Measure Your Volunteers the Way You Measure Your Participants

A funder asks how many volunteers you will need to hit next year’s target. You have a headcount — say 42 on the books — and a rough sense that “some drop off”. You cannot answer the question, because a headcount is a stock and the funder is asking about a flow.

Most connection programmes measure participants carefully and volunteers barely at all. Attendance registers, wellbeing scores and case notes on one side; a spreadsheet of names and DBS dates on the other. That asymmetry is a measurement failure with two consequences. You cannot forecast capacity, and you cannot see that a large share of the people benefiting from your programme are the ones delivering it.

This is a procedure for fixing both. It assumes you have no volunteer data worth the name and are starting from a standing start.

Step 1: Build the funnel before you build the dashboard

Count volunteers as a sequence of stages, each with a defined trigger. The point of defining the trigger is that two people in your team should be able to look at the same volunteer and agree which stage they are in.

Stage	Trigger that counts	What the number tells you
Enquiry	Any inbound contact expressing interest	Whether your recruitment channels work
Application	Completed form submitted	Whether your ask is too heavy
Screening cleared	References and checks returned	Where your administrative bottleneck sits
Training complete	Attended all required sessions	Whether training is scheduled often enough
First shift delivered	Volunteer has run or supported one session	Your real conversion rate
Third shift delivered	Three sessions completed	Early retention
Six months active	At least one shift in each of the last two months	Sustainable capacity

Do not set targets for these stages in your first two cohorts. You are establishing a baseline, not hitting a benchmark, and any target you pick now will be invented. Record the numbers, then look at them after cohort two and decide which conversion rate is unacceptable.

The single number most programmes cannot produce is enquiry-to-first-shift conversion. It is the one that tells you how many people you must attract to staff a rota.

Step 2: Put a clock on every stage

Counts alone will mislead you. Add elapsed days between each stage for every individual, and report the median rather than the mean — a handful of volunteers who take eight months to clear checks will drag an average into meaninglessness.

The two intervals that matter most are enquiry to screening cleared, and training complete to first shift. The first is almost entirely within your control and is usually where enthusiasm dies. The second is where you will find the failure mode below.

The trained-and-never-deployed problem

A person applies in March, waits five weeks for checks, attends training in mid-April, and is then told the next suitable slot is in June. They do not refuse. They simply stop answering. By the time you notice, you have spent the recruitment cost and the training cost and received nothing, and your headcount still shows 42 because nobody formally left.

Two metrics catch it:

1. Share of trained volunteers who have not delivered a first shift within 21 days of training. This is your leakage rate at the most expensive point in the pipeline.
2. Share of first-shift volunteers who never deliver a second. A single shift with no follow-up is usually a scheduling failure, not a motivation failure.

The fix is almost always sequencing rather than persuasion: book the first shift during the training session itself, with a named session and a date, before anyone leaves the room. Measure whether you did that as a process indicator — percentage of trainees leaving with a booked first shift — because it is the lever, not the outcome.

Step 3: Measure training by what it changes

Attendance registers tell you nothing about capability. If your training is meant to equip volunteers to hold a conversation with someone who is isolated, notice risk and refer on, then measure those three things directly.

Use a short pre and post self-rating on specific tasks — “I know what to do if a participant discloses that they are in crisis” — rather than a global confidence score. Then add one behavioural check at 90 days: has this volunteer made a referral, escalated a concern, or asked for supervision? A cohort in which nobody ever escalates anything is not a well-trained cohort; it is an under-supervised one.

Keep the instrument stable across cohorts. The review of loneliness and isolation research published in BMC Public Health in June 2023 identified inconsistent measurement as a central barrier to comparing findings, and the same problem operates inside a single organisation that redesigns its feedback form every year.

Step 4: Track retention by cohort, not by headcount

Monthly headcount hides churn. Forty-two in January and forty-two in June can conceal eleven leavers and eleven joiners, which is a very different programme with very different training costs.

Group volunteers by the month they completed training and track each cohort’s active share at 1, 3, 6 and 12 months. This gives you a curve, and the shape of the curve tells you where to intervene. It also lets you test changes properly: if you move to booking first shifts at training, the September cohort’s three-month figure is the test.

What this means in practice

If you collect only two new things this quarter, collect the date of each volunteer’s first delivered shift and the month of their training cohort. Those two fields make retention curves and pipeline forecasting possible retrospectively. Almost nothing else you might add has that property.

Step 5: Count your volunteers as beneficiaries

Volunteering is plausibly an intervention on the volunteer, and the qualitative meta-synthesis published in BMC Health Services Research in 2022 found that participants in social prescribing describe benefit extending past social contact into restored purpose and meaningful participation — which is a fair description of what a volunteer role offers. The systematic review in the International Journal of Environmental Research and Public Health in 2021 reported gains in self-esteem and self-confidence as key outcomes, while noting how thin the trial evidence is.

There is now at least one properly controlled test. The HEAL-HOA dual randomised trial, published in The Lancet Healthy Longevity this month, tested prosocial engagement and volunteering against a control condition among lonely older adults in Hong Kong. That matters because the intervention literature is otherwise dominated by small uncontrolled studies, a gap the American Heart Association named explicitly in its 2022 scientific statement as the central research problem in this field.

So administer the same loneliness measure to volunteers that you administer to participants, at intake and at six months. Use a validated instrument, not a bespoke one. The AARP Foundation’s 2018 survey of adults 45 and older used the 20-item UCLA Loneliness Scale precisely so its findings would sit alongside the academic literature; if 20 items is too long for a volunteer induction, the shorter set the UK embedded into Office for National Statistics collection from its 2018 loneliness strategy gives you national comparators.

What the evidence will support

Claim	Evidence status
Volunteers benefit from the role, not just participants	Consistent qualitative and observational support; one randomised trial now tests it directly
Structured, purposeful activity beats contact alone	Qualitative meta-synthesis; plausible but not trial-established
Social participation programmes raise self-esteem and confidence	Systematic review support, limited and heterogeneous trials
Training improves volunteer effectiveness	Essentially no published evidence specific to connection programmes. Measure it locally or do not claim it
Health systems should routinely assess isolation	Recommended by the National Academies in 2020; implementation evidence still thin

Be blunt with funders about row four. A claim that your training works, backed only by attendance figures and satisfaction scores, invites the obvious question and you will not have an answer.

What this does not solve

None of this changes who volunteers. Volunteer pools skew towards people who are already socially connected, already confident in institutional settings, and already have discretionary time — the same selection problem that runs through most programme design, where nearly everyone you reach came looking. Better pipeline data will tell you that your last four cohorts were demographically near-identical. It will not tell you how to recruit the fifth one differently.