Scent Marketing Research and Statistics: What the Peer-Reviewed Evidence Actually Says

Scent marketing statistics travel badly. A modelled best-case estimate becomes "scent lifts sales 23%" in a sales deck, a single small study becomes "research shows," and a null result disappears. This page lists the primary sources we rely on, what each one measured, and the number it actually supports. Every entry links to the journal record or an open-access copy; we do not cite papers we have not read.

1. The meta-analysis: Roschk & Hosseinpour (2020)

Journal of Marketing, 84(1). doi:10.1177/0022242919881137 (open access).

• Integrates 671 effects from ambient-scent experiments.
• Pleasant ambient scent "on average produces a substantial increase in the level of customer responses (3%–15%)."
• Spending: modelled increases of 3% for an average condition and 23% for a most favorable condition. These are model outputs, not observed sales. Quote them that way.
• Effects are "more cognitive than affective": scent moves evaluation and judgment more than mood.
• Effects are larger when the scent is congruent with the setting, reads as one clear idea (unidimensional), is familiar, and is used in a service setting rather than goods retail.
• Perceived concentration has nonlinear effects: intensity has an optimum.

2. The field experiment: Girard, Lichters, Sarstedt & Biswas (2019)

Journal of Service Research, 22(4). doi:10.1177/1094670519842333 (open access).

• German regional railway; jasmine, melon and violet-leaf fragrance delivered through the trains' existing air-conditioning, which is HVAC delivery.
• Study 1: 204 passengers, scented vs. unscented coaches. Service quality 4.24 vs 3.77, service experience 3.82 vs 3.19, service value 3.69 vs 3.05; short-term effects small to medium.
• Study 2: four months, nine waves, 35 commuters completing all waves. Long-term effects larger; two weeks after the scent was removed, experience and value ratings had not significantly declined.
• Only 27% of passengers consciously recognized the scent.
• Caveat: the samples are modest and the setting is a train. Describe it as a four-month field experiment on a working railway, never as large-scale.

3. The systematic review: Giacalone, Pierański & Borusiak (2021)

Applied Sciences, 11(13), 6195. doi:10.3390/app11136195 (open access).

• 20 studies met the bar: real retail environment, ambient scent, behavioral variables, experimental data.
• Most report positive effects on emotional states; evidence for behavioral effects is inconsistent.
• Neither the S-O-R nor the thematic-congruency framework is consistently supported, so congruency is supported but not settled.
• Methodological gap: scent composition rarely reported, intensity "generally not reported," and only 3 of 20 studies specified the diffuser or its placement, which are exactly the variables that decide whether a program works.

4. The null: McGrath, Aronow & Shotwell (2016)

SpringerPlus. PMC4899340 (open access).

• Randomized controlled trial, 31 days, chocolate scent in a Canadian bookstore and café.
• No statistically significant effect on total, book, café, or bulk sales.
• Caveat the authors themselves raise: the store was 800 square feet with about 21 customers a day, so statistical power was low. It is an honest null, not proof of no effect.

5. Simplicity and processing fluency: Herrmann, Zidansek, Sprott & Spangenberg (2013)

Journal of Retailing, 89(1), 30–43 (no stable open link; cited from the article).

• Home-decoration store, 18 days, 402 purchasing customers.
• Average spend: simple orange scent 56.83 CHF; complex basil-orange-green-tea blend 43.11 CHF; no scent 45.96 CHF. The complex blend performed no better than no scent.
• Three lab studies attribute the mechanism to processing fluency. "Simple" means the scent reads as one idea, not that it has few ingredients.
• Caveat: only two scents were used across all four studies.

6. Intensity: Leenders, Smidts & El Haji (2016)

Journal of Retailing and Consumer Services. doi:10.1016/j.jretconser.2016.05.007.

• Major supermarket, melon scent at three intensity levels; outcomes included actual time in store and store-level sales.
• Only the high intensity moved evaluations, time in store, and sales; lower levels did not. Mood mediated the effect.
• We do not quote a sample size for this study because we could not confirm it from the accessible pages.

7. Individual variation: Liu, Awada, Becerik-Gerber, Lucas & Roll (2023)

Journal of Environmental Psychology, 91, 102135.

• 48 young adults, bergamot via electric diffuser, simulated office task under time pressure.
• Both sexes self-reported less stress; physiology diverged, with men showing increased heart rate and rating the office less pleasant than control.
• Limits: one scent, one intensity, student-age sample, simulated task.

How to read all of this together

Scent reliably changes how a space is evaluated. It sometimes changes behavior. The size of the effect depends on fit, intensity, and delivery, and the published literature mostly fails to report those variables. That is the evidence-based reason the test that matters is the one run in your own building, which is why every Aroma Impressions program starts with a seven-day on-site trial. For the plain-language version of this page, see what scent marketing is and why it matters; for the social-behavior evidence from hotel and bar settings, see does scent make people more social?

Numbers we do not use

"Scent increases sales by 20%" without a source. "Scent increases dwell time by 40%." "75% of emotions are triggered by smell." None of these trace to a peer-reviewed study we can read. If you see one of them on this site, it is a mistake; tell us and we will remove it.