Exposure depends on what is burned, in what device, where, for how long, with what ventilation and with whom nearby. Ezzati and Kammen (2001) showed in rural Kenya that individual exposure was shaped by time spent close to the fire and by peaks during lighting and stirring, so that members of one household differed widely. A review for the research community concluded that reliance on fuel category had limited the field and that better exposure assessment was its most pressing methodological need (Clark et al., 2013). At the same time, modelling work in India found that a small number of household characteristics, including fuel, kitchen location and ventilation, explained a useful share of the variation in measured kitchen concentrations (Balakrishnan et al., 2013). Questionnaire and observation data are therefore imperfect but informative.

Families that acquire a cleaner stove rarely abandon the old one. They use each for the tasks it suits, a pattern known as stove stacking (Ruiz-Mercado et al., 2011). The consequence for health is serious, because the relation between fine particle exposure and disease is steep at low concentrations and flattens at high ones, so that modest continued use of a traditional fire can cancel much of the benefit of a clean one. Large trials of improved biomass stoves and of liquefied petroleum gas have reported smaller health effects than hoped, and incomplete reduction of exposure is one of the explanations offered (Smith et al., 2011; Clasen et al., 2022). A tool that records only the primary fuel will miss stacking entirely.

Gravimetric sampling of fine particulate matter over a full day is the reference method and requires pumps, filters, a microbalance and return visits. Carbon monoxide is far easier to measure with small electrochemical monitors and has often been used as a surrogate for particles, although the relation between the two varies with fuel and stove and is weaker for personal exposure than for kitchen concentration (Northcross et al., 2010). Temperature loggers fixed to stoves give an objective record of use over weeks (Ruiz-Mercado et al., 2012). Low-cost optical particle sensors have become widely available, but their readings depend on the type of smoke and on humidity and need local calibration. The World Bank's Multi-Tier Framework shows how survey and measurement information can be combined into ordered tiers of access to clean cooking (Bhatia and Angelou, 2015).

The interview is addressed to the person who does most of the cooking. Instead of asking for the main fuel, the fieldworker asks the respondent to name every device used in the past seven days for cooking, heating water, heating the room and lighting, and for each device the fuel and the tasks it served. A photograph card of locally common stoves helps identification. Secondary uses are prompted explicitly: the fire kept for a staple dish that needs long simmering, the kerosene lamp, the charcoal brazier brought indoors on cold evenings. The output is a list from which stacking can be read directly.

The second set of questions locates cooking in space and time. The respondent indicates whether cooking is done in the main living room, a separate room, a detached building or the open air, and whether this changes with the season. Duration is asked meal by meal for the previous day, which is easier to recall than a typical day. The fieldworker then records who is usually within arm's reach of the stove while it is alight, with specific prompts for infants carried on the back, young children, and older or unwell relatives who stay indoors.

Observation does not depend on recall and is less open to the wish to give an approved answer. The checklist covers the construction of the cooking space, the number of windows and doors that open to the outside and whether they are open at the time, the presence and state of a chimney or hood, visible leaks or cracks in the stove, and soot on walls and roof. Each item has defined response categories with pictures in the fieldworker's guide. Where a stove said to be the main one is cold and clean while another is warm and blackened, the discrepancy is noted and not resolved by the fieldworker.

Where monitors are available, carbon monoxide is recorded at the cook's breathing height for the length of the cooking event under way during the visit, together with a short period before lighting. The reading is treated as an ordinal indicator of the event observed and not as an estimate of daily exposure. Programmes with access to calibrated particle sensors may add them on the same terms. Any quantitative reading reported against a health-based value should use the current guideline levels, which for fine particulate matter are an annual mean of 5 and a 24-hour mean of 15 micrograms per cubic metre (WHO, 2021), and should make clear that a single short measurement cannot show compliance with either.

The final step assigns each household to one of a small number of ordered tiers. The proposed rule is deliberately simple and transparent. The lowest tier is reserved for households that used only gas, electricity or another fuel that is clean at the point of use during the week. Households that stack a clean fuel with a polluting one are placed above them, and those relying on solid fuels or kerosene are graded by whether cooking is outdoors, in a ventilated separate kitchen, or in an enclosed living space. A flag is added when an infant or a person with chronic respiratory illness is routinely near the fire. The tiers are ordinal labels. They carry no implied concentration and should not be converted to one.