The Accountability Architecture Problem

The first executive search I lost a client over was not a bad search. The candidate was right for the role. The process was thorough. The presentation was strong. What failed was something I had not designed for: the handoff between the search process and the client's internal decision-making, where nobody on either side had clear accountability for what happened next.

The candidate was excellent. The client was interested. And then nothing moved for six weeks because the people who needed to make the internal decision each believed someone else was responsible for driving it. The candidate accepted a different offer. The search failed.

The failure was not a people problem. The people on both sides were capable and well-intentioned. It was an accountability architecture problem: a gap in the system where responsibility was assumed rather than assigned, and where the cost of the gap — a failed search — was distributed across everyone involved rather than concentrated on the person who should have driven the outcome.

That pattern — responsibility assumed, accountability diffuse, cost distributed — is the most common failure mode in organisations that I work with.

The distinction that matters

Responsibility is assigned. Accountability is real.

Every organisation assigns responsibility. Job descriptions, project ownership, RACI matrices — these are responsibility assignment tools. They describe who is nominally in charge of what. They do not, by themselves, create accountability.

Accountability is real when the person who was assigned responsibility actually bears the consequences if the outcome fails. Reputationally. Economically. Structurally. When the outcome of what they owned is visible and attributed — when there is no way to diffuse the result across the team, the market conditions, or the timing.

Research by London Business School professor Dan Cable found that the most consistent predictor of high performance in organisations was not the quality of the stated goals or the sophistication of the measurement system. It was whether the people responsible for outcomes genuinely believed they would bear the consequences of those outcomes. The belief changed the behaviour. The accountability changed the quality of decision-making before any outcome was produced.

This is the mechanism: accountability shapes behaviour before the fact. The person who knows they will genuinely own the outcome invests differently in understanding the problem, surfaces the issues they see, and escalates more honestly than the person who knows the accountability is diffuse or symbolic.

Most organisations assign responsibility liberally and have accountability that is diffuse, negotiable, or symbolic. Accountability is real only when the person who owns a domain genuinely bears the consequences if the outcome fails. That belief — that the consequences are real — is what changes behaviour before the fact.

Why organisations avoid real accountability

Genuine accountability is uncomfortable. This is not a failure of character — it is a structural feature of how organisations work.

Holding someone accountable for an outcome requires having a difficult conversation when the outcome fails. It requires distinguishing between the person and the result — acknowledging that a capable person produced a poor outcome and that this matters — without destroying the relationship or the person's willingness to take on the next difficult task. Most managers are not trained for this and find it easier to diffuse the accountability than to have the conversation.

It also requires accepting imperfect information. Outcomes are not always cleanly attributable. Markets move. Clients change. External factors intervene. The person who owns an outcome cannot always control all the variables that determine it. Building a genuine accountability system requires building the capacity to distinguish between good decisions that produced bad outcomes — which happens and which should not punish the decision-maker — and bad decisions that produced bad outcomes, which should.

Most organisations collapse this distinction. They either avoid accountability entirely (making it collective, symbolic, or invisible) or they apply it crudely (blaming individuals for outcomes regardless of the quality of their decision-making). Both extremes are easier than the genuine version and both destroy organisational performance in different ways.

The Majhi Group version

In executive search, accountability is unusually visible. The outcome of a search — whether the hire works, whether the relationship holds — is directly attributable to the people who ran it. There is no diffusion. The client knows who did the search. The result either vindicates or implicates the firm.

This visible accountability is one of the reasons the retained model produces different behaviour than contingency. The contingency recruiter's accountability ends at placement. Whether the hire works at twelve months is not their problem — they are already running the next search. The retained firm's accountability explicitly extends beyond placement: the 90-day replacement guarantee is an accountability architecture. It creates an economic and reputational consequence for a hire that fails, which concentrates attention on quality rather than speed.

The accountability structure is the incentive structure. And the incentive structure is the culture.

Building this into Majhi Group required being explicit about what we are accountable for: not just the placement, but the outcome. Not just the shortlist, but the quality of the hire at six months. This is uncomfortable when a search produces a hire that struggles. It is also what makes the search process better, because the people running it know they will be judged by the outcome they can't walk away from.

The Majhi OS design

The same problem appears in the technical architecture of Majhi OS.

The observability layer monitors mandate health and fires alerts when something is degrading. But an alert without accountability is noise. If nobody owns the mandate, nobody owns the alert. If the accountability for recovery is diffuse — spread across the recruiter, the search lead, the client contact — then the alert sits unacknowledged while everyone waits for someone else to act.

The architecture of the accountability system determines whether the system works. Each alert has an owner. The owner is the person whose outcome is affected if the alert is ignored. The accountability is built into the system design, not left to informal negotiation about who should probably look at this.

This is a software design decision. It is also an institutional design decision. The two are the same problem: when you surface a problem, who owns the response?

Building it without fear

The counterargument to genuine accountability is that it creates a culture of fear — that people become risk-averse, hide problems, and stop experimenting when they know they will personally bear the consequences of failure.

This is a real risk when accountability is built badly. Badly built accountability punishes outcomes regardless of decision quality — blaming people for results they could not control, treating bad luck and bad judgment as equivalent. That version does create fear.

Well-built accountability distinguishes between the quality of the decision and the quality of the outcome. It holds people accountable for the things they could control: their analysis, their communication, whether they surfaced the problem when they saw it, whether they made a reasonable decision with the information they had. It does not hold them accountable for outcomes determined by factors outside their control.

Harvard Business School research by Amy Edmondson on psychological safety found that the highest-performing teams combined high psychological safety with high accountability standards — not as a contradiction but as a design. People felt safe raising problems, disagreeing, and admitting mistakes because they trusted the accountability system to distinguish between honest errors and negligent ones. The safety and the accountability were mutually reinforcing.

This combination — genuine accountability for what people can control, with genuine safety to be honest about what they can't — is harder to build than either extreme. It requires leadership that can have difficult conversations about outcomes while remaining curious rather than punitive about the decisions that produced them. It requires a system that makes the distinction visible rather than leaving it to individual judgment in the moment.

In Kalahandi, the institutions I saw fail most often failed because accountability was either absent (nobody bore the consequences of poor administration) or arbitrary (consequences were distributed by politics rather than by outcomes). The institutions that worked had accountability that was real and that was legible — people understood what they would be held to and what they wouldn't. That legibility was itself a kind of justice that made the accountability bearable.

Building with that in mind — in organisations where the accountability is real enough to change behaviour and legible enough to feel fair — is the institutional design problem that matters more than almost any other.

Sources

London Business School — The Accountability Problem

Harvard Business Review — What Is Psychological Safety? (Amy Edmondson, 2023)

McKinsey — Accountability by design in the agentic organisation