Alignment Is the Constraint

Posted Sep 16, 2025

By Muness Castle 20 min read

tl;dr - Acceleration is seductive. Alignment is scarce. The first multiplies motion. The second creates progress.

Overview

For 30 years, across startups and public companies, I have watched the same failure mode repeat. When results lag, leaders reach for acceleration. They add people, stack more projects, tighten deadlines, and automate. Velocity goes up. Rework, frustration, and missed outcomes go up too. The bottleneck was not delivery. It was alignment.

The Arsenal of Democracy story on EconTalk shows the order of operations in the extreme. Alignment unlocked speed, not the other way around.

Thesis: Unless you have reason to believe otherwise, assume that alignment is the system constraint. Until shared aims, a clear mechanism of action, and fast feedback are in place, more speed produces more rework. The first step is to demystify strategy itself. If alignment is the constraint, then we need a shared, minimal definition of what strategy really is.

History gives us a vivid example of what happens when alignment, not delivery, is treated as the real constraint.

Alignment as the Path to Speed: Lessons from the Arsenal of Democracy

Here’s how alignment in WWII mobilization—the Arsenal of Democracy—showed up as process, not just rhetoric.

By the end of the war, an American assembly line was producing a B-24 bomber in less than an hour. But that success was far from inevitable. (See the EconTalk episode.)

At first, misalignment made success nearly impossible. Factories built aircraft to outdated specifications, suppliers produced parts that didn’t fit, and workers optimized locally with no connection to the larger mission. Incentives and urgency weren’t enough. Without alignment on what the Aim was and how to get there, early output was wasted motion.

What alignment looked like in practice

Concretely, alignment in mobilization looked like coordinated authority, controlled change, fast feedback, enabled labor, and quality counted as output.

Shared aim and command: a single national aim (equip the Allies to win) with coordinating bodies that could resolve conflicts and set priorities across agencies and firms (e.g., War Production Board priority ratings, Office of Production Management, Army/Navy coordination) The WPB “allocated scarce materials, established priorities … and prohibited nonessential production” source.
Change control and standardization: production‑control systems, change‑order pipelines, and interchangeability drives so factories and suppliers built the same version at the same time; modification centers as a buffer while factories caught up. The USAAF used Modification Centers so that updates “avoided disruption to the production lines” while factories caught up to the newest design source.
Feedback from the field: combat and maintenance reports driving design fixes that were pushed back through the change pipeline and, later, incorporated directly on factory lines.
Aligned incentives: cost‑plus contracts and priority access to materials so firms were rewarded for delivery and quality, not just bids; oversight (e.g., Truman Committee) to focus effort and surface bottlenecks. The Truman Committee conducted hundreds of hearings and field trips to uncover waste and bottlenecks, and is credited with saving the government billions through its oversight source.
Workforce enablement: process, tooling, and training redesigned so a rapidly expanded, often inexperienced workforce could do precision work at scale.
Quality as part of output: inspectors and test regimes staffed and funded as core work, not overhead, to ensure “usable, combat‑ready” output.

Alignment enabled adaptation to design change

Wartime designs changed constantly. Alignment made that survivable. Priority systems and centralized change‑order pipelines told factories and suppliers which revision to build and when to switch. Early on, modification centers retrofitted aircraft to the latest spec so assembly lines did not stall; as production control matured, major updates were incorporated directly on the line. The effect was that constant change produced synchronized improvement rather than fragmented variants.

The majority of newly produced combat aircraft were channeled to the modification centers immediately after leaving the production facility … The use of modification centers avoided disruption to the production lines to incorporate continuous improvements or other changes to the aircraft design. –(USAAF Modification Centers)

For example, the P-47 had five major redesigns; the B-29 saw 900 design changes before production stabilized; Chrysler’s tank engine required more than 6,000 modifications. Without alignment, those changes would have fractured the system: one plant building variant A, another stuck on variant B. But because the aim was clear (“combat-ready aircraft that win the war”), the production ecosystem adapted in lockstep. Alignment turned design churn into coordinated improvement instead of chaos.

