Sources of power skill

Sources of Power

Table of Contents

• Choosing a Source of Power
• Leverage
• Proximate Objectives
• Chain-Link Systems
• Design-Type Strategy
• Focus
• Using Advantage
• Stacking Sources

Choosing a Source of Power

A guiding policy without a source of power is a wish. Before formulating policy, identify which asymmetry the strategy will run on:

Source	Core question	Reach for it when
Leverage	Where does a unit of effort buy the most result?	You can anticipate others' behavior or see a pivot point
Proximate objectives	What can we actually hit from here?	Ambiguity is high; teams stall on grand targets
Chain-link	Which link caps the whole system?	Effort everywhere, results nowhere
Design	What configuration of pieces wins?	Stakes are high and resources scarce
Focus	Which narrow target can we dominate?	Rivals serve everyone adequately, no one well
Advantage	What asymmetry do we already hold?	You have something rivals structurally cannot copy

Leverage

Strategic leverage comes from three multiplied ingredients: anticipation, a pivot point, and concentration. Each alone helps; together they let a small force move a large outcome.

Anticipation. Most behavior is predictable: customers keep their habits, incumbents defend their margins, platforms follow announced roadmaps, regulators follow published calendars. You do not need to forecast the future — you need to notice commitments already made. When a major OS vendor announces privacy changes 18 months out, every business built on third-party tracking will predictably scramble; a startup that builds the first-party-data alternative before the deadline harvests the scramble. Practice: keep a one-page register of "announced futures" — platform deprecations, regulation effective dates, rivals' public commitments — and ask which one your roadmap exploits.

Pivot points. A pivot point is a place where effort is amplified — where a small, feasible change releases a disproportionate result. Signatures of a pivot point: a constraint shared by all rivals that you alone can relax; a moment of peak user emotion (first run, failure, renewal) that nobody has designed; an imbalance between how much something matters to customers and how little attention it gets. Finding them is empirical: walk the customer journey and rank each step by (impact if 10x better) ÷ (cost to make it 10x better).

Concentration. Effects become visible only past a threshold. Marketing in five channels at 20% intensity each typically produces nothing measurable in any of them; one channel at 100% breaks through, and the visible win compounds — it attracts talent, references, and belief. Concentration is painful precisely because it makes you absent elsewhere; that pain is the cost of crossing the threshold. Rule of thumb for early-stage products: if you cannot name the segment where you are over-investing relative to its current size, you are spread below threshold everywhere.

When to use: whenever you can know or decide something rivals haven't — leverage is the default source of power for resource-poor attackers.

Proximate Objectives

A proximate objective is a target close enough that the team can see how to hit it. Leaders who hand down objectives as hard to reach as the original challenge ("become the category leader") have not done their job; resolving ambiguity into feasible targets is the job.

Calibrate distance to ambiguity:

Ambiguity level	Situation	Right objective distance	Example
Low	Stable market, known playbook	12-month outcome targets	"Grow EU self-serve revenue 40%"
Medium	Known direction, unknown mechanics	One-quarter capability targets	"Stand up a working PLG funnel with 3 instrumented cohorts"
High	New market, new tech, shifting rules	2-6 week ambiguity-resolving targets	"Run 15 problem interviews; decide wedge by March 15"

Deciding the ambiguity away. When the Surveyor program could not design a Moon lander because no one knew the lunar surface, a JPL engineer wrote a specification — firm enough to land on, specific slopes and granularity — and the program designed against it. The spec was a judgment, not a fact, but it converted an undesignable problem into an engineering problem. Product equivalent: when the team stalls on unknowables ("what will enterprise buyers require?"), write the assumption down as a design spec, date it, and build against it. Revise the spec when evidence arrives; never stall waiting for certainty.

Writing a proximate objective: one owner; a done-test a outsider could verify; a deadline inside one planning horizon; and the team's honest answer to "can you see how?" is yes. Cascade them: each accomplished proximate objective creates new capabilities, which bring previously blue-sky targets into proximate range — strategy as a ladder, not a leap.

When to use: always, as the bridge from policy to action — and especially when smart teams are stalled, because stalling usually means the objective is too far away, not the team too weak.

Chain-Link Systems

A system is chain-linked when overall performance is capped by its weakest link — excellence in one stage adds nothing until the others match. The Challenger was lost to one O-ring. A product funnel is a chain: acquisition → activation → retention → expansion; a 10x improvement in acquisition poured into 15% activation is mostly waste.

