The Retention Cliff: Where Attendance Actually Collapses

Ask a commissioner what they measure for a twelve-week social connection programme and the answer is usually enrolment: how many people signed up, how many attended session one. That number is the least useful thing in the whole report. Programmes rarely fail at the door. They fail three or four weeks in, when the people who showed up out of curiosity or obligation stop showing up, and the cohort that remains is a fraction of the one that was funded.

If you commission or run group-based connection interventions — walking groups, befriending schemes, social prescribing referrals into community activities — the question that matters is not “did people join” but “at what week did they leave, and why.” Most programme reporting cannot answer that, because most funding structures never ask for session-by-session attendance in the first place.

Why week one numbers lie

Enrolment figures are inflated by two things that have nothing to do with whether a programme works: referral pressure and novelty. A link worker under pressure to hit a referral target will sign people up who have low motivation to attend. A new group, whatever its content, benefits from curiosity that a running group does not get to rely on. Both effects fade fast, and they fade at roughly the same point — once the person has sampled the activity enough times to know whether it’s for them, and before it has become a habit they’d miss.

The social prescribing literature is honest about how little is known here. A 2025 systematic review protocol notes that despite widespread adoption, the effectiveness of social prescribing for older adults remains unclear, and that only one peer-reviewed randomised controlled trial exists in the area at all. A 2021 systematic review of social prescribing and wellbeing reports gains in self-esteem and confidence but flags limited trial evidence and heavy heterogeneity across programmes — which means the studies that do exist are not comparable enough to pin down a common attrition point. No source in the current literature reports a specific week at which attendance collapses across programmes. That gap is itself a finding: the field has not been tracking the thing that determines whether an intervention survives contact with real schedules.

What the evidence does support is a mechanism, not a number. A 2022 qualitative meta-synthesis of how people experience social prescribing found that participants describe the benefit as extending beyond contact itself to restored purpose and meaningful participation — and that structured, purposeful activity appears to hold people better than unstructured contact alone. That is consistent with what programme managers report anecdotally: groups built around a task or a role people can grow into retain longer than groups built around simply being present.

Evidence status: what you can and cannot claim

Claim	Evidence status
A specific week (e.g. week 3–4) is when attendance drops across most connection programmes	Not established in the published literature; consistent with field observation, not confirmed by trial data
Structured, purposeful activity retains participants better than unstructured contact	Supported by qualitative synthesis; not yet tested experimentally against retention as an outcome
Social prescribing produces positive self-reported outcomes for those who complete it	Supported across multiple systematic reviews, but reviews report outcomes for completers, not attrition rates
Volunteering-based prosocial engagement can reduce loneliness in older adults	Supported by one randomised controlled trial (Hong Kong), one of very few controlled trials in this space
Structured befriending reduces loneliness scores over 8–16 weeks in aged care	Supported by a 2025 randomised controlled trial, with effects sustained to 16 weeks
Attrition data is routinely collected and reported in this field	Not established — most published evaluations do not report session-level attendance at all

The two randomised trials in the table are useful precisely because they had to solve the retention problem to run at all. The 2024 HEAL-HOA trial tested prosocial engagement and volunteering against a control among lonely older adults in Hong Kong and is one of the few controlled tests of a loneliness intervention rather than an uncontrolled programme evaluation. The 2025 befriending trial in aged care reduced UCLA Loneliness Scale scores by 2.39 points at eight weeks and 2.71 points at sixteen weeks against control — durability that an uncontrolled community programme, without the trial’s built-in structure and monitoring, has no comparable mechanism to guarantee.

What this means in practice: stop funding and reporting on enrolment numbers alone. Require session-level attendance data broken down by week, for every group programme you commission, from week one through the end of the funded period. You cannot manage a drop-off you have never measured, and right now almost nobody is measuring it.

The week four problem

Call it what programme managers already call it informally: the week four problem. The pattern is not a statistic drawn from a trial — no source here quantifies it precisely — but it is the point most experienced link workers and group facilitators name when asked where a cohort thins out. Three sessions is enough to sample an activity. It is not enough to have built a relationship with anyone in the room, and not enough for the activity to have become part of someone’s week rather than an item on a list. People who came out of referral compliance, rather than desire, tend to stop exactly here.

The fix is not a better opening session. It’s designing the middle. That means:

1. Build in a role or task by week three — something the person is asked to do for the group, not just receive from it, before the novelty period ends.
2. Track attendance weekly, not just at intake and at three-month follow-up, so you can see the cliff coming rather than explaining it after the fact.
3. Fund a facilitator’s time specifically for individual outreach to anyone who misses two sessions in a row — a phone call, not an email — before week four rather than after.
4. Ask funders to accept a completion-rate metric alongside an enrolment metric, so the incentive to over-refer at the start is balanced by an incentive to actually retain.

None of this is validated by a trial. It is a reasonable response to a mechanism — novelty fading before habit forms — that the qualitative social prescribing evidence does support, applied to a gap the trial literature has not yet filled.

What this does not solve

This is a retention fix, not a reach fix. Everything above assumes someone has already been referred or has already turned up once. It does nothing for the much larger group who were never referred, never heard about the programme, or heard about it and decided it wasn’t for them — the people the loneliness literature consistently shows are hardest to reach and least likely to self-select into any structured intervention at all. Fixing week four gets you a better return on the people you already have. It does not touch who you don’t have, and no attendance-tracking system will change that.