Alignment enabled operational optimization

Design changes were only half the battle; factories also had to make the work itself executable by a brand-new workforce. Factories didn’t just build faster with existing tools; they redesigned the production system itself. Jigs and fixtures were rebuilt so an untrained workforce (half a million women, many new to industry) could assemble precision aircraft. Rivet guns were counterweighted so smaller operators could use them. Tasks were simplified and redistributed so unskilled workers could contribute without breaking tolerances. Alignment on the aim made it obvious: the constraint wasn’t just labor supply, it was how to make this workforce effective. That clarity drove the retooling.

Additional hydraulic lifts, smaller rivet guns, and other tools quickly put women on an equal footing with their male counterparts. –(Lockheed Martin history of Rosie the Riveter)

These were not ad hoc shop‑floor tweaks; they were coordinated programs of training, tooling standards, and job redesign that flowed from a shared aim and were funded and inspected as such.

Alignment enabled quality at scale

Alignment defined output to include reliability. Leadership treated quality as part of the aim, so factories staffed inspection and test as primary work. High-speed production would have been worthless if engines failed in the field. At Ford’s Willow Run plant, 3,000 of 15,500 workers were inspectors (20% of the workforce). That allocation only makes sense if everyone is aligned on the mechanism: output without quality is not success. By aligning quality control with the overall aim, the system could scale without collapsing under its own defects. Because quality was aligned to the aim, managers could devote scarce labor to inspection without being penalized for lower throughput.

Evidence of inspection-at-scale is abundant in wartime records; for example, contemporary accounts describe multi‑stage tear‑down inspections and stringent tolerances as standard practice in aircraft engine plants (see US wartime production overviews and plant archives; also discussed in EconTalk: Arsenal of Democracy). Even so, getting to this level of reliability took time and painful iteration.

Alignment enabled persistence through ramp-up failure

Even with all of this, Willow Run “for the first two years … produced virtually nothing.” Governance and oversight kept the focus on fixing the mechanism rather than abandoning the mission. Leaders had the authority and political cover to retool layouts, retrain crews, and resequence work until the line flowed.

Interwar planners expected mobilization to require just 18 months, [but] it took between two and three years to reach full production for many items. –(War on the Rocks)

A misaligned system would have abandoned the effort or doubled down on brute force. Instead, alignment allowed persistence: redesign the equipment, retrain the workforce, re-sequence the flow until the mechanism worked. Once alignment was locked in, speed finally compounded and eventually produced bombers every hour by war’s end. The system persisted because leaders and factories were aligned on the end goal and empowered to act on it.

Alignment enabled massive productivity

Once those fixes stuck and the mechanisms ran together, output did not just recover; it compounded. Once these alignment mechanisms were in place and working together, output did not just meet need: it overshot it. Central priority systems, common definitions of done, and closed feedback loops removed cross‑purpose work and created compounding throughput. By late 1944, aircraft, tanks, and munitions arrived faster than the front could consume them.

At Willow Run, peak throughput was famously reported as a bomber about every hour, a pace only achieved after alignment and production control were in place (The Henry Ford Museum).

That surplus capacity meant supply was never the limiting factor again. Once the loop of Aim, Mechanism, and Feedback was firmly in place, productivity became not just adequate but overwhelming.

When alignment is in place, speed compounds. But when it is missing, speed seduces, and wrecks.

When speed outruns alignment

Doing more feels good because it is visible and measurable. But speed without direction is theater. Feature factories and splashy flameouts share a cause: acceleration eclipsed alignment. Teams adopt “best practices” (sprints, OKRs, microservices, AI copilots) without saying how they move the Aim. No mechanism, no progress.

This is the same trap I described in User Value Comes First: teams chase financial proxies or feature output, but when those are not aligned to user value, churn and wasted effort compound. If this rings true, the constraint is not delivery; it is alignment.

