Every so often, a single number about rough sleeping in the UK makes the news, reported as though it were a precise, settled fact about a fixed and easily observed population. It rarely is. Counting how many people are sleeping rough on any given night is one of the hardest measurement problems in social policy — not because nobody is trying hard enough, but because the population being counted is, almost by definition, trying not to be seen. Understanding the gap between what a count can capture and what is actually happening on the ground matters enormously for how charities, local authorities and funders like New Capital Link Foundation think about the scale of the problem and where to put resources.
How a snapshot count actually works
Official rough sleeping figures in the UK are typically built from what's known as a snapshot count: volunteers and outreach workers go out on a single, chosen night and record verified sightings of people sleeping rough within a defined area, sometimes supplemented by local knowledge from services that work with people day to day. It's a sensible, practical method. There is no register of rough sleepers and no central database that updates in real time, so a physical count on an agreed night is one of the few ways to produce a comparable figure at all. But the method's strength — a single, consistent snapshot — is also its central limitation. A count of this kind can only ever record the people its volunteers actually see, in the places they happen to look, on the night they happen to look there.
Why hidden homelessness doesn't show up
A great deal of rough sleeping is, by its nature, hidden. People sleep in cars, in stairwells, in industrial units, behind hoardings, or in spaces well away from the routes a count typically follows. Others move between friends' sofas, night buses or all-night venues specifically to avoid being seen sleeping outside at all — a pattern usually described as "hidden homelessness" rather than rough sleeping, even though the underlying insecurity is much the same. Some people actively avoid count locations, whether because of past bad experiences with services, distrust of anyone in an official capacity, or simply a preference to be left alone. None of that is a flaw in the volunteers' effort; it's an inherent limit of trying to count a population that includes people who, for good reason, don't want to be found.
The night itself shapes the number
Because a snapshot count only covers one night, conditions on that specific night can move the figure up or down without any real change in the underlying scale of the problem. Bitter weather can push people into squats, stairwells or emergency shelter that falls outside a count's usual routes, artificially lowering the number even as the underlying need deepens. How many volunteers turn out, how thoroughly a local area is covered, and simple chance can all shift the result too. Comparing one count to the next, then, is comparing two snapshots taken under two different sets of conditions — useful as a rough indicator of direction, but a much blunter instrument than the single, precise-sounding figure usually reported.
"Rough sleeping" is narrower than "homelessness"
It's also worth being precise about what the term actually covers. Rough sleeping — people bedding down outdoors or in places not designed for habitation — is only one part of a much wider homelessness picture that includes people in temporary accommodation, hostels, refuges, insecure or overcrowded housing, and people sofa-surfing with no home of their own. Someone can be entirely without a stable, secure home and never appear in a rough sleeping count at all. Treating the rough sleeping figure as a stand-in for "how much homelessness there is" understates the scale of housing insecurity considerably, simply by definition.
A floor, not a ceiling
None of this means official counts are worthless — quite the opposite. They are a genuinely useful, consistently gathered baseline, and movement over time is real information worth paying attention to. But the right way to read any given figure is as a floor rather than a ceiling: a reliable minimum, almost certainly an undercount, rather than a complete picture of everyone currently without a bed. For an organisation deciding where to direct funding, time or volunteering hours, that distinction changes how much weight a single headline number should carry on its own.
Why ground-level knowledge counts as data too
This is exactly where organisations working directly on the street earn their value as more than delivery partners. A group like Homeless Street Angels, out in Leeds night after night, builds a granular, current picture of who is sleeping rough, where, and why — a picture that a periodic snapshot simply cannot replicate. Knowing which doorways are occupied through a cold snap, which faces are new, which are missing, or who has moved into a squat rather than being found on a count route, is not a soft substitute for "real" statistics. It is data, gathered continuously rather than as a single snapshot, and it deserves to be weighed alongside official figures rather than treated as anecdote that sits beneath them.
For a funder, that means asking a second question after "what does the number say": namely, how was this number produced, and what does it most likely miss. It means treating a flat or falling official count with some caution rather than as an automatic sign the problem is shrinking, and treating outreach organisations' first-hand knowledge of who they're actually seeing on the ground as a genuine, complementary source of insight — not a lesser one. The statistics tell us the floor. The people working the streets every night tell us, more often than any single count can, how much further the real picture is likely to extend beyond it.