A Step-By-Step Method For Tracking Who Your Social Prescribing Programme Never Reaches

A programme that reports 78% improvement on a loneliness scale and a programme that reports 12% improvement can both be doing equally poor jobs at the thing that matters most: getting to the people who are actually isolated. Outcome data only describes people who enrolled. If your referral pathway systematically misses the housebound, the recently bereaved, the non-English-speaking, or the person too depressed to answer a phone call, your outcomes look fine and your reach is failing. Most social prescribing evaluation stops at the first number.

This is a procedure for building a reach measure alongside your outcome measure, not instead of it. Both are needed because they answer different questions: outcomes tell you if the intervention works for the people it touches; reach tells you whether it is touching the people it was funded to touch.

Why outcomes alone mislead

Systematic reviews of social prescribing consistently report positive outcomes for people who complete a programme — improved self-esteem, reduced loneliness scores, in some cases lower use of GP and emergency services. But the same reviews are candid about a structural weakness: the trial evidence is thin, heterogeneous, and says almost nothing about who was excluded before enrolment. One review protocol assessing social prescribing for older adults found only a single peer-reviewed randomised controlled trial existed in the entire field, despite widespread adoption. A qualitative synthesis found participants describing genuine, meaningful benefit — but by definition, that synthesis can only report on people who stayed long enough to be interviewed.

None of this means outcomes are worthless. It means an outcomes-only report tells a funder about programme quality for the enrolled population, and says nothing about programme reach into the target population. Those are different claims, and commissioners increasingly ask for both, partly because national loneliness strategy work has pushed standardised measurement as a precondition for comparing programmes at all.

The six-step procedure

Step 1: Define your target population before you define your enrolled population. Write down, in a sentence, who this programme was funded to reach — not who typically walks through the door. If the funding brief says “socially isolated older adults in this postcode,” that population includes people who never contact a GP, never use a smartphone, and never hear about the group. Your denominator is not “referrals received.” It is an estimate of the eligible population, drawn from local demographic and health data.

Step 2: Build a referral funnel with a stage for every drop-off point. Map the full pathway: eligible population → aware of the service → referred → attended once → attended regularly → completed. Most programmes only measure the last two or three stages. Put a number, even a rough one, against every stage. The gaps between stages are where reach is lost, and they are usually invisible in an outcomes report because outcomes reports start counting at “attended once.”

Step 3: Track who refers, not just how many. If 90% of referrals come from one enthusiastic GP practice or one link worker, your reach is a proxy for that person’s caseload, not a population-level result. Record referral source by practice, team, or channel, and check the demographic profile of each source against the target population from Step 1.

Step 4: Compare enrolled demographics against the eligible population, not against national averages. A programme that is 90% female and mostly under 70 might look fine against national loneliness statistics, but not if the local eligible population — say, isolated men over 75 — looks nothing like that. National surveys of adults 45 and older show loneliness driven strongly by network size and physical isolation, and those risk factors are unevenly distributed by neighbourhood and household type. Local comparison, not national benchmarking, is what exposes a skewed intake.

Step 5: Record non-completion reasons, not just non-completion rates. “Did not attend” is not a category, it is a shrug. Distinguish between: could not get there (transport, mobility), did not feel it was for them (stigma, mismatch), tried once and did not return (fit problem), and never made first contact (awareness or referral failure). Each has a different fix, and lumping them together produces a retention number with no diagnostic value.

Step 6: Report reach and outcomes on the same page, not in separate documents. A funder reading only the outcomes section sees a success story. Put the funnel table directly above the outcomes table so the two are read together. This is uncomfortable — it means presenting your own coverage gap — but it is the only version of the report that is actually honest about what the programme has and hasn’t done.

Evidence status

Claim	Evidence status
Social prescribing produces positive outcomes for people who complete it	Reasonably supported across multiple systematic reviews, though trial quality is weak
Social prescribing reduces service use (GP, A&E, inpatient)	Reported in a minority of included studies; not a reliable general claim
Current evaluation practice adequately measures who is excluded	Not supported — the literature is almost silent on pre-enrolment attrition
Third-place and community-space closures reduce reach into certain neighbourhoods	Supported; closures since 2019 concentrated in socially vulnerable and rural areas

The self-selection ceiling

Give this failure mode a name, because it recurs in every reach conversation: the self-selection ceiling. A programme improves its outcomes by unconsciously tightening its intake — link workers refer people who seem likely to engage, participants who struggle quietly drop out early and aren’t chased, and the remaining group produces good numbers. The ceiling isn’t a policy decision anyone made; it’s the aggregate effect of everyone doing something locally sensible. The fix in Step 3 and Step 5 is designed specifically to surface it, because it will not surface on its own.

What this means in practice: if your annual report has an outcomes table but no funnel table, you do not yet know whether your programme works or whether it has simply learned to select for people who were always going to do well. Build the funnel table first. It costs you nothing but a spreadsheet and some referral-source tagging, and it is the difference between a plausible success story and a defensible one.

What this does not solve

Tracking reach tells you where you are losing people. It does not tell you how to reach the person who was never in any referral pathway to begin with — the person with no GP contact, no community group membership, no digital access, living in a neighbourhood where the local library or community centre has already closed. Third-place availability research shows those closures are not evenly spread; they are concentrated in exactly the areas with the highest social vulnerability, which means the population hardest to reach is also the population with the fewest physical spaces left where reach might happen informally. A better funnel measure makes the gap visible. It does not fund the venue, staff the outreach, or rebuild the infrastructure that closed years before your programme existed.