Common misfires

Speed and execution that outrun alignment create expensive noise, not outcomes. The patterns below show up everywhere.

More, faster: assumes delivery is the constraint. Result: more of the wrong work.
Metrics without a mechanism: numbers move, outcome does not.
Snapshots without flow: counts entries in each phase of a funnel, but ignores cycle time or path distribution.
Business as usual: new aims, old routines. As Roosevelt warned, “This can only be accomplished if we discard the notion of ‘business as usual.’”

More output is not progress if it fails to change user behavior. Outcomes (not feature counts) tell you whether the mechanism is working.

What it looks like on product teams

Many product teams I’ve observed experience the feature factory syndrome: rapidly releasing a flurry of features that seem impressive, only to find they do not make a dent in business outcomes. This often happens when teams operate with implicit goals or conflicting interpretations of strategy. Without a unifying north star, velocity turns into Brownian motion: lots of busyness, no progress.

Teams end up working on things that do not matter, and effort is wasted because the right hand and left hand are not coordinated. Another common pitfall is when leadership declares an urgent deadline (“We must ship by Q4!”) and teams crunch to hit it, but in the rush they skip the due diligence of cross-functional alignment. The product might ship on time, yet sales, marketing, or support were not prepared and the launch flops. These failures reinforce that speed is not a virtue in isolation—it must be channeled toward a clear, shared aim. See also: John Cutler on Feature Factories and Amplitude’s follow-up.

Many failures in tech can be traced to teams focusing on delivery velocity while losing sight of strategic coherence. One stark example was the short-form video platform Quibi. Flush with $1.75 to $2.0 billion in funding, Quibi raced to deliver a splashy product launch in under two years. In the rush, the offering was misaligned with actual user behavior and needs—expensive, 10-minute exclusive shows built for mobile at a time when short-form, user-generated video dominated. Founders later wrote that Quibi failed “either because the idea itself wasn’t strong enough … or because of our timing” (open letter coverage). Post-mortems highlighted the demand mismatch and format rigidity (WSJ; The Guardian). In other words, Quibi moved fast on building something, but it was not aligned to a real market demand. Result: a highly optimized delivery of the wrong strategy.

If this looks familiar, it is because most orgs optimize the wrong constraint. Critical Chain gives a language for choosing the one place where speed actually helps.

A Critical Chain Lens

Eli Goldratt’s view is that every system has a constraint. Optimizing anywhere but the constraint is waste. Leaders often assume the constraint is delivery. More engineers, tighter sprints, more automation, more AI. In my experience, the constraint is usually alignment:

Without shared aims, teams push in different directions.
Without mechanisms, work cannot be validated.
Without feedback, confidence drifts away from reality.

In Real-World Application of Strategic Clarity, I showed how platform teams demonstrated this in practice: their real constraint is usually alignment with the orgs they serve. When alignment is the constraint, adding speed multiplies misaligned work. Faster drift, more rework, more burnout.

Indeed, misalignment is very expensive. One global survey estimated around $1 million wasted every 20 seconds, roughly $2 trillion a year, due to ineffective implementation of strategy (PMI 2018 Pulse).

Strategy, made small

Here’s the smallest working fix when alignment is the constraint.

Strategy does not need mystique. At its core, strategy is three things:

Aim: the outcome you want to create
Mechanism of action: the causal lever you believe will move that outcome
Feedback: the signal that will validate or disprove the mechanism

Feedback acts as a real-world sufficiency test: are our tactics actually enough to move the Aim?

An Aim is not just an internal target like “ship features” or “increase revenue.” It should anchor to real user outcomes. As I argue in User Value Comes First, growth comes from delivering genuine value, not from chasing profit-first outputs. As such, an Aim should be phrased as the change in user behavior we seek, not as internal outputs. As I argued in Outcomes Over Output, outputs are deliverables; outcomes are the user behaviors and benefits that prove value.

