Twelve executives, one disagreement, ninety minutes of meeting. Should the company create a chief innovation officer position? According to the account given by Lindy Greer, Jennifer Jordan and Maxim Sytch in the July-August 2026 issue of Harvard Business Review, the question took up an entire room of executives at a large technology company, without any decision coming out of it. Everyone had an opinion. Nobody, in reality, had the final say.

There is nothing exotic about the episode. Every growing organization, product included, eventually runs into the same unspoken question: who exactly decides when several legitimate claims overlap? The researchers from the University of Michigan and IMD give it a precise name: decision rights, meaning the clear allocation of who decides, who is consulted and who executes when facing a given choice. A piece of organizational plumbing so discreet that you only notice it when it leaks.

A forty-year-old tool, still misused

Since the 1980s, most companies have relied on the same tool to clarify this plumbing: the RACI matrix (Responsible, Accountable, Consulted, Informed), a table that assigns each person, for each decision, one of these four roles: who does the work, who is ultimately answerable for it, who must be consulted before the choice is made final, and who simply needs to be informed once it is. On paper, the tool is crystal clear. In practice, according to the three Harvard Business Review authors, it fails so often that it ends up worsening the confusion it was supposed to dispel.

The firm McKinsey, in a separate analysis drawn from its own consulting engagements, reaches a similar diagnosis: by multiplying the people who have oversight or a veto over the same decision, the RACI matrix dilutes the very thing it claims to clarify. Too many voices, not enough deciders: the meeting drags on, the decision slips back.

The four mistakes the study reveals

The article identifies four recurring traps. First, haste: companies assign roles before they have even clarified the objective being pursued, which multiplies ownership disputes as soon as several people believe, in good faith, that they carry the same responsibility. Second, rigidity: the role chart is drawn once, usually by a single executive, then forgotten in a binder, while priorities and teams keep moving. Third, semantic confusion: a role such as “consulted” is understood differently by everyone, sometimes as a simple opinion to gather, sometimes as a de facto veto, which distorts the decision without anyone really being aware of it. Finally, hierarchical precedence: when the pressure rises, the highest-paid person imposes their choice, regardless of what the matrix had planned, a reflex so widespread that it has a name in English-language management literature, the HiPPO effect (the “highest paid person’s opinion”).

Product, a full-scale testing ground

No discipline experiences this blind spot as regularly as product management. A roadmap, whose construction we have already detailed elsewhere, by nature brings together competing opinions: the salesperson who promises a feature to a strategic customer, the engineer who warns about technical debt, the designer who defends the experience, the executive who demands a quarterly result. Who decides in this chorus? The best-known prioritization grids, such as RICE (a score that combines reach, impact, confidence in the estimate and the effort required) or MoSCoW (must have, should have, could have, won’t have, that is, what is essential, desirable, optional or out of scope), often fail for the same reason as the RACI matrix: a score table never says who ultimately decides. We had already detailed the limits of these tools elsewhere for prioritization; the decision rights problem runs even deeper, because it touches authority itself, not just the calculation method.

This is precisely what Marty Cagan, cofounder of the Silicon Valley Product Group, has theorized, without ever using the term “decision rights”. His notion of the autonomous product team (empowered team, a team entrusted with a problem to solve rather than a list of features to deliver) rests on a principle that is simple to state and hard to uphold: an engineer who judges a solution to be bad has the right to refuse to build it, provided they propose another one. Autonomy, for Cagan, is not the absence of hierarchy: it is a hierarchy of decisions, where each question belongs to an identified person, not to a soft consensus or an order from above.

RACI, DACI, DARE: changing tools is not enough

Faced with these repeated failures, several organizations have looked for a better grammar. Atlassian, the company behind Jira and Confluence, has for several years promoted DACI (Driver, Approver, Contributors, Informed), a variant centered no longer on the tasks to be carried out but on the decision itself: a Driver prepares the case and the options, a single Approver decides, contributors bring their expertise without a veto, and the informed people learn the outcome once it is final.

How to run a DACI as a team, presented by the official Atlassian Training program (Atlassian YouTube channel).

McKinsey, for its part, proposes DARE (deciders, advisors, recommenders, execution stakeholders), whose principle fits in one sentence: give more people a voice, but fewer people a vote. Three acronyms, three organizations, one shared finding: a decision must have a single named owner, everything else being only degrees of listening.

What is really at stake

The stakes go beyond the semantics of charts. Product organizations have spent fifteen years flattening their hierarchies in the name of agility, betting on cross-functional teams and short rituals. But flattening the hierarchy did not abolish the question of power: it made it informal, and therefore harder to correct when it malfunctions. A team that speeds up without having settled who decides does not become more agile, it simply imports its disagreement earlier in the cycle. This is the paradox that the research by Greer, Jordan and Sytch documents in its own way: clarity of decision-making power is not a bureaucratic add-on to agility, it is one of its conditions.

What the study does not say yet

One caveat is in order. The three professors’ research covers general organizations, not product teams specifically, and the article, written for an executive readership, does not detail a quantified, publicly accessible methodology at this stage. The diagnosis of the four mistakes has the solidity of a recognized academic synthesis, not yet that of a longitudinal study carried out specifically on product teams. Another limit, an older one: changing acronyms changes nothing if the organization keeps treating the role chart as a formality to fill in once and then forget. The authors themselves insist: co-building the roles with the people concerned, defining the behavior expected of each, and reviewing the allocation at regular intervals matter more than the name of the framework chosen.

What a product manager can do starting Monday

Before the next roadmap trade-off meeting, one exercise comes down to four simple moves. Write the decision in one precise sentence, not the general objective around it: “choose which feature goes into the next development cycle” rather than “improve customer satisfaction”. Name a single approver, even if it is the product manager themselves, and own it publicly before the meeting, not after the fact. Specify, for each person consulted, what their opinion really commits: simple input, or a right to block; the nuance changes the whole debate. Finally, set a review date, in a month or a quarter, to check that the allocation still holds, rather than treating it as set in stone in a forgotten document.

The twelve executives in the study ended up parting ways without deciding, each convinced they were right. The risk for a product team is not so different: by trying to involve everyone in every decision, you get meetings that look like deliberations but never give birth to a choice. The speed that agile methods promise is only worth anything if, somewhere in the room, one person knows they have the final say, and the others know it too.

Sources

Read the full study on hbr.org.