Running a board self-assessment on a small board

A six-person board is not a fifteen-person board with nine people missing. It runs differently, it fails differently, and almost everything written about board self-assessment quietly assumes the larger version. Follow that advice literally on a small board and you get a survey nobody trusts, a report you never see, and a set of questions that describe somebody else's room.

Three things change when the board is under about seven people. Anonymity stops being a technical guarantee and starts being a social one. The threshold that releases a report becomes the hardest part of the whole exercise. And a handful of questions stop meaning anything at all.

Anonymity is thinner in the room than in the data

Two different things get called anonymity, and on a small board they come apart.

The first is what the system does. Nothing connects an answer to the person who gave it. Invitations and responses share only the cycle they belong to, so there is no lookup to run, no admin view that reveals it, and no version of the report where the chair can see who said what. That holds whether six people answer or sixty.

The second is what the people in the room can infer. If five colleagues who have sat together for four years read a written comment, one of them may recognize a turn of phrase. Nothing in the data connects it. Somebody may still guess. Pretending otherwise is the fastest way to lose the board's trust, because they already know it.

So say it plainly before question one, in a meeting, out loud. The scored questions are reported as an aggregate and never individually. Written comments are shown word for word or withheld entirely, never summarized, because a summary of four comments identifies their authors about as well as the comments do. And the written prompts are optional, so anyone who would rather stay on the scored items can.

That last point does more work than it looks like it does. A board member who is told the free text is optional usually writes something. One who suspects the chair will be reading between the lines writes the safe answer, and the safe answer is worth nothing to anybody.

The threshold is the hard part, and it does not bend

A board report is withheld unless it has at least 4 responses and at least 60% of the roster. Whichever of those two is larger is the number you need. That is the rule everywhere in the product, with no admin override and no exception for a board that really wants to see it.

Run the arithmetic for a small board and it gets concrete fast:

• A 4-member board needs all 4.
• A 5-member board needs 4.
• A 6-member board needs 4.
• A 7-member board needs 5.
• A 9-member board needs 6.

A board of three cannot clear it at all, and it is better to know that before you send anything than after.

Below the threshold the section is withheld with a plain explanation of why. It is not shown partially, not shown with a warning label, and not shown to the chair only. There is no half-report. The chief executive's separate column follows the same gate, because a column of one person's answers sitting beside a suppressed board report would identify them.

Breakouts are stricter still. Slicing results by tenure or by committee needs 5 responses inside the slice, which on a six-person board is never going to happen. Expect those to stay withheld even in a round that otherwise releases. That is the design working, not a fault.

The practical consequence is that chasing the last response is the entire job. On a fifteen-person board, a 60% response rate is a decent outcome. On a six-person board, one person going quiet is the difference between a report and nothing. Get verbal commitment in the meeting before the invitations go out, name a closing date, and tell people the number. "If two of us skip this, there is no report" is a sentence that gets responses. Vague encouragement does not.

Some questions were written for a bigger room

Standard board instruments contain items that only make sense above a certain size. The clearest example in this one is a culture question that reads, exactly: "All board members participate; a small group does not dominate."

On a fifteen-person board that is a real measurement. Three people run every discussion while the rest watch, and the item catches it. On a five-person board where everyone speaks at every meeting because there is nobody to hide behind, the question measures nothing. People answer it, the answer means whatever they decided it meant, and the score is noise.

Small-board mode swaps that item out rather than adding another one on top. Turn it on and two items are replaced, so the assessment does not get longer. The replacement asks what actually goes wrong on a board this size: "The work of this board is shared rather than carried by one or two people." The second replacement trades a question about personal engagement for "I know what I am specifically responsible for," which on a small board is the thing that is usually unclear.

The general lesson survives even if you use a different instrument or write your own. Read every question and ask whether the failure it describes is available to a board your size. Domination by a subgroup is not. Two people carrying everything is. A committee structure nobody has reviewed is not a small-board problem; having no committees and no plan for the work is. You can see the full board list at the published board questions and make that judgment before anyone answers.

What to do before you send anything

• Decide in a meeting, not over email, and get each person to say they will do it.
• Give a two-week window with a named close date, and say what happens if the threshold is not met.
• Turn on small-board mode if you are under six, and tell people you did.
• Do not promise breakouts. On this size board they will be withheld.
• Ask the chief executive the same core questions and keep their answers in their own column rather than averaging them in. The gap between the two is usually the most useful thing you will get.
• Book the meeting where you will discuss the results before the results exist. Boards that skip this step read the report and do nothing.

BoardSource publishes sector data on board size and composition if you want a reference point for how your board compares structurally, though the numbers there describe boards considerably larger than most of the ones reading this.

If you want the answers rather than the questions, this assessment collects them anonymously, applies the threshold above without exception, and shows the board and the chief executive side by side. It is free, and there is no paid version of it. It takes about eight minutes per person.

Small boards often assume this exercise is for organizations bigger than they are. It is closer to the opposite. On a board of six, one disengaged member is 17% of your governance, and nobody has a way to raise that in a meeting. A written round is the least awkward version of that conversation anyone has come up with. Whether it is worth running every year is a separate question with a fairly clear answer.