Diagnosing a chain-link situation:

1. Map the links — the stages where value is created or lost (funnel stages, or the quality chain: data → model → UX → trust).
2. Set a "good enough" threshold per link, benchmarked against the best rival or user expectation.
3. Measure each link against its threshold. A chain-link diagnosis holds when one or two links sit far below threshold while others are at or above it.
4. Confirm the cap: improvements in strong links haven't moved the global metric. That is the signature.

Why chain-link systems get stuck: fixing one link shows no system-level result (the next weak link now caps it), so the effort looks wasted and gets abandoned. Escaping requires a leader who takes responsibility for the whole chain, fixes links one at a time with slack resources, and refuses to judge progress by global metrics until the last weak link is repaired. Tell the team explicitly: "Activation work will not move revenue this quarter — it removes the cap so that next quarter's acquisition work can."

Excellence as a moat: the same logic in reverse. When every link is matched and mutually adapted — IKEA's in-house flat-pack design, catalog, and out-of-town warehouse showrooms each presuppose the others — a rival copying any single link gains nothing, and copying all of them means becoming a different company. Matched chains are among the strongest isolating mechanisms available.

When to use: when effort is everywhere and results are nowhere — and as a build target when you want advantage rivals cannot cherry-pick.

Design-Type Strategy

Some strategies are choices among known positions; the most powerful are designs — premeditated configurations of resources and actions, like designing an aircraft rather than picking one from a catalog. Hannibal's victory at Cannae was not "fight harder"; it was a designed sequence — feigned retreat, enveloping cavalry, anticipation of Roman aggression — every piece presupposing the others.

The three marks of design: premeditation (worked out in advance, not emergent improvisation), anticipation (built on others' predicted behavior), and coordination (parts configured to reinforce each other, accepting trade-offs a menu-chooser would refuse).

The performance/flexibility trade-off. Tight integration extracts maximum performance from limited resources but resists change; loose, modular configurations sacrifice peak performance for adaptability. Choose tight design when stakes are high, resources are scarce, and the competitive window is now — a vertical AI product owning data pipeline, fine-tuned model, and workflow UI will beat an assembled API stack on quality, but will re-platform slower when models leap. Choose modularity when rich in resources or when uncertainty dominates.

When to use: high-stakes plays where you must beat better-resourced rivals — design is how the weak beat the strong, and why second-rate strategies suffice for those with overwhelming advantage.

Focus

Focus is a coordinated set of policies aimed at a narrow target, producing power that rivals serving everyone cannot match without wrecking their own economics. Crown Cork & Seal prospered in the "commodity" can business by configuring everything — plant location, spare capacity, technical service — around short runs and rush orders for smaller customers, and earned pricing power where larger rivals saw none.

Application discipline: pick the segment where the incumbent's strengths are irrelevant or self-blocking; configure multiple policies (product, pricing, support, distribution) on it simultaneously — focus is the coordination, not just the narrowing; and verify the segment either values you enough to pay or leads somewhere larger. A startup "focused" on a segment it serves with the same generic product is narrow, not focused.

Using Advantage

Advantage is rooted in asymmetry — a difference between you and rivals that translates into lower cost or higher willingness-to-pay. Two disciplines:

Advantage is positional, not general. No company is simply "better"; it is advantaged in some contests and disadvantaged in others. And an advantage matters only at the point of contention: nobody cares about your costs except where you actually compete for a customer's decision. Strategy means steering competition toward the contests where your asymmetry decides.

An advantage must be used to be valuable. Like a silver mine, an advantage you cannot extract more value from is sterile. Four ways to work it:

1. Deepen it — widen the gap between value delivered and cost in your existing wedge (the default, most neglected move).
2. Broaden it — carry the underlying asymmetry to adjacent fields: proprietary job-cost data extends from dispatch software into benchmarking and insurance pricing; a beloved character franchise extends from films into parks and merchandise.
3. Create demand for what you are uniquely good at — grow the segment where your asymmetry decides.
4. Strengthen isolating mechanisms — the barriers that stop imitation from eroding returns: patents, brand and reputation, network effects, switching costs, tacit team know-how, exclusive relationships. If returns are good and undefended, your strategy's clock is running.

When to use: whenever you already hold an asymmetry — and as a test of every proposed advantage: if it changes neither cost nor willingness-to-pay at a real point of contention, it is decoration.

Stacking Sources

Real strategies stack two or three sources. The startup in kernel.md stacks concentration (one motion), a pivot point (time-to-first-value), and anticipation (rivals' enterprise distraction). The established platform stacks focus (mechanical trades), advantage (proprietary data), and de-entropy. When reviewing a strategy, name its sources of power explicitly; if you cannot name any, the strategy is effort dressed as insight.