The Decision Constraint
Ross Sylvester, Co-Founder & CEO, Adrata | Feb 2026 | ~10 min read
Picture this. You are a CRO with 10 AI agents and 50 reps. Your agents can research any account in seconds, generate personalized outreach in milliseconds, and execute follow-up sequences at machine speed. Your reps are equipped with buyer group intelligence, competitive positioning, and real-time deal signals. Everything in your revenue engine is executing.
But nothing is directed.
Your Slack is flooded with agent outputs waiting for approval. Your pipeline review has become a triage session where you adjudicate which of 200 AI-recommended actions to authorize. Your frontline managers are bottlenecked on you for account prioritization, discount authority, and messaging strategy. The agents are fast. The organization is paralyzed.
This is the decision constraint. And it is the defining challenge of revenue leadership in the AI era.
The Execution Constraint Is Dead
For fifty years, the constraint in revenue organizations was execution capacity. Can we make enough calls? Send enough emails? Research enough accounts? Create enough personalized content? Every generation of sales technology -- from the auto-dialer to the CRM to the sequencing platform -- was designed to push execution volume higher. The playbook was simple: more activity equals more pipeline equals more revenue.
AI has obliterated that constraint. Execution is now virtually unlimited. An AI agent can research a prospect's entire digital footprint, generate a personalized multi-touch sequence, and execute it -- all in the time it takes a rep to write a subject line. The volume ceiling is gone. A team of 50 reps with AI agents has the execution capacity that would have required 500 reps five years ago.
But something strange happens when you remove a bottleneck. The system does not flow freely. It backs up somewhere else.
Goldratt's Ghost in the Machine
In 1984, Eliyahu Goldratt published The Goal and introduced the Theory of Constraints. His core insight was deceptively simple: "An hour lost at a bottleneck is an hour out of the entire system. An hour saved at a non-bottleneck is worthless." Every system has exactly one constraint that governs its throughput. Improve anything except the constraint and you improve nothing.
Goldratt's Five Focusing Steps are: identify the constraint, exploit it, subordinate everything else to it, elevate it, and repeat. When you break one constraint, another emerges. This is not a failure. It is the nature of systems.
For decades, the constraint in revenue organizations was execution. AI broke that constraint. Now apply Goldratt's logic: what is the new bottleneck?
It is the decision-maker.
When your AI agents can execute 100 actions per hour but each action requires human direction -- which accounts to target, what messaging to use, when to pivot strategy, whether to discount -- the human decision-maker throttles the entire system. The math is unforgiving. If each decision takes 10 minutes of human deliberation, your 100x execution engine operates at 1% capacity. The constraint has shifted from doing to deciding.
The data confirms this at scale. McKinsey found that a typical Fortune 500 company wastes more than 500,000 days per year on ineffective decision-making -- the equivalent of approximately $250 million in annual wages. Executives spend nearly 40% of their time making decisions and believe most of that time is poorly used. These numbers predate the AI era. When execution speed increases by orders of magnitude while decision speed stays constant, the waste compounds.
Boyd's OODA Loop: The Orient Phase Is Everything
John Boyd was a United States Air Force Colonel who studied why American F-86 pilots consistently defeated superior MiG fighters during the Korean War. The MiGs were faster and could climb higher. By every hardware metric, they should have won. They did not.
Boyd developed the OODA loop -- Observe, Orient, Decide, Act -- to explain why. The entity that cycles through this loop fastest does not just win more often. It gets inside the opponent's decision cycle, forcing them to perpetually respond to a reality that has already changed. The competition becomes asymmetric. One side is acting. The other is reacting to actions that are no longer relevant.
Here is what matters for revenue leaders: AI has compressed the Observe phase (data is everywhere, instantly) and the Act phase (execution is near-instantaneous). But Orient -- synthesizing observations through experience, context, and judgment to build a mental model of reality -- remains profoundly human. So does Decide.
Boyd considered Orient the schwerpunkt -- the decisive point. In his more complex diagrams, every feedback loop flows through Orientation. He drew on Godel's incompleteness theorems and Heisenberg's uncertainty principle to argue that any closed system of thought will eventually become mismatched to reality. Competitive advantage belongs to those who detect and resolve that mismatch fastest.
Boyd also identified something he called Fingerspitzengefuhl -- fingertip feel. This is the experienced decision-maker's ability to skip the explicit Decide phase entirely and act from intuition built through deep orientation. He labeled this pathway "Implicit Guidance and Control" and argued that organizations should "emphasize implicit over explicit in order to gain a favorable mismatch in friction and time."
The application to revenue leadership is direct. When your reps and AI agents have been properly oriented -- trained on strategy, steeped in customer context, aligned on competitive positioning -- they can skip the explicit "ask the boss" decision step and act from implicit understanding. The CRO's job is not to make every decision. It is to build the orientation that makes most decisions unnecessary.