Example:

Aim: increase retention
Mechanism: personalized onboarding builds early confidence, which reduces early churn
Feedback: 30-day activation rate, plus time to first value

Quick Reference

Element	Question to Ask	First Artifact
Aim	What outcome are we trying to create?	One sentence with a definition of done
Mechanism	Why do we think this will work?	A causal “because” statement
Feedback	How will we know fast if we are wrong?	A proxy metric + stage gate

If they cannot name the Aim, teams end up staring at the next ticket instead of knowing where they are headed. If the team cannot name the Mechanism, the team cannot steer; they might pull levers that conflict or do not connect to the outcome. If the team cannot name the Feedback, the team cannot learn; they will repeat work without getting closer to the Aim. And if teams only watch metrics without a mechanism, it is just scoreboard watching with no playbook behind it. See the one‑page alignment worksheet at the end of this post for teams to fill out together.

Naming the Aim, Mechanism, and Feedback gives a team strategy in miniature. But to make it real, teams need mechanisms that institutionalize those choices.

Mechanisms that connect strategy to action

A brilliant strategy on paper means little without a concrete path to implement it. Pragmatic leaders define mechanisms that link goals to daily execution. As Amazon puts it, “Good intentions don’t work, mechanisms do”. In practice, a mechanism is a repeatable process or tool that is adopted by the team and regularly inspected for effectiveness (AWS Operational Readiness Review).

Two processes teams can borrow:

Working Backwards (PR/FAQ): write the press release and FAQ before building to force clarity on customer value and assumptions (Working Backwards).
Weekly Business Review: review controllable input metrics weekly to course-correct early (Commoncog on Amazon’s WBR).

Clear, structured strategy documents are also mechanisms. As I described in Documenting Strategy, tools like S&T Trees help ensure every tactic is necessary, viable, sufficient, and connected, the hallmarks of alignment.

Without such mechanisms, even well-intentioned teams drift. Studies and reviews estimate about two thirds of strategic efforts fail in execution, not because the strategy is bad but because the “how” breaks down (HBR; Fortune summary). Good strategy names the how, and great organizations deliberately build mechanisms (frameworks, processes, incentives) to carry that how into effect.

A compact alignment template and playbook are included at the end of this essay to help teams translate these mechanisms into practice. Treat each mechanism slice as an MVP in Seiden’s sense: the smallest thing you can do to learn if the hypothesis is correct (Outcomes Over Output).

Even the best mechanisms fail if they target the wrong problem. Validate the constraint before you scale speed.

Limits, tells, and when to add speed

This is not an anti‑productivity screed. Flow efficiency, tooling, automation, and AI assistance absolutely matter once the aim, mechanism, and feedback loop are in place. But starting with productivity sends teams in the wrong direction:

Proxy worship. Local throughput goes up, but the outcome does not move. Teams celebrate cycle‑time wins while the user behavior we care about is unchanged.
Premature optimization. We tune the non‑constraint. Effort piles onto dashboards, tooling, and process tweaks while the real bottleneck, shared aims and a testable mechanism, remains untouched.
Narrative drift. People get very good at shipping something and begin to mistake activity for progress. The organization’s story shifts from Why this? to How fast?

If you lead with productivity, you can optimize your way into the ditch. Start by aligning on the right thing to do (Aim), why this should work (Mechanism), and how we will know quickly if we are wrong (Feedback).

Then, by all means, pour on the speed: fix flow using smaller batches, introduce visible WIP limits, reduce handoffs, and apply automation where it trims cycle time. The WWII story makes the same point: massive output only came once the true alignment constraints were solved.

Once the loop runs, LLMs amplify alignment by codifying context, closing the loop faster, and clarifying decisions. They are multipliers, not a substitute for Aim and mechanism.

If you are excited about LLMs as I am, rejoice: LLMs can be excellent tools for alignment, feedback, and communication when pointed at the right problems.

