Understanding shelter needs

1Shelter: much more than a roof

Shelter fulfils several functions at once: it protects from the weather, heat and cold, it provides security and privacy, it preserves dignity and family life, and it enables people to resume activities, including economic ones. Inadequate shelter comes at a cost to health: respiratory infections, hypothermia, malaria, violence, particularly against women and girls.

The response is not limited to the structure. Sphere refers to shelter and settlement: a shelter is part of a neighbourhood, village or site, with its access routes, drainage, services (water, health, school, market) and livelihoods. In most crises, families first find shelter by their own means: with relatives, in rented housing, in public buildings or in their damaged homes. The humanitarian role is to support these strategies, not to systematically replace them with camps.

2The Sphere shelter and settlement standards

The 2018 Sphere Handbook contains seven standards: planning, location and settlement planning, living space, household items, technical assistance, security of tenure and environmental sustainability. Some essential benchmarks:

• Covered floor area: at least 3.5 m² per person, excluding cooking, bathing and sanitation spaces; in cold climates or urban settings where cooking and washing take place indoors, rather 4.5 to 5.5 m² per person.
• Planned camp-type settlements: at least 45 m² per person, including household plots (30 m² if communal services are provided outside the site).
• Fire prevention: a 30 m firebreak every 300 m of built-up area and a distance of at least 2 m between shelters, ideally twice the height of the structure.
• Location: safe land, outside areas prone to flooding or landslides, with a slope that allows drainage, and safe access to services.

These figures are indicators, not stand-alone targets: a shelter of 3.5 m² per person that is poorly ventilated, without privacy or a door that locks from the inside does not meet the standard.

3Assessing shelter and NFI needs

Assessment starts in the first hours and is refined afterwards. It combines secondary data (censuses, maps, reports), the multi-cluster/sector initial rapid assessment (MIRA) and shelter and NFI assessments through observation, key informant interviews, focus group discussions and household surveys. The aim is to find out:

• Where affected people are living: damaged house, host family, rented housing, spontaneous site, collective building, on the street.
• The level of damage to housing (minor, major, destroyed) and the causes of that damage.
• NFIs owned and missing, access to markets (materials, household items, rent) and prices.
• Tenure status (owner, tenant, informal occupant) and eviction risks.
• Local capacities: craftspeople, know-how, salvageable materials, mutual aid mechanisms.

Questions must be harmonised with the Shelter cluster so that data from several partners can be aggregated. Always consult women, men, older people and persons with disabilities separately: their needs and risks differ.

4Vulnerabilities, risks and climate

The same shelter does not have the same value everywhere. You must take into account the climate (rainy season, cyclones, extreme cold, extreme heat), hazards (floods, earthquakes, landslides), local building culture and practices (indoor or outdoor cooking, separation of spaces).

Some households need particular attention: female-headed or child-headed households, older people living alone, persons with disabilities, very large families, people without occupancy documents. The analysis must also identify shelter-related protection risks: overcrowding, no lock, insufficient lighting, distant latrines, exploitation by landlords. These elements guide both the choice of assistance option and the targeting criteria.

5Conducting a shelter and NFI assessment step by step

A shelter assessment is not a questionnaire administered once and for all. It is a process in several waves, each of which answers a specific decision: should tarpaulins be distributed tomorrow morning, which combination of options should be proposed to the donor in three weeks, which transfer value should be adopted for repairs. The experienced practitioner always starts by asking which decision the assessment must inform, and then collects only what is needed for that decision.

Wave 1: the first 72 hours

• Secondary data: census and average household size, housing typologies by area (wall and roofing materials), hazard maps, data from previous assessments, reports from civil protection authorities, satellite imagery if an emergency mapping service has been activated. An information management officer from the organisation or the cluster consolidates these elements in a simple matrix (source, date, area, figure, reliability).
• Planning estimate: number of affected households multiplied by a damage distribution assumption drawn from comparable events. It is imperfect, but it makes it possible to launch stock orders and to request an emergency allocation.
• Direct observation: a team of two (a technician and a protection or community engagement staff member) walks through the accessible areas, noting the types of damage, spontaneous gathering points and access constraints.

