You can assemble a board that ticks every box, the right number of independent directors, a former CFO, a digital expert, a balanced mix of backgrounds, and still watch it rubber-stamp a doomed strategy. Composition tells you who is in the room. Effectiveness is about what happens once they are. Evaluation is how you find out the difference, honestly, before the market does it for you.

The quick version

  • Composition is the board's make-up: how many directors, the split between executives and independent non-executives, and the spread of skills, experience and background. It is necessary but not sufficient.
  • Effectiveness is whether the board actually does its job, challenging management, setting direction, and overseeing risk. Research keeps finding that how a board works together predicts performance better than its structure on paper.
  • Evaluation (the UK Code now calls it a board performance review) is the annual, evidence-based look at composition, dynamics and individual contribution, with an external facilitator at least every three years for larger listed companies.
  • The trap is treating a tidy composition table or a completed evaluation questionnaire as the goal. The point is better decisions, and that lives in behaviour, not box-ticking.

The idea in depth: structure is the floor, not the ceiling

Most governance rules concern themselves with composition because composition is observable and enforceable. The UK Corporate Governance Code, updated in 2024 by the Financial Reporting Council, devotes a whole section to "Composition, Succession and Evaluation," asking that at least half the board (excluding the chair) be independent non-executive directors, that appointments be merit-based and promote diversity and inclusion, and that no one accumulate so many other roles that they cannot give the job real time (the 2024 Code and its accompanying Guidance on Board Effectiveness set this out). These are sensible defaults. A board stacked with executives and the founder's friends has no one positioned to say no.

But independence on an org chart is not the same as independence in a meeting. The most useful academic correction to the box-ticking instinct came from Daniel Forbes and Frances Milliken, whose paper "Cognition and Corporate Governance" (Academy of Management Review, vol. 24, 1999, pp. 489–505) reframed boards as strategic decision-making groups rather than legal structures. Their model argues that board performance turns on three behavioural ingredients: effort norms (do directors actually do the homework?), the use of knowledge and skills (does the board draw out what its members know?), and cognitive conflict (do they disagree about substance, not personalities?). That reframing changes the appointment question. Stop recruiting to fill a CV gap and start asking the harder one: will this person make the group think better, or just look better balanced?

A board is a strategic decision-making group, not a legal structure, its CVs matter less than how it argues.

Recent practitioner research points the same way. In a 2024 study of 120 directors published on the Harvard Law School Forum on Corporate Governance, Yuki Sakasai, Gaizka Ormazabal and Jordi Canals of IESE Business School concluded that "board structural characteristics as defined by most regulators and asset managers are insufficient", that professional competencies and soft skills are what separate effective boards from compliant ones. Directors in their survey ranked strategy-execution capability, leadership and the ability to collaborate as a team above formal credentials. For a chair, that suggests mapping the board not by job titles but by behaviours: who challenges well, who asks the question everyone is avoiding, who goes quiet when it matters.

flowchart TD
  A(["Composition
who is in the room"]) --> B(["Board dynamics
effort, knowledge use,
productive conflict"])
  B --> C(["Effectiveness
challenge, direction,
oversight"])
  C --> D(["Better decisions
+ firm performance"])
  E(["Evaluation
annual performance review"]) -.->|"reads"| B
  E -.->|"reshapes"| A
Composition feeds dynamics; dynamics drive effectiveness; evaluation is the feedback loop that reads the dynamics and reshapes the composition. Leaders Loop

Productive conflict, and where the evidence gets honest

The seductive part of the Forbes–Milliken model is cognitive conflict: the idea that boards decide better when directors disagree about the substance of a problem. It is intuitive, a board that nods along is a board not doing its job. The practical implication is to engineer disagreement on purpose: rotate who speaks first so the senior voice does not anchor the room, ask the newest director for their read before the chair signals a view, and treat a unanimous vote on a hard question as a warning sign rather than a triumph.

An honest limitation. The evidence for cognitive conflict is weaker than the theory's popularity suggests. A meta-analytic review of the Forbes–Milliken model found strong support for effort norms and the use of knowledge and skills, but inconclusive results for cognitive conflict, partly because the same disagreement that sharpens decisions can curdle into relationship conflict that fractures a board. Substance-disagreement helps; person-disagreement does not, and the two are hard to keep apart. So treat productive conflict as something to cultivate carefully, not a dial to crank to maximum. The chair's real skill is letting a board argue hard about the issue and still leave the room as a team.

