Research · Paper 07 · August 2026
Portfolio owners reasonably want to understand an estate without surveying every building in it. A sample can support some conclusions and not others, and the distinction is not a technicality. This paper sets out what a sample of roofs can legitimately be used for, what it cannot, and how to structure one so the answer is worth having.
A sample characterises the population it was drawn from. Nothing more.
It supports prioritisation. It does not support a conclusion about any individual building.
Stratify before sampling, or the sample describes a population that does not exist.
01 · The core distinction
These are different exercises with different requirements, and conflating them is the most common and most expensive error in portfolio work.
The question is: across this estate, where should attention and budget go first? A sample can answer this well. It identifies which building types, construction eras or roof forms are carrying the most risk, and it does so at a fraction of the cost of surveying everything.
The question is: can this roof carry this array? A sample cannot answer this at all, at any sample size. The conclusion applies to the building assessed, and no proportion of an estate surveyed changes that. A roof that has not been assessed has not been assessed.
A sample presented as characterising an estate is frequently received as licensing installation across it. The result is that buildings never assessed are treated as cleared, and the first genuine assessment happens after commitment — which is precisely the position sampling was meant to avoid.
02 · Structuring the sample
A random sample of an estate assumes the estate is homogeneous. Commercial estates almost never are, and the variation runs along axes that matter structurally.
Pre-1980, 1980 to 2000, and post-2000 stock behave differently, are documented differently, and fail differently. A sample that does not distinguish them will produce an average that describes none of them.
A profiled metal sheet on a portal frame, a ballasted flat roof, a north-light saw-tooth and a converted mill roof are four different problems. Averaging across them produces a figure that is arithmetically correct and practically useless.
Under BS EN 1991-1-3 ground snow load is derived from the site and rises with altitude, and under BS EN 1991-1-4 wind demand varies with terrain and distance to shoreline. An estate spread across the country is not one loading environment. Two buildings of identical construction can return different answers for reasons that have nothing to do with the buildings.
The stratification is itself a useful deliverable. Frequently a portfolio owner does not know, in structured form, what they own — and establishing that costs little and changes how the rest of the programme is planned.
03 · What the output should say
Reported per stratum rather than as an estate-wide average, because the average conceals precisely the variation the exercise was commissioned to find.
Selected at random within strata, selected because they were accessible, or selected because somebody already had a concern. These produce very different figures, and the third is not a sample of the estate at all — it is a sample of the estate’s known problems.
Prioritisation, budget shaping, and identification of which parts of the estate warrant full assessment first.
Any conclusion about an individual unassessed building. This should be stated in the output rather than left to be inferred, because it is the sentence that gets lost between the report and the decision.
04 · Basis and limitations
05 · Citation
Suggested citation
Raihan, S. (2026) How Many Roofs Characterise an Estate: Sampling, Stratification and What a Sample Cannot Support. Solar Surveys Research, Paper 05. Solar Surveys Ltd. Available at: https://solarsurveys.co.uk/research/how-many-roofs-characterise-an-estate
Published under Creative Commons Attribution 4.0. Reproduce, quote and build on this with attribution. Writing about it and want something clarified, or a point expanded? Ask. We would rather it was reported correctly than quickly.
Solar Surveys Research
On-site structural surveys and roof condition assessments for commercial buildings across the UK, signed by a qualified structural engineer.