Wave 2: days 3 to 14, the rapid assessment

It is aligned with the Multi-Cluster/Sector Initial Rapid Assessment (MIRA) when one is organised, and with the harmonised minimum questionnaire disseminated by the Shelter Cluster at national level. The combined methods are as follows:

1. Key informants at village or neighbourhood level: village chief, teacher, health worker, representative of a women's group, owner or trader of building materials. As a rule, three to five informants per location, making sure that at least one third are women.
2. Separate focus group discussions (women, men, older people, persons with disabilities and, in urban settings, tenants) of 6 to 10 participants, lasting no more than an hour and a half.
3. Structured observation of a sample of dwellings using a damage classification form (see the fourth section).
4. Market price monitoring: corrugated iron sheets, timber, cement, nails, tarpaulins, blankets, kitchen utensils, rent for one-room and two-room units, daily wages of a mason and an unskilled labourer.

Wave 3: weeks 3 to 8, the in-depth assessment

It is used to design the programme and set the targeting criteria. It takes the form of a sample-based household survey, often with digital data collection (forms on phones or tablets), or of a full registration in the intervention areas. To estimate a proportion with a 95% confidence level and a 5% margin of error, the sample size is about 385 households for a large population; it is multiplied by a design effect, often 1.5 to 2, if clusters of villages are sampled, and 10% is added for non-response. If results are needed by district or by tenure status group, a sample is needed for each stratum.

Who does what

The modules of a shelter and NFI questionnaire

• Household composition, disaggregated by sex, age and disability (Washington Group Short Set of Questions).
• Current place of residence and pre-crisis place of residence; intention to return, stay or move on within three months.
• Type and level of damage to the original dwelling; materials; perceived capacity to repair.
• Conditions of the current shelter: floor area, number of rooms, number of people per room, weatherproofing, ability to lock, lighting, ventilation.
• Tenure status and available documents; eviction threats; rent paid and arrears.
• Household items owned and missing, with a closed list aligned with the cluster's standard kit.
• Income, debts, housing-related coping strategies (reducing meals to pay rent, sharing a room with another family).
• Assistance preferences (in-kind, cash, technical support) and preferred information and complaint channels.

Field rule: never ask a question if you do not know how the answer will be analysed. A shelter questionnaire longer than 40 minutes produces lower-quality data and wears out households already approached by several organisations.

6Applying the Sphere standards in figures and with judgement

The seven Sphere shelter and settlement standards are read at three levels: the standard (the universal qualitative statement), the key actions and the key indicators, together with guidance notes. The figures are benchmarks that must be contextualised; what is non-negotiable is the statement of the standard. A proposal writer who quotes 3.5 m² per person without explaining why this benchmark suits the local climate and practices shows that they know the handbook but not its spirit.

The key quantitative benchmarks

Interfaces with the other chapters

Shelter is not planned in isolation. The benchmarks of the water supply, sanitation and hygiene promotion chapter determine siting: in the acute phase, a maximum of 50 people per latrine, then 20, a latrine within 50 m of shelters, a water point within 500 m of the dwelling. The health and protection benchmarks influence lighting, separation of spaces and accessibility. The Humanitarian inclusion standards for older people and people with disabilities point out that a ramp, a door handle at the right height and a stable floor are part of the shelter, not optional extras.

The Core Humanitarian Standard (CHS) applied to shelter

• Commitment 1 (appropriate and relevant response): assistance options derive from an assessment, not from available stock.
• Commitment 3 (strengthening local capacities, avoiding negative effects): avoid destabilising building material or rental markets, avoid deforestation caused by timber harvesting.
• Commitment 4 (communication, participation): households understand why they receive a kit rather than cash assistance and can give their views on the shelter design.
• Commitment 5 (complaints): an accessible mechanism, including for reporting exploitation by a landlord or a distribution agent.

Exercising professional judgement

Sphere explicitly recognises that an indicator may not be met. In that case, the gap must be documented, the reasons explained (lack of land, restrictions imposed by the authorities, funding), mitigation measures described (two-storey shelters, reduction of communal areas offset by managed shared spaces) and a plan made to close the gap. A donor or an evaluator forgives a documented gap; they do not forgive an ignored one.

An unmet Sphere indicator is not a failure in itself. The failure lies in not having measured it, not having reported it and not having planned anything to reduce its effects on people.

7Classifying damage, tenure and vulnerability: the tools

Three tools structure the analysis: a common damage scale, a tenure grid and a vulnerability grid. They must be harmonised with the cluster; otherwise the figures of several organisations cannot be added up and gaps are poorly estimated.

Tool 1: damage classification scale

Enumerators who are not engineers must never declare a category 3 building "safe". In seismic contexts, the authorities often use a colour code (green, yellow, red) applied after inspection by certified engineers: the humanitarian assessment aligns with this code instead of creating a competing one.

Tool 2: tenure grid

Tool 3: vulnerability grid for targeting

The grid combines shelter need (damage, current conditions) and capacity to recover (income, available labour, social network). A simple weighting, discussed with communities, gives better results than a complex score that nobody understands.

Thresholds (for example, priority from 7 points upwards) are set after testing the grid on a real sample and checking that it does not produce absurd exclusions. A community verification committee, including women, reviews borderline cases. Criteria must never rely on sensitive data that cannot be protected.

8Full example: the shelter assessment after the floods in the municipality of Basse-Ternes

Fictional context. The Ternes river bursts its banks at the end of the rainy season and floods the rural municipality of Basse-Ternes: a semi-urban main town and seven villages. Census data give 9,800 households with an average size of 5.6 people. The rainy season is expected to last another six weeks, followed by a cool season in which night temperatures drop to 12 °C. Your NGO is present in the area for a water and sanitation programme; the national Shelter Cluster asks you to cover the municipality.

Step 1: planning estimate (day 2)

Civil protection estimates that 4,200 households have been affected. Data from a comparable flood in 2019 showed 25% of dwellings destroyed, 35% with moderate to severe damage and 40% with light damage. Estimate: 1,050 households without housing, 1,470 with a dwelling to be repaired with technical support, 1,680 needing light repairs. On this basis, you request 1,500 shelter kits and 2,500 NFI kits from the cluster's prepositioned stock, and you inform the donor that a funding request will follow within ten days.

Step 2: rapid assessment (days 4 to 10)

Four pairs of enumerators visit the eight locations: 31 key informants (including 12 women), 22 focus group discussions, structured observation of 410 dwellings selected at random along transects, and price monitoring in three markets. Consolidated results:

Markets: the main town has two wholesalers of corrugated iron sheets and timber; the price of corrugated iron sheets has risen by 35%, while the price of mud bricks is stable because production has resumed. The villages are supplied from the main town, an hour and a half away by road. One-room rents in the main town have risen by 20% in two weeks.

Step 3: analysis

Households without housing (category 4): 26% of 4,350, i.e. 1,131 households. 29% of them, i.e. 328 households, live on the embankment, exposed to a new flood: a life-saving priority. 204 households (18%) occupy schools, which are due to reopen in eight weeks.

Space requirement: for the 328 households on the embankment, an average of 5.6 people and a benchmark of 3.5 m² per person give about 19.6 m² per household. The cluster's standard shelter kit (two 4 × 6 m tarpaulins) covers a structure of about 18 to 20 m² once overlaps are made: acceptable in the short term, insufficient for the cool season. Families of more than 7 people (14% of the sample) will receive a third tarpaulin.

Repairs: 37% in categories 2 and 3, i.e. 1,610 households, about a quarter of which are category 3 and must be inspected by an engineer before any repair. As mud bricks are available and their price is stable, wall repairs are feasible locally; the roof is the critical point.

NFIs: 50% of affected households, i.e. 2,175 households, lost bedding and kitchen utensils. The markets in the main town sell these items; those in the villages do not.

Step 4: recommendations sent to the cluster and the donor

What makes this example good quality

• The planning figures were replaced by measured figures, and the difference (4,350 instead of 4,200) was communicated to the cluster.
• The analysis distinguishes between the main town and the villages: a municipal average would have masked the high proportion of tenants in the main town and the exposure of villagers on the embankment.
• Each recommendation is linked to a quantified group, a justification (market, risk, tenure status) and a timeframe.
• The limitations are stated in writing: an observation sample not representative of inaccessible neighbourhoods, prices collected in only three markets, return intentions not measured.

A convincing shelter assessment always ends with a table that links each group of households to a figure, an assistance option and a timeframe. Without this table, the analysis informs no decision.

9Common mistakes, donor expectations and contextual variations

The most common mistakes

• Counting houses, not households: one destroyed house may have sheltered three households. The unit of analysis must be the household, with the size of each household.
• Ignoring people hosted by relatives: invisible in sites, they often make up the majority of those without housing. Site-focused assessments overestimate the need for tents and underestimate the need for host family support.
• Confusing needs with distribution wishes: asking "do you need a tarpaulin?" almost always produces a yes. The condition of the shelter must be observed and factual questions asked.
• Classifying damage without training: two untrained enumerators will classify the same house differently. Half a day of training with photographs and a day of field calibration greatly reduce this discrepancy.
• Forgetting the market: without price and availability monitoring, it is impossible to defend a cash modality before the donor.
• Collecting unnecessary sensitive data: ethnicity, precise legal status, identity of perpetrators of violence. If such data do not serve a decision, they must not be collected.
• Not giving feedback: communities that never receive feedback on an assessment refuse the next one.

What donors, evaluators and auditors look at

• DG ECHO: in its guidance on shelter and settlements, the European Commission emphasises tenure analysis, the settlements approach, priority given to solutions that support household strategies (repair, hosting, renting) and the justification of modalities. The project document (Single Form) must show the link between measured needs, options and indicators.
• USAID Bureau for Humanitarian Assistance (BHA), when it funds: for the shelter and settlements sector, its application guidelines require a description of damage, tenure status and markets, standard indicators by sub-sector and, for construction, the involvement of qualified technical staff and compliance with environmental regulations.
• OCHA-managed pooled funds: alignment with the priorities of the response plan and the cluster, coverage of gaps identified in the 4W, no duplication.
• Evaluators (OECD DAC criteria, ALNAP guide on evaluating humanitarian action): relevance of options to measured needs, coverage of invisible groups, coherence with other sectors.
• Auditors: traceability of beneficiary lists back to assessment forms, consistency of the criteria applied, evidence of independent verification.

Variations by context

10Interview questions and key references

Interview questions and elements of a strong answer

1. "How would you organise the shelter assessment in the ten days following an earthquake?" A good answer distinguishes the waves (secondary data and planning estimate, rapid assessment aligned with the cluster and the MIRA, in-depth assessment), cites the methods (key informants, separate groups, structured observation, market monitoring), and mentions harmonisation of tools with the cluster, inspection of buildings by engineers and feedback to communities.
2. "A site shows 3.2 m² per person. What do you do?" Check the figure per household rather than as an average; identify the households furthest below the benchmark; analyse the causes (land, restrictions, population growth); propose measures (site extension, two-storey shelters, new land, host family support); document the gap and report it to the cluster and CCCM.
3. "Why analyse tenure status from the assessment stage onwards?" Because it determines the possible options and the risks of eviction; because tenants and informal occupants are often excluded for not having been counted; because donors and the Sphere standard require it.
4. "How do you prevent the assessment from creating expectations?" Clearly present the objective and the fact that no assistance is promised; obtain consent; coordinate to avoid multiple assessments; share the results and the decisions taken.
5. "What sample size for a household survey in three districts?" About 385 households per district if 5% precision at 95% confidence is required for each, adjusted for the design effect and non-response; explain that the precision required depends on the decision to be made.
6. "How do you include persons with disabilities in the shelter assessment?" Washington Group Short Set of Questions, specific focus group discussions, observation of the accessibility of shelters and services, targeting criteria that include the need for construction assistance.

Key references

• The Sphere Handbook, 2018 edition: Humanitarian Charter and Minimum Standards in Humanitarian Response, Shelter and Settlement chapter.
• The Core Humanitarian Standard on Quality and Accountability (CHS) and its indicators.
• Global Shelter Cluster: guidance on shelter assessments, technical notes from working groups and coordination tools published on its website.
• Multi-Cluster/Sector Initial Rapid Assessment (MIRA) guidance of the Inter-Agency Standing Committee (IASC).
• Shelter Projects: collection of case studies and overviews published by the Global Shelter Cluster, useful for comparing contexts.
• IFRC and OCHA: Shelter after disaster (2010 edition).
• Humanitarian inclusion standards for older people and people with disabilities (Age and Disability Consortium).
• Washington Group Short Set of Questions on Disability.
• European Commission, DG ECHO: guidance document on humanitarian shelter and settlements.
• ALNAP: Evaluation of Humanitarian Action Guide.

In an interview, quoting a Sphere figure is not enough. What sets a senior profile apart is the ability to say what they do when the figure is not met, and how they document it.