Onboarding ends on day one. Retention starts in week three.

A new volunteer arrives on a Tuesday evening. Somebody finds her a lanyard, hands her a folder with the safeguarding policy in it, walks her round the building, introduces her to eleven people whose names she will not remember, and warns her that the kettle is temperamental. She signs something. She goes home.

That is volunteer onboarding at most organizations I have seen. One evening, front-loaded, mostly paperwork, and then nothing structured until she either sticks or doesn’t.

The evidence says the first evening is the least important part of the process and the eight weeks after it are the most.

Different things keep people at different points

Chacón, Vecina and Dávila published a three-stage model of volunteer duration in Social Behavior and Personality in 2007, and it has held up better than most of what surrounds it. Their finding was that the thing predicting whether a volunteer is still there depends on how long they have already been there.

In the short term, meaning the first months, the predictor is satisfaction. Not commitment to your organization, which has had no time to form. Not any sense of themselves as a volunteer of yours, which takes about a year. Just satisfaction. Was that evening worth the evening?

This is a more demanding standard than it sounds, because satisfaction is produced by experience and nothing in a policy folder produces any. A person who spent her first three shifts standing slightly to one side of the actual work, waiting to be told what to do, has already formed the judgment that decides whether there is a fourth. She will not phrase it as dissatisfaction. She will phrase it as being busy.

The strongest lever is training, and it is not close

Piatak and Carman surveyed 323 volunteers between January and April 2020 for a paper in the Journal of Public and Nonprofit Affairs. They tested a long list of things about the volunteer experience against two outcomes: whether the volunteer would recommend the organization to someone else, and whether they intended to volunteer again.

Appropriate training was the strongest predictor of both. Volunteers who received it were far more likely to recommend the organization and far more likely to say they would be back. Inclusion and diversity climate also predicted both, and 13.4 percent of the open-ended suggestions volunteers wrote in asked the organization to become more inclusive.

Now the part that should reorganize your Monday. Logistics were not significant. Choice of project, ease of registration, the range of activities available: none of them predicted either outcome, despite being where a great deal of volunteer program improvement effort goes, and despite being what most sign-up software is sold on.

I take that to mean the front door is not where you are losing people. The room behind it is.

What “training” means here, and what it doesn’t

It does not mean a module. A sixty-minute slide deck delivered to a room of eleven new people is an induction with a different name.

The version that matches the research is closer to apprenticeship. NCVO’s induction guidance says to arrange for a new volunteer to shadow an experienced one, and this is the single line in that guidance most often treated as optional. It is the one that does the work. Somebody who has done the Saturday sorting shift ninety times knows the eleven small things nobody wrote down, including what to do when a donation arrives that you cannot accept, and she transmits them in an afternoon.

Two shifts of shadowing costs you two shifts of a volunteer’s time. Set against the fact that around a third of organizations keep their volunteers less than ten months, which is what the 2023 Volunteer Management Progress Report from Tobi Johnson & Associates found across 1,247 respondents in 36 countries, it is cheap.

The induction list, and the two items everyone skips

NCVO’s guidance for a volunteer induction is a good checklist and I would not rewrite it. Confirm they understand the role and what is expected. Introduce the staff. Tour the site. Explain the resources, the breaks and how to claim expenses. Arrange the shadowing. Share the safeguarding, equal opportunities and data protection policies along with the complaints procedure. Assign a current volunteer as a buddy. Give them a handbook. Ask whether they need any of it in another format, such as large print or audio.

Two of those get skipped almost universally, and they are the two that signal most.

Expenses. NCVO’s Time Well Spent 2023, based on YouGov fieldwork with 7,006 UK adults, found only 55 percent of organizations reimburse volunteer expenses at all, and 16 percent of volunteers were unsure what their organization’s policy even was. Among young people, 20 percent named expenses as a barrier to volunteering. A volunteer who is quietly paying four dollars in bus fare to give you three hours, and who has not been told whether that is claimable, is being charged for the privilege of helping you. Say the policy out loud on the first evening, including if the answer is that you cannot reimburse, because then at least she can decide with the facts.

The complaints procedure. Telling a brand new volunteer how to complain about you feels like inviting trouble. It is the opposite. It is the clearest available signal that this is a real organization with a spine and not a group of people who will close ranks. It costs one sentence.

Weeks three to twelve are the part nobody plans

Here is a ninety-day shape that fits what the research actually says. It is not elaborate and it does not need software.

Week one. The induction above, plus the first shift alongside their buddy. Do not leave them to observe. Give them something small that is genuinely theirs by the end of the evening.

Week three. A five-minute conversation, in person, from the person who runs the shift. One question: is there anything you have been unsure about that nobody has explained? New volunteers accumulate small confusions and will not raise them unprompted, because raising them feels like admitting they were not paying attention on the first night.

Week six. The competence checkpoint. By now they should be able to do the core of the role without asking. If they cannot, that is a training gap and it is yours, not theirs. This is also the point to hand over the first thing with a boundary around it, because what you ask someone to do is the strongest thing you control.

Week twelve. Ask what they would change. Not “how is it going,” which produces “fine.” Ask what one thing they would change about the way the shift runs, and then change something.

I started out running youth groups in Ottawa, long before I had any theory about any of this, and the thing I got wrong for years was treating the first night as the event. I would put real effort into a welcome and then hand the person a rota. The volunteers who stayed were the ones who happened to fall in with somebody who taught them the work. That was luck. It should have been the design.

The obvious objection

You do not have time for a three-touch schedule across ninety days for every new volunteer. I know. The same Volunteer Management Progress Report found only 47 percent of volunteer managers spend more than three quarters of their hours on volunteer management, only 20 percent are fully dedicated to it, and the mean supervisor is carrying somewhere between 21 and 30 volunteers. Twelve percent of programs have no budget at all.

Two answers. The first is that this schedule costs time rather than money, and most of it is time you are already spending in worse form, on recruiting replacements for the people who left in month four. The second is that the touches do not all have to come from you. The week-three conversation is better coming from the shift lead than from a coordinator the volunteer has met once, and the buddy relationship is designed precisely to move this load off the one person whose name is on the program.

There is a real caution attached to all this, and I would rather say it than have you find it later. Hager and Brudney’s 2004 work for the Urban Institute found that supervision and communication were, oddly, negatively associated with volunteer retention, and their own explanation was that heavy structured oversight starts to make volunteering feel like a job. Ninety days of scheduled contact can tip into surveillance if you let it. The difference is whether the conversation is about how they are doing or about how you are doing, and volunteers can tell the difference instantly. I have written separately about why volunteering starts to feel like paid work.

What happens after ninety days is a different problem

Everything above buys you a volunteer who is satisfied and competent at the end of month three. That is the whole of what onboarding can do, and it is not the whole of retention. The same three-stage model that says satisfaction carries the early months says something else carries the year after, and the tactics do not transfer. That is the second cliff, and it catches more programs than the first.