Governance in Reverse: Diagnosing the Quiet Coup of Undemocratic IT Decision-Making
Photo: Iodonline, CC BY-SA 4.0, via Wikimedia Commons
No board ever voted to let the technology organization set company strategy. Yet in a remarkable number of American enterprises, that is effectively what has happened. Not through malice, and usually not through ambition — but through a slow, quiet transfer of decision-making authority that occurs wherever business leaders defer to technical jargon they don't fully understand.
As enterprises race to adopt AI and automation, the stakes of this quiet power shift are rising fast. The executives who choose which models to deploy, which data to train on, and which decisions to automate are, in a very real sense, choosing what the company becomes. That choice deserves to be made deliberately, with governance — not by default. Here are five signs the balance has slipped, and what to do about each.
1. Budget Lines That Balloon Without a Business Case
The sign: Technology spending grows year over year, presented in aggregate terms ('infrastructure modernization,' 'AI readiness') that no non-technical leader can interrogate. Requests are framed as non-negotiable.
The fix: Require every material initiative to be tied to a named business outcome with a measurable target, reviewed by the business units it affects. Capital allocation committees should include at least one member empowered to ask the naive question — because the naive question is usually the one nobody in the room has asked in three years.
2. Tool Choices Presented as Faits Accomplis
The sign: The organization learns it has adopted a new platform, AI vendor, or data-sharing arrangement after the contract is signed. Business stakeholders are informed rather than consulted, and the announcement arrives framed as settled.
The fix: Institute a stage-gate review for any system that touches customer data, financial reporting, or core operations. Business sign-off must occur before procurement, not after. This is not bureaucracy; it is the same due diligence applied to any other material commitment.
3. Automation Decides Things No Human Remembers Deciding
The sign: AI-driven systems are making consequential decisions — credit thresholds, hiring filters, inventory commitments, customer routing — with no documented record of who approved the decision logic or when it was last reviewed.
The fix: Maintain a decision inventory: every automated decision with material business impact, its owner, its approval date, and its review cadence. If a system cannot be explained to a board member in two sentences, it is not ready to operate unsupervised.
4. Expertise Becomes a Veto
The sign: Options die in technical review without ever reaching the people who sponsored them. 'The architecture team rejected it' ends conversations that strategy should have been allowed to have.
The fix: Distinguish between technical constraints (real, often binding) and technical preferences (negotiable). Require that any rejection of a business-sponsored initiative come with documented reasoning and, where feasible, an alternative path. Expertise should inform decisions, not quietly make them.
5. Risk Reporting Flows Only Downward
The sign: Leadership hears about cyber incidents, compliance gaps, or failed implementations only when they surface publicly. The information channel between the technology organization and the board runs through a single filter.
The fix: Establish an independent reporting line — through audit, risk, or a technology committee with its own advisory capability — so leadership receives unfiltered visibility into technology risk. Transparency here protects everyone, including the technology leaders themselves.
Balance, Not Paralysis
None of this should be read as an argument for weakening technology leadership. The strongest enterprises pair capable technical teams with equally capable business governance — each sharpening the other. The goal is not to slow delivery but to ensure that the direction of delivery reflects the whole organization's intent.
The companies that get this right treat technology decisions as what they have always been: business decisions with a technical vocabulary. The ones that get it wrong will discover, usually at the worst possible moment, that no one ever voted for the people who were making the calls.