Codifying the mechanism:

Teams can use an LLM to draft the One-page Alignment Sheet from meeting notes and docs. Have it normalize team jargon and map terms to the metric tree and definitions.
Connect to shared data sources through a semantic layer so queries reference the same facts and names. As I noted in Operational Data Primer, consistent vocab and event models are what keep mechanisms and metrics in sync across teams.

Close the loop faster:

Let an agent fetch the latest proxy signals each week and propose a short review note: what moved, what did not, and why. Treat it as a first draft, then decide as a team.
Replace dashboard sprawl with task-specific agents that answer the on-the-spot questions people actually have. Use them to check whether the mechanism is working in the flow of work.
Be careful with junk signals. Tickets and ad hoc tags rarely form an objective rubric on their own. You still need clear definitions of done and trusted data.

Communicate context:

Generate decision logs, briefs, and user-ready explanations that anchor to Aim, Mechanism, Feedback, and Guardrail.
For complex orgs, consider role-specific agents that arrive with the organization’s vocabulary and patterns. They carry context so people can focus on the judgment calls.

LLMs multiply value once the alignment loop exists. Aim and mechanism first, feedback second, then add speed.

Closing

Aim more, plan less, communicate better, learn faster. When the mechanism is explicit and the feedback is fast, speed becomes a multiplier, not a mirage. Alignment is scarce; acceleration is everywhere. Choose the constraint on purpose.

This piece sits inside a larger, growing field guide: start by defining an Aim as a change in user behavior, not a list of features or near term revenue targets, as in User Value Comes First and Outcomes Over Output. Write down the Mechanism and connect tactics so teams share the same why, what, and how; use Strategy and Tactics Trees to keep tactics necessary, viable, sufficient, and connected. Instrument Feedback with flow signals and shared definitions so teams learn fast and adjust together, as in Operational Data Primer. To see the loop running in practice, see Real-World Application of Strategic Clarity. For a personal- and team‑friendly antidote to speed‑first optimization, use Open Horizons: Aim. Do. Reflect..

Worksheets and Playbook (click to expand)

Alignment, on one page

One crisp Aim with a definition of done
One plausible Mechanism that names the causal lever
Two or three Feedback signals, including at least one proxy metric you expect to move first
A small stage gate: what we will learn in the next two weeks, and a decision rule for continuing or changing course

Guardrails

Protect the user with a simple SLI or SLO while you experiment
Favor short cycles over big pushes. Planning matters, the plan changes

One-page Alignment Sheet

Field	Description
Aim	A clear statement of the outcome you want to create.
Mechanism	“We believe __ will move __ because ___.”
Feedback signals	Two to three metrics: one proxy, one outcome, and one leading indicator.
Stage gate (2 wks)	“We will continue if ___.”
Guardrail	One SLI or SLO to protect users while testing.
Owner & cadence	Person responsible • Weekly 15-minute review on ___.

Mini playbook

Use this to upgrade alignment before you add speed.

1. Coauthor a metric tree for the Aim

Name the outcome and its leading indicators
Write one sentence that links mechanism to those indicators

2. Ship a mechanism slice

Build the smallest change that would move the leading indicator
Decide in advance what you will learn and what would make you stop

3. Close the loop, fast

Review the signals weekly. Ask what changed, what did not, and why
Capture the decision and the rationale in a short log

4. Add guardrails

Choose one reliability or quality signal to watch while you experiment

5. Only then add speed

When the loop is working, scale with automation, staffing, or AI

Run this in 30 minutes

Prep (10 min): Write one Aim and one Mechanism sentence.

Aim: the outcome in plain words.
Mechanism: “We believe X will move Y because Z.”

Working session (15 min):

Sketch a 3-node metric tree: Aim → 2 leading indicators → 1 proxy you expect to move first.
Define a two-week stage gate: “We will continue if proxy ≥ X by week 2.”

Commit (5 min):

Create a 3-line decision log entry: Date • Aim • Mechanism • Gate.
Schedule a 15-minute weekly check to review the proxy and decide: continue, change, or stop.

This post is licensed under CC BY 4.0 by the author.