Research report template in IMRaD format, with tables and figures

Background. Cafes are designed for conversation, yet many customers describe them as too loud to talk in. Little local data links measured sound levels in cafes to what customers experience. Aim. To measure sound levels in urban cafes across the week and relate them to customers' reported difficulty holding a conversation. Method. A-weighted equivalent continuous sound levels (LAeq) were measured for 15 minutes in each of 12 cafes on three occasions, 36 sessions in all, between 3 August and 13 September 2026, while 312 customers completed a five item questionnaire.

Results. The mean session level was 72.4 dB(A), ranging from 63.8 to 81.6 dB(A). Weekday lunch was the loudest period at a mean of 75.8 dB(A). Cafes with mainly hard surfaces averaged 75.2 dB(A), 6.6 dB louder than those with soft furnishings. The share of customers reporting difficulty conversing rose from 11.1 percent below 65 dB(A) to 85.2 percent at 80 dB(A) and above. Conclusion. Most cafes studied were loud enough at peak times to make conversation difficult for a majority of customers, and surface finishes were the factor most clearly associated with lower levels.

Keywords: room acoustics, hospitality, speech communication, sound level measurement, customer experience.

Cafes have become places to meet, work and talk, and their interiors have changed to match: exposed concrete, timber floors, glass shopfronts and open kitchens are now common. These surfaces reflect sound rather than absorb it, so the noise of an espresso machine, music and other customers builds up. Restaurant and cafe reviews often mention noise, but owners rarely have measurements to act on.

Guidance on noise usually addresses either the community or the workplace. The World Health Organization's environmental noise guidelines concern noise from transport and leisure affecting people in their homes and surroundings, and occupational limits such as the European Union's lower exposure action value of 80 dB(A) over an eight hour day concern hearing damage to workers. Neither sets a level for comfortable conversation in a hospitality venue, which is the question this study addresses.

Twelve independent cafes within two kilometres of the city centre were selected from a list of 41 to cover a range of sizes, from 24 to 96 seats, and interior finishes. Each was classified by the two researchers on a walk through as mainly hard surfaced, mixed, or soft furnished, using the floor, ceiling and seating materials. Owners gave written permission, and the cafes are identified here only by number.

Each cafe was visited three times: a weekday morning between 8 and 9 am, a weekday lunch between 12 and 1 pm, and a weekend late morning between 10 and 11 am. In each session a Class 2 sound level meter meeting IEC 61672-1, calibrated before and after with an acoustic calibrator, recorded LAeq over 15 minutes. The meter was placed at a central table at 1.2 metres, seated head height, at least 2 metres from the counter and any speaker. Occupancy was counted at the start and end of each session.

During each session, one researcher invited customers who had been seated with at least one companion for ten minutes or more to complete a five item paper questionnaire, shown in Appendix A. The key item asked whether, in the last ten minutes, the customer had found it difficult to hear the person they were talking with. Responses were linked to the session level, not to individual positions in the room. Participation was anonymous and voluntary; 312 of 377 customers approached took part, a response rate of 82.8 percent.

During each session, one researcher invited customers who had been seated with at least one companion for ten minutes or more to complete a five item paper questionnaire, shown in Appendix A. The key item asked whether, in the last ten minutes, the customer had found it difficult to hear the person they were talking with. Responses were linked to the session level, not to individual positions in the room. Participation was anonymous and voluntary; 312 of 377 customers approached took part, a response rate of 82.8 percent.

During each session, one researcher invited customers who had been seated with at least one companion for ten minutes or more to complete a five item paper questionnaire, shown in Appendix A. The key item asked whether, in the last ten minutes, the customer had found it difficult to hear the person they were talking with. Responses were linked to the session level, not to individual positions in the room. Participation was anonymous and voluntary; 312 of 377 customers approached took part, a response rate of 82.8 percent.

Session levels were summarised by period and by surface type. Customers were grouped into five bands of session level, and the share reporting difficulty was calculated for each band; the association between session level and the session share reporting difficulty was measured with Pearson's correlation. The study was approved by the Halverston University Human Research Ethics Committee, reference HREC-2026-118.

