When to Kill a Workplace Connection Programme

Most guidance on workplace connection programmes tells you how to start one. Almost none tells you how to know it’s time to stop. That gap costs money and, worse, it costs credibility: a programme that limps on past its useful life is the reason the next good idea gets a harder time in the budget meeting.

You are not looking for a reason to cut things. You are looking for the signal that separates “this needs a redesign” from “this has run its course.”

The three questions that matter

Before you touch attendance numbers, answer these in order.

1. Is anyone still joining who didn’t already know each other? A peer support group, lunch club, or interest circle exists to create ties that wouldn’t otherwise form. If the same eight people who were already friends are the only ones showing up, you’re subsidising a friendship group, not building social infrastructure. That may still be worth doing, but it’s a different budget line with a different justification.

2. Has attendance become a floor, not a trend? Every recurring programme settles into a baseline of regulars. The question is whether that baseline is stable or declining. A stable core of 12 out of an eligible 200 might be fine for a niche offer. A core that was 40 eighteen months ago and is now 12, with no new joiners, is not a plateau — it’s a slope you’re still on.

3. Would the organisers tell you if it wasn’t working? This is the one nobody asks, and it’s the one that matters most. Programme staff and volunteer facilitators have every incentive to report steady-state success, because the alternative is dismantling the thing they built. Build in a channel — a survey, an exit interview, a third party — that doesn’t route through the people running it.

Named failure mode: the loyalist trap

Programmes that survive well past their usefulness almost always survive because of a small, vocal, high-attendance group who love it. Call this the loyalist trap. The loyalists show up to every session, rate it highly on every survey, and will lobby hard if you propose changes. Their satisfaction is real. It is also not the metric you need.

A programme built to address workplace loneliness is meant to reach people with thin networks — not deepen ties among people who already have them. Robert Putnam’s distinction between bonding and bridging social capital is useful here: bonding capital strengthens existing ties, bridging capital creates new ones across groups that wouldn’t otherwise connect. A programme that has drifted into pure bonding for a fixed group of loyalists is not doing what most connection budgets are meant to fund. That doesn’t make it worthless. It makes it a different, smaller thing than the one you approved, and it should be priced and evaluated as that.

What the evidence actually supports here

Be honest about how thin the underlying evidence base is before you build a stop/continue decision on it.

Claim	Evidence status
Lonely workers use more sick leave and report more stress-related absence	Supported — Cigna’s 2020 workplace loneliness report found lonely workers miss work roughly twice as often for illness and five times as often for stress
Structured peer or social prescribing-style programmes reduce loneliness or improve self-reported wellbeing	Weak positive signal — systematic reviews report improvements in self-esteem and confidence, but trial evidence is limited and heterogeneous across programme designs
Social connection functions as a preventive health factor comparable to diet or exercise	Supported at population level — but this is a case for having some connection infrastructure, not for keeping any specific programme running
A given workplace programme is reaching people with genuinely thin networks, not just socially confident volunteers	Almost never measured — most programmes do not track this at all

That last row is the one to sit with. If you don’t know whether your programme is reaching people who were otherwise isolated, you don’t have grounds to claim it’s addressing loneliness — you have grounds to claim it’s popular with people who liked it enough to keep coming.

A decision procedure, not a vibe check

Run this at the six-month and twelve-month marks, not just at annual budget review.

1. Pull attendance data and split it into first-time attendees vs. repeat attendees for the last quarter.
2. If first-time attendees are under 10% of total attendance and falling, the programme has stopped recruiting and started maintaining.
3. Ask, separately from organisers, whether attendees report having made a new working relationship they wouldn’t otherwise have had. Not “did you enjoy it” — that question always scores well and tells you nothing.
4. Check whether the programme was originally scoped to reduce isolation among a specific group (new starters, remote workers, a team going through change) and whether that group is still who’s showing up.
5. If the answer to 2 and 4 both point toward drift, don’t cancel outright — reset scope, target, or format for one more cycle, then apply the same test.
6. If the second cycle produces the same result, stop it. Redirect the budget to something with a defined start and end point rather than an open-ended standing offer.

What this means in practice: don’t measure a connection programme by whether people like it. Measure it by whether it is still doing the one thing it was funded to do — creating ties that didn’t exist before. When it stops doing that, it has become a perk for people who already had what the programme was meant to provide. That’s a legitimate thing to fund. It is not the thing you should keep calling a loneliness intervention.

What this does not solve

This procedure tells you when to stop a programme. It does nothing to tell you whether the underlying need has gone away. The American Enterprise Institute’s 2021 survey found 12% of Americans report no close friends at all, up from 3% in 1990 — a shift in baseline social capacity that no single workplace initiative will offset. Killing a stale programme is a governance decision. It is not evidence that the problem it was built for has been solved, and it should not be reported to a funder as if it were.