How to get honest answers from your board

Ask a board to rate itself for the first time and the ratings come back high. Almost uniformly high, with very little spread, and a free-text box that says "great group of people" nine times. The chair reads it, concludes the board is in good shape, and files it.

Sometimes that's true. More often it means people gave the answer that costs them nothing. Getting past that is not a matter of better questions. It's a matter of whether the people answering believe, with reason, that nobody can work out who said what.

Why people give the safe answer

Board service is social before it's anything else. You joined because someone you respect asked you. You sit next to the treasurer at the annual event. You will see all these people again in six weeks.

Now someone hands you a form asking whether the board understands the organization's finances, and you don't think it does. Rating that honestly means saying something critical about a person in the room, in writing, in a document the chair will read. The safe answer costs you nothing and helps nobody. Most people, most of the time, pick the safe answer, and they're behaving rationally when they do.

This gets worse the smaller and more collegial the board is. A nine-person board that likes each other is harder to get candor from than a twenty-person board of near-strangers, which is the opposite of what most chairs expect.

What anonymity has to mean in the plumbing

The word "anonymous" in an email does no work at all. Board members have filled in enough surveys to know that most tools quietly record who submitted what, and they'll assume yours does too unless you can tell them precisely why it doesn't.

So the mechanism has to be specific enough to describe in a sentence. In this assessment, the invitation and the answers live in separate places that share only which assessment cycle they belong to. There is no key between them. Not a hidden one, not a restricted one. It doesn't exist. Answers carry no name, no email, and no IP address.

They also carry no submission time, which sounds like a technicality and isn't. If your answers recorded when they were submitted, anyone with the invitation list could line up submission times against who had completed and work backwards to a name. So no time is stored at all.

That design has real costs, and it's worth being honest about them rather than presenting it as free. You can't resume a half-finished response on another device. You can't send a reminder to only the people who haven't answered. And nobody, ever, can follow up with whoever rated something low, because there is no way to know who they were. If your plan for the results depended on that, the plan needs to change.

What your board does see is totals: 9 of 12 responded, never which nine.

The last piece is what happens when the numbers are small. Results are withheld entirely below a threshold rather than shown with a caveat. A board report needs at least four responses and at least 60% of the roster. Written comments are shown exactly as written or not at all, because a summary of four comments identifies their authors about as reliably as the comments do. Tell people this before question one. It's the single most convincing thing you can say, because it's a promise the tool keeps whether or not you want it to.

How to frame the ask

Do this out loud, in a meeting, before any email goes out. Two sentences, from the chair, in this order.

Say why now, and make it specific and non-accusatory. "We've never asked ourselves how this is going, and with the strategic plan coming up it seems worth knowing" works. "Some concerns have been raised" guarantees you learn nothing.

Say what happens to the answers, using the same words the survey uses, not a friendlier paraphrase. Then say what you'll do with the results and when: a named meeting, a specific amount of time on the agenda.

Then get out of the way. Set a deadline about ten days out, say roughly how long it takes, and don't chase individuals. If a chair is visibly tracking who has and hasn't answered, everyone reasonably concludes the chair can also see how they answered.

One more thing that helps more than it should: answer it yourself, first, and say so. A chair who treats it as something the board does to itself rather than something the chair administers changes the temperature of the whole exercise.

Four things that kill honesty

The chair announces a target. "It'd be great if we came out above a four on everything." Said warmly, meant well, and it converts the survey into a loyalty test. Nobody undercuts the chair's stated hope in writing. If you want a number to move, say so after the results, never before.

The chair collects the responses. Most general-purpose form tools record the respondent's email address by default, and some display it. Even when it's switched off, board members can't verify that, so they assume the worst. If you're using a general tool rather than a purpose-built one, show someone the settings screen. Better, use something where the linkage doesn't exist to be switched on.

Asking for names optionally. This is the one that seems generous and is the most damaging. If the name field is optional, the people who feel safe fill it in, and the people who don't become conspicuous by omission. Worse, its presence tells everyone that identification is possible in principle, which contaminates the answers of people who leave it blank. Either responses are unattributed or they aren't. There's no middle setting.

Ignoring what a small group can and can't hide. On a six-person board, a comment written in a distinctive voice is a signature, and a breakout by tenure or committee is worse. Breakouts need at least five responses within the group before they'll show, which on most small boards means they simply won't, and that's the right outcome. If your board is under seven people, skip demographic slices entirely, expect fewer written comments, and tell the board in advance that some sections may be withheld. A board that's warned about suppression trusts it. A board that hits it unexpectedly assumes something broke.

What honest answers look like

They're messier. Spread widens, "don't know" shows up in places you assumed were settled, and free text gets specific enough to be uncomfortable. Ratings often fall in the second year, and the common misreading is that the board got worse. Usually the board got honest, which is why year-over-year comparison only works when the questions stay identical and the confidentiality terms don't change between rounds. That stability is also the argument for running one annually rather than every third year.

If it's useful to see how the wording handles this, the board questions are published in full, and the National Council of Nonprofits has a good overview of self-assessment practice from outside any one vendor. The assessment itself is free, takes about eight minutes, and stores no link between a person and their answers.

The uncomfortable version of all this is that candor is not something you ask for. It's something you make structurally safe and then stop interfering with. Say the terms plainly, mean them, and then leave your board alone until the deadline.