Evaluation: the annual review that has to bite

If effectiveness lives in behaviour, you cannot manage it without looking at it, which is what a board performance review is for. The UK Code asks boards to review their own performance, that of their committees and of individual directors annually, and for FTSE 350 companies to have that review externally facilitated at least every three years, with the external evaluator named in the annual report (per the FRC's Code and guidance). The logic is the same one that makes 360-degree feedback work for individuals: a group cannot see its own blind spots from the inside, and the chair who runs the meetings is the last person who can objectively assess them.

The honest snag is that most evaluations change nothing. Industry surveys repeatedly find a gap between directors who believe evaluations are valuable and boards that actually act on them, a completed questionnaire that produces a polished report and no decisions is theatre. Design the review for action, not assurance: ask not "are we a good board?" but "what is the one thing about how we work that we will change before the next cycle?", then put that change on a future agenda and check it happened. An evaluation that does not move at least one behaviour, agenda item, or board member has failed, however reassuring its scores.

flowchart LR
  A(["Gather evidence
survey + interviews,
external every 3 yrs"]) --> B(["Read the dynamics
not just the scores"])
  B --> C(["Name 1-2 real changes
behaviour, agenda,
composition"])
  C --> D(["Act
put it on a future agenda"])
  D --> E(["Re-check next cycle
did it actually change?"])
  E -.-> A
A board evaluation that closes the loop, evidence to a named change to a follow-up, versus one that stops at a reassuring report. Leaders Loop

A worked example

Picture the board of a mid-sized listed retailer, call it Mersey Goods. (Illustrative throughout; this is a teaching example, not a real company.) On paper the composition is exemplary: nine directors, five independent, a former retail CEO, an ex-CFO, a digital-commerce specialist, gender-balanced, the requisite committees in place. The annual self-assessment questionnaire scores the board 4.4 out of 5. Everyone is relieved.

Then a triennial external review lands. The facilitator interviews each director privately and sits in on two meetings. The structural scorecard is indeed excellent, and the dynamics are quietly broken. The former CEO dominates every strategy discussion; the digital specialist, hired precisely to challenge the store-led plan, has stopped speaking because her first two interventions were brushed aside. Decisions are unanimous because dissent has been trained out of the room. This is a board with model composition and no cognitive conflict, exactly the failure the structural table cannot see.

The review names two changes, not twenty. First, the chair restructures strategy sessions so the executive presenting goes last and the independent directors are asked for their read first, an effort-norms-and-conflict fix straight out of the Forbes–Milliken playbook. Second, the digital specialist is given the lead on a board sub-discussion of the online channel, putting her knowledge to use rather than letting it sit unused. Twelve months on, the questionnaire score has barely moved, but the board killed a costly store-expansion plan that the old, agreeable version would have waved through. The number that mattered was never the 4.4. It was whether the right argument finally happened.

Frequently asked questions

What is the difference between board composition and board effectiveness?

Composition is the static make-up, how many directors, the executive-versus-independent split, the mix of skills and backgrounds. Effectiveness is the dynamic question of whether that group actually challenges management, sets direction and oversees risk well. Good composition makes effectiveness possible but does not guarantee it; a well-composed board can still defer, drift, or fall into groupthink.

How many independent directors should a board have?

Codes vary by jurisdiction, so check the one that applies to you. The UK Corporate Governance Code asks that, apart from smaller companies, at least half the board excluding the chair be independent non-executive directors. The principle behind the number matters more than the number: you need enough genuinely independent voices to make saying "no" to management normal rather than career-limiting.

What does a board evaluation actually involve?

Typically a mix of self-assessment questionnaires, confidential interviews and sometimes observation of meetings, covering the board as a whole, its committees and individual directors. For FTSE 350 companies the UK Code expects this to be externally facilitated at least every three years, with the evaluator named in the annual report. The format matters less than whether it produces specific, followed-up changes rather than a reassuring summary.

Does board diversity improve performance?

The honest answer is "it depends on what kind, and on how the board uses it." Diversity of background and cognition can widen the range of perspectives a board brings to a decision, which is the mechanism that helps. But diversity only pays off if the dissenting voices are actually drawn out and listened to, a diverse board run as a rubber stamp captures none of the benefit. Treat diversity as raw material for better debate, not a result in itself.

Can a board be too conflict-driven?

Yes. The research distinguishes conflict about the substance of a decision (useful) from conflict about people and relationships (corrosive), and the two are easy to confuse in the heat of a meeting. A board that argues constantly about personalities decides worse, not better. The chair's job is to keep disagreement aimed at the problem and to make sure the board still functions as a team once the vote is taken.

Related in the Toolkit

Composition only makes sense once you know what a board is for, the division of labour across roles and committees shapes who needs to be in the room, and the distinction between executive and non-executive directors is the structural backbone of board independence.

Where to go next