What Scent Does in a Shop

Answer to the question

Does scent marketing increase sales?

Nobody can say, and that is the state of the evidence rather than a dodge. In the largest synthesis of ambient scent, 671 effects from 64 articles and 15,447 respondents, scent moved almost every customer response measured except mood control and the amount of money spent. The best-controlled field experiment found no behavioural effect at all, and showed the standard design drew its positive results from groups that already differed.

Key takeaways

• Across 671 effects from 64 articles and 15,447 respondents, the two customer responses ambient scent did not move reliably were mood control and the amount of money spent.
• The three per cent sales lift quoted from that meta-analysis is a model prediction, and the same model's range runs from plus 23 per cent to minus 17 per cent.
• Christoph Teller and Charles Dennis interviewed every mall shopper on entry and again on exit, found no behavioural effect, and showed the standard design reporting pre-existing group differences as scent effects.
• Mary McGrath's randomised replication of the chocolate-scented bookstore result found no evidence that chocolate scent affects sales, and that replication has been cited once.

You are three steps inside a clothes shop and something in the air is pleasant, warm, faintly like clean laundry. Somebody chose it, and somebody invoiced for it. Scent marketing is a real industry, and the evidence base it quotes is the largest synthesis of ambient scent ever assembled: 671 effect sizes drawn from 64 articles and 15,447 respondents. That synthesis found scent moving almost every customer response it measured. The two it could not move reliably