Two-Way Doors and the 70% Rule
Jeff Bezos formalized a distinction in his 2016 Letter to Amazon Shareholders that every CRO should internalize. Type 1 decisions are one-way doors: consequential, irreversible, requiring careful deliberation. Type 2 decisions are two-way doors: reversible, low-cost to get wrong, easy to walk back through.
The critical organizational failure, Bezos observed, is that "as organizations get larger, there seems to be a tendency to use the heavyweight Type 1 decision-making process on most decisions, including many Type 2 decisions. The end result of this is slowness, unthoughtful risk aversion, failure to experiment sufficiently, and consequently diminished invention."
Revenue organizations are rife with this. A rep wants to adjust messaging for a specific vertical. A manager wants to test a new outreach cadence. An AI agent has identified an unconventional entry point at a target account. These are two-way doors -- easily reversed if they don't work -- but in most organizations they require approval chains designed for one-way decisions. The entire engine idles while a Slack message waits for a thumbs-up.
Bezos added a second principle: "Most decisions should probably be made with somewhere around 70% of the information you wish you had. If you wait for 90%, in most cases, you're probably being slow." Patrick Collison of Stripe put it more sharply: "If you can make twice as many decisions at half the precision, that's often better."
McKinsey's research validates this empirically. Their studies of 1,200+ managers found that "faster decisions tend to be higher quality, suggesting that speed does not undercut the merit of a given decision." The intuitive fear -- that moving fast means deciding poorly -- is not supported by the evidence. Speed and quality travel together because the discipline required to decide fast is the same discipline required to decide well: clarity on what matters, willingness to act on incomplete information, and a culture that corrects quickly rather than deliberates endlessly.
Barrels and Ammunition
Keith Rabois, the venture capitalist and former executive at PayPal and Square, uses a framework that is brutally relevant to the AI era. Most great people, Rabois argues, are "ammunition" -- they execute with high velocity within a defined framework. What organizations desperately need are "barrels" -- people who can take a vision from concept to completion, making the hundreds of small decisions along the way without escalating each one.
"The bottleneck to progress," Rabois says, "has rarely been resources -- it's almost always been about who can effectively direct those resources."
AI gives you infinite ammunition. An organization equipped with AI agents has functionally unlimited execution capacity. But execution without direction is noise. What you need are barrels -- leaders at every level who can aim the ammunition, who can orient their teams, who can make the operational and tactical decisions that turn raw capacity into revenue.
Your organization's throughput is not limited by how many AI agents you deploy. It is limited by how many barrels you have. And barrels, unlike ammunition, cannot be manufactured by AI. They must be developed, one decision at a time.
Mission Command: Tell Them What and Why, Not How
The Prussian military formalized a concept in 1888 that modern revenue organizations desperately need: Auftragstaktik, or mission-type tactics. The principle is simple. The commander communicates intent -- the objective, the timeframe, the forces available -- and subordinate leaders determine the methods independently. Centralized planning. Decentralized execution.
This works because the commander's intent acts as a decision filter. When a soldier in the field encounters a situation not covered by the plan (which is every situation, because plans do not survive contact with reality), they ask: "Does this action advance the commander's intent?" If yes, act. If no, don't. No radio call required. No approval chain. No bottleneck.
General Stanley McChrystal brought this principle into the modern era when he restructured Joint Special Operations Command in Iraq. His innovation was combining two elements that neither sufficed alone. First, shared consciousness: "extreme, participatory transparency" that gave every team holistic awareness of the entire operating picture. Second, empowered execution: "individuals and teams closest to the problem, armed with unprecedented levels of insights from across the network, offer the best ability to decide and act decisively."
McChrystal's insight was that you need both. Shared consciousness without empowered execution creates awareness without action -- everyone sees the battlefield but waits for orders. Empowered execution without shared consciousness creates action without alignment -- everyone acts but in different directions.
The CRO's version of this is: give every rep and every AI agent access to the full picture -- the same pipeline data, the same competitive intelligence, the same strategic priorities -- and then give them the authority to act on what they see. The CRO defines the what and the why. The frontline determines the how.
The Data on Decision Effectiveness
The relationship between decision-making quality and business performance is not a correlation you can dismiss. Bain's ten-year research program involving more than 1,000 companies found a 95% confidence level correlation between decision effectiveness and business results. Organizations in the top quintile of decision-making scores achieved revenue and earnings growth more than 3x greater than their peers. Their employee happiness scores were 27 points higher on average.
Only 15% of companies practice effective decision-making.
McKinsey's data is equally stark. When respondents say decisions are made at the right organizational level -- often delegated down rather than escalated up -- they are 6.8x more likely to be part of a winning company. Those empowered to make decisions and receiving sufficient coaching from leaders were 3.2x more likely to report that their company's delegated decisions were both high quality and speedy.
The pattern in these studies is consistent. The winning companies are not making better individual decisions. They are making decisions faster, at lower organizational altitudes, with clearer accountability. They are treating decision-making as a system to be optimized, not a privilege to be hoarded.
The losing companies are doing something specific and recognizable: they are escalating two-way door decisions to one-way door processes. They are requiring consensus where commitment would suffice. They are accumulating what researchers now call "decision debt" -- the compounding consequence of postponed, avoided, or poorly structured decisions. Unlike technical debt, which stems from haste, decision debt is born from hesitation. And its cost is not linear. Each day of delay adds scaffolding, workarounds, and entrenched assumptions that make the eventual decision exponentially more expensive to implement.
Andy Grove put it plainly: "Most companies don't die because they are wrong; most die because they don't commit themselves. They fritter away their valuable resources while attempting to make a decision."
The Decision Constraint Framework for the AI Era
Here is what I believe CROs must do. Not incrementally. Structurally.
1. Classify every recurring decision by type.
Map the decisions your organization makes repeatedly. Account prioritization. Messaging strategy. Discount authority. Competitive positioning. Pipeline management. For each one, ask: is this a one-way door or a two-way door? The honest answer, in most revenue organizations, is that 80% or more of decisions are two-way doors being processed as if they were irreversible. Reclassify them. Push two-way door decisions to the lowest competent altitude.
2. Assign decision rights explicitly.
For every category of decision, one person or role must have what Bain calls the "D" -- the authority to commit the organization. Not a committee. Not a consensus. One decision-maker with clear accountability. For strategic decisions (ICP definition, pricing architecture, market positioning), the CRO holds the D. For operational decisions (account prioritization within segments, campaign messaging), frontline leaders hold the D. For tactical decisions (individual outreach timing, follow-up cadence, research depth), reps and AI agents hold the D.
3. Build orientation, not approval chains.
Your highest-leverage investment is not reviewing more decisions. It is building the shared mental models that make most decisions obvious. This means regular strategy sessions where you share your reasoning, not just your conclusions. It means post-mortems that build shared understanding of what works and why. It means giving every rep and agent access to the full context -- win/loss analysis, competitive intelligence, customer data -- that feeds Boyd's Orient phase. When orientation is deep enough, decisions become implicit. The system flows without the bottleneck.
4. Develop barrels at every level.
Identify the people in your organization who can carry a vision from idea to completion without escalating every sub-decision. Give them scope. Give them authority. Coach them aggressively -- McKinsey's data shows that coaching combined with empowerment produces 3.2x better outcomes than empowerment alone. These are your decision-makers at operational altitude, and your organization's throughput is directly proportional to how many of them you have.
5. Set guardrails, not instructions.
Define the boundaries within which autonomous action is expected. Don't discount below X. Don't promise capabilities we can't deliver. Always lead with these three value propositions. Engage these buyer roles in deals above this threshold. Within those guardrails, reps and agents act without asking. This is Auftragstaktik applied to revenue: the CRO communicates intent, the frontline executes.
6. Create shared consciousness.
Radical transparency is not optional. Every rep and every agent should see the same pipeline data, the same strategic priorities, the same competitive positioning. McChrystal's principle applies: empowered execution without shared consciousness produces action without alignment. The CRO's operating rhythm -- the weekly pipeline review, the quarterly strategy session, the daily standups -- should be designed not to make decisions but to distribute context.
7. Manage decision debt with cadence.
Create a forcing function for decisions. Unresolved choices accumulate monitoring effort, communication overhead, and context-reloading costs. They sap momentum. They disengage your best people, who tire of unclear direction and perpetual waiting. Establish a weekly cadence for clearing the decision backlog. If a decision has been discussed in three meetings without resolution, it gets decided in the fourth -- with 70% information, imperfect consensus, and a commitment to correct course if the data changes. Disagree and commit.
The Constraint Has Shifted
The revenue organizations that will win in the AI era are not the ones that deploy the most agents, generate the most outreach, or execute the most sequences. Execution is now table stakes. Everyone will have the same AI capabilities within eighteen months of anyone else.
The winners will be the organizations that can decide -- quickly, at the right altitude, with clear accountability -- what all that execution capacity should be aimed at. They will have CROs who understand that their job is no longer to make decisions but to build a decision-making system. They will have oriented teams that act from shared understanding rather than waiting for explicit approval. They will have barrels at every level who can direct the infinite ammunition that AI provides.
As the IMD Business School concluded in their 2026 research: "The winners will not be those deploying the most models, but those reinventing how decisions, teams, and accountability are organized around AI."
The execution constraint is dead. The decision constraint is the new bottleneck. And per Goldratt, every hour lost at the bottleneck is an hour lost for the entire system.
The question for every CRO is simple: are you building a faster execution engine, or are you building a faster decision-making system? Because in 2026, only one of those is the constraint that matters.