails. The popular "70% rule" — delegate if someone else can do it 70% as well as you — captures the spirit but misses the structure. It treats each task in isolation instead of comparing tasks against each other for the same scarce resource: your attention. The question is not "can someone else do this adequately?" The question is "given everything competing for my time, is this where my relative advantage is highest?"
Comparative advantage also explains why delegation decisions must be dynamic. Your advantage shifts as your skills develop, as your team grows capability, as new tools emerge. The task you should keep today may be the task you should delegate next quarter — not because it became less important, but because your advantage moved to a higher-leverage activity.
Three dimensions that determine delegation fitness
Effective delegation frameworks decompose the decision into independent dimensions that can be evaluated separately and then combined. After analyzing the major models — the Eisenhower Matrix, the Skill-Will Matrix from Hersey and Blanchard's situational leadership research, and contemporary automation evaluation frameworks — three dimensions consistently emerge as the structural drivers of delegation decisions.
Dimension 1: Irreversibility. How costly is it to fix a poor outcome? Some tasks are easily correctable: a formatting error in a report, a scheduling conflict, a data entry mistake. Others are difficult or impossible to reverse: a strategic commitment, a public communication, a relationship-defining conversation. Irreversibility determines how much quality risk you can tolerate in delegation. High-irreversibility tasks demand either that you retain them or that you delegate only to agents with demonstrated competence and provide tight verification loops.
The Eisenhower Matrix captures one facet of this through its urgency-importance grid. Tasks in the "important but not urgent" quadrant often have high irreversibility — strategic planning, relationship building, skill development — which is precisely why they are dangerous to delegate thoughtlessly. The matrix's instruction to "schedule" these tasks is incomplete. The real instruction is: evaluate whether the irreversibility of the outcome requires your direct involvement or whether a well-specified delegation can preserve quality.
Dimension 2: Identity-centrality. Does this task develop a capability that defines who you are becoming? Some tasks are identity-constructing: they build skills, relationships, or judgment that compound over time and shape your trajectory. Others are identity-neutral: they must be done but do not contribute to your growth edge. Writing the first draft of a strategic framework is identity-constructing for a strategist. Formatting that framework into slides is identity-neutral.
The Skill-Will Matrix, developed from Hersey and Blanchard's situational leadership research in the 1970s, approaches this from the delegation receiver's perspective — matching delegation style to the recipient's competence and motivation. But the same logic applies to you as the delegator. When a task sits at your growth edge — where your skill is developing but not yet automatic — keeping it builds capability. When a task sits well below your growth edge — where your skill is fully developed and the work is routine — keeping it consumes time without generating growth. Identity-centrality is not about whether a task matters. It is about whether doing it yourself generates compounding returns.
Dimension 3: Cognitive uniqueness. Does this task require judgment, context, or creativity that only you possess? Some tasks demand deep contextual understanding — knowledge of a client's unstated preferences, awareness of political dynamics in an organization, intuition built from years of pattern recognition. Others require execution of well-defined procedures that any competent agent can follow. Cognitive uniqueness determines the feasibility of delegation: tasks high in cognitive uniqueness are expensive to specify, because the knowledge required to do them well lives in your head in forms you cannot easily externalize.
This dimension connects directly to John Sweller's cognitive load theory. Sweller established that working memory is limited in both capacity and duration — when exposed to excessive information demands, cognitive overload degrades performance. Tasks high in cognitive uniqueness consume germane cognitive load — the effortful processing that builds schema and generates insight. Tasks low in cognitive uniqueness consume extraneous cognitive load — processing effort that produces no learning or growth. Delegation is, at its core, a strategy for managing your cognitive load budget: offload extraneous load to free capacity for germane load.
The Delegation Resistance Score: a practical scoring method
Combine the three dimensions into a single metric. Score each task on each dimension from 1 (low) to 5 (high). Multiply the scores to produce a Delegation Resistance Score (DRS) ranging from 1 to 125.
| DRS Range | Routing Decision | Action | | --------- | ------------------------------------------ | ---------------------------------------------------------------------- | | 1-30 | Strong delegation candidate | Delegate to a person, system, or AI tool. Write a specification. | | 31-60 | Partial delegation or automation candidate | Delegate components, automate routine elements, retain judgment steps. | | 61-90 | Retain with periodic re-evaluation | Keep for now. Re-score quarterly as capabilities and context shift. | | 91-125 | Core retention | This is your highest-leverage work. Protect it from displacement. |
The multiplication rather than addition is deliberate. A task that scores 5 on irreversibility but 1 on identity-centrality and 1 on cognitive uniqueness (DRS: 5) is still a strong delegation candidate — the high consequence simply means you need better verification, not that you should do it yourself. Multiplication ensures that a task must score high on multiple dimensions to resist delegation.
Consider an example. A team lead scores her weekly tasks:
- Writing performance reviews: Irreversibility 4, Identity-centrality 4, Cognitive uniqueness 5 = DRS 80. Keep.
- Scheduling team meetings: Irreversibility 1, Identity-centrality 1, Cognitive uniqueness 1 = DRS 1. Delegate or automate immediately.
- Reviewing code architecture: Irreversibility 3, Identity-centrality 4, Cognitive uniqueness 4 = DRS 48. Partial delegation — delegate implementation review to senior engineers, retain architecture review.
- Compiling weekly metrics reports: Irreversibility 2, Identity-centrality 1, Cognitive uniqueness 2 = DRS 4. Automate.
- Onboarding new hires on team culture: Irreversibility 3, Identity-centrality 5, Cognitive uniqueness 4 = DRS 60. Retain but create supporting materials that others can deliver for logistics.
The framework does not make the decision for you. It structures the inputs so that your judgment operates on explicit criteria rather than habitual defaults.
The five routing destinations
The DRS score tells you whether to delegate. But delegation is not a single action — it is a routing decision with multiple destinations. Each destination has different specification requirements and failure modes.
Route 1: Do it yourself. High-DRS tasks that sit at your growth edge, require your unique context, and carry consequences you cannot afford to get wrong. The key discipline is not deciding to keep these tasks — that part is easy. The discipline is protecting them from being crowded out by low-DRS tasks you failed to delegate.
Route 2: Delegate to a person. Tasks where human judgment matters but your specific judgment is not required. The specification must define the outcome, the quality threshold, and the constraints — not the method. Dan Sullivan's "Who Not How" framework captures this: once you decide to delegate, the question shifts from "how do I do this?" to "who can achieve this outcome?" The Skill-Will Matrix then helps you calibrate your delegation style to the recipient: high-skill, high-will recipients need autonomy; high-skill, low-will recipients need motivation; low-skill, high-will recipients need training; low-skill, low-will recipients need a different assignment.
Route 3: Delegate to a system. Tasks where consistency and reliability matter more than contextual judgment. Checklists, standard operating procedures, templates, and workflows are system delegation targets. L-0522 established these as legitimate delegation receivers. The specification here is the system itself — the checklist, the template, the documented process. The failure mode is designing the system once and never updating it as requirements evolve.
Route 4: Automate. Tasks where the inputs are predictable, the steps are well-defined, and the outputs are consistent. The FAIR framework from automation research provides a useful filter: evaluate Frequency (how often the task recurs), data Availability (whether inputs are structured and accessible), Input consistency (whether the task follows predictable patterns), and Required accuracy (whether the error tolerance allows automated execution). Tasks scoring high on all four are automation candidates. Tasks scoring low on input consistency or required accuracy need human oversight even if automated.
Route 5: Eliminate. Tasks that score low on all three dimensions and produce no meaningful outcome. These are not delegation candidates — they are deletion candidates. The Eisenhower Matrix's "delete" quadrant captures this: tasks that are neither urgent nor important should not be delegated to someone else. They should not exist. Delegating a worthless task does not create value. It creates waste in someone else's workflow.
The AI routing layer: your Third Brain as delegation infrastructure
Contemporary AI tools add a sixth consideration that cuts across all five routes: which tasks can be partially or fully delegated to AI systems?
The Bessemer Venture Partners framework for operationalizing AI distinguishes four automation levels that map directly to the DRS. Full automation applies to tasks with DRS below 10 — mechanical, low-consequence, predictable-input work like data formatting, scheduling, and notification routing. AI-assisted processes apply to tasks with DRS between 10 and 40 — the AI handles mechanical components while you review outputs and provide judgment on edge cases. Human-led with AI support applies to tasks with DRS between 40 and 70 — you do the core work, but AI handles research, summarization, first drafts, and logistical coordination. Intentionally manual processes apply to tasks with DRS above 70 — sensitive conversations, identity-defining creative work, and strategic decisions where the process of doing the work is itself the value.
The practical protocol is progressive delegation: start with the lowest-DRS tasks, route them to AI, verify that quality meets threshold, then move up the DRS scale one task at a time. This is the opposite of the common impulse to hand AI your hardest problems first. Hard problems have high DRS scores — high irreversibility, high identity-centrality, high cognitive uniqueness — which means they are the worst candidates for initial AI delegation.
Your Third Brain — the external cognitive infrastructure of tools, systems, and AI agents — becomes the primary delegation receiver for everything below DRS 30. The framework treats AI not as a replacement for human delegation but as a routing option alongside people, systems, and automation. The decision criteria are the same: irreversibility, identity-centrality, cognitive uniqueness. The only thing that changes is the set of available receivers.
The re-scoring discipline
A delegation decision framework that produces static scores is a framework that decays. Your comparative advantage shifts. Your team develops new capabilities. New tools emerge. Tasks that required your cognitive uniqueness six months ago may now be well-specified enough for delegation because you externalized the required context into documentation or training materials.
Build a quarterly re-scoring habit. Re-score each task and look for two signals: tasks whose DRS has dropped (delegation opportunities you are missing out of habit) and tasks whose DRS has increased (areas deserving more of your protected attention). The framework's value is not in the initial sort. It is in the ongoing calibration of your attention allocation to your evolving advantage.
From what to delegate to what to never delegate
The Delegation Resistance Score gives you a structured method to identify tasks that should leave your plate. But the framework raises a question it cannot fully answer: among the tasks that score high, which ones should you protect absolutely — not just retain by default, but commit to never delegating regardless of how capable your team or tools become?
L-0524 addresses this boundary directly. Some tasks are identity-defining in a way that transcends the DRS scoring — strategic thinking, relationship building, and work that constructs the very framework through which you make all other decisions. These are not simply high-scoring tasks. They are categorically different from tasks that happen to score high on the current rubric. Understanding that categorical difference is what separates effective delegation from delegation that hollows out your core capability.
You now have the decision framework. Next, you learn where the framework's boundary is — and why that boundary exists.
Sources:
- Ricardo, D. (1817). On the Principles of Political Economy and Taxation. John Murray.
- Sweller, J. (1988). "Cognitive Load During Problem Solving: Effects on Learning." Cognitive Science, 12(2), 257-285.
- Paas, F., & van Merriënboer, J. J. G. (2020). "Cognitive-Load Theory: Methods to Manage Working Memory Load in the Learning of Complex Tasks." Current Directions in Psychological Science, 29(4), 394-398.
- Hersey, P., & Blanchard, K. H. (1969). "Life Cycle Theory of Leadership." Training and Development Journal, 23(5), 26-34.
- Sullivan, D., & Hardy, B. (2020). Who Not How: The Formula to Achieve Bigger Goals Through Accelerating Teamwork. Hay House.
- Bessemer Venture Partners. (2025). "From Tasks to Systems: A Practical Playbook for Operationalizing AI."
- Eisenhower, D. D. (1954). Address at the Second Assembly of the World Council of Churches. Northwestern University Archives.