The 36 session levels had a mean of 72.4 dB(A) and ranged from 63.8 dB(A), a weekday morning in a soft furnished cafe, to 81.6 dB(A), a weekday lunch in a hard surfaced one. Lunch was the loudest period in 11 of the 12 cafes. Figure 1 shows the mean for each period.

Occupancy explained part of the difference: lunch sessions averaged 78 percent of seats filled against 46 percent in the morning. But levels did not rise in step with occupancy alone. Two cafes with similar lunch occupancy, cafes 2 and 11, differed by 7.9 dB, and both were classified differently for surfaces.

Table 1 gives each cafe's mean across its three sessions. Grouped by surfaces, soft furnished cafes averaged 68.6 dB(A), mixed 71.8 dB(A) and hard surfaced 75.2 dB(A). The five hard surfaced cafes were all at or above 72.6 dB(A), and the three soft furnished cafes were all at or below 70.8 dB(A), despite including some of the busiest rooms.

Overall, 142 of the 312 customers, 45.5 percent, reported difficulty hearing the person they were talking with. The share rose steadily with the session level, as Figure 2 and Table 2 show. Across the 36 sessions, the correlation between session level and the share of customers reporting difficulty was strong, r = 0.71.

Staff spend whole shifts in these rooms. Using the three period means for a weekday shift covering morning and lunch, the estimated eight hour level for counter staff is about 73 dB(A), below the 80 dB(A) lower exposure action value. Levels at the coffee machine itself, which this study did not measure, are likely to be higher, so this estimate is a floor, not an assessment.

This section interprets the three results against the research questions, sets out what the design cannot show, and draws practical points for the people who run cafes.

Typical cafe levels in this sample sat around 72 dB(A), and at lunch most cafes passed 75 dB(A), the band in which nearly two thirds of customers reported difficulty conversing. The relationship between level and difficulty was close and steady, without an obvious threshold, which suggests that every few decibels of reduction matters to customers rather than a single target.

Surface finish was the feature most clearly associated with level. Hard surfaced cafes were on average 6.6 dB louder than soft furnished ones, a larger gap than the difference between morning and weekend sessions. Because sound reflected from hard surfaces builds on itself, adding absorption, such as acoustic ceiling panels, upholstered seating or wall hangings, is a practical lever that does not require fewer customers.

The sample is twelve cafes in one city, chosen for variety rather than at random, so the figures describe this sample, not all cafes. Each session lasted 15 minutes at one position; levels near the counter or in corners will differ. Music volume was not controlled or recorded. Surface classification was a researcher judgement rather than a measured absorption value, and the questionnaire relied on self report. Customers who found a cafe too loud may have left before being invited, which would lead the study to understate difficulty.

Customers in the cafes studied were often in rooms loud enough to make conversation difficult, especially at weekday lunch. The level customers experienced was closely tied to the difficulty they reported, and hard interior finishes were associated with markedly higher levels. Future work should measure absorption directly, record music levels, and test whether acoustic treatment in a sample of cafes changes both the measured level and customers' experience.

Customers in the cafes studied were often in rooms loud enough to make conversation difficult, especially at weekday lunch. The level customers experienced was closely tied to the difficulty they reported, and hard interior finishes were associated with markedly higher levels. Future work should measure absorption directly, record music levels, and test whether acoustic treatment in a sample of cafes changes both the measured level and customers' experience.

Customers in the cafes studied were often in rooms loud enough to make conversation difficult, especially at weekday lunch. The level customers experienced was closely tied to the difficulty they reported, and hard interior finishes were associated with markedly higher levels. Future work should measure absorption directly, record music levels, and test whether acoustic treatment in a sample of cafes changes both the measured level and customers' experience.

Customers sitting alone were not approached, so all 312 respondents were in conversation when they answered item 3. No names or contact details were collected.

LAeq over 15 minutes for every session, dB(A). The mean column is the arithmetic mean of the three sessions and matches Table 1; the bottom row gives the period means shown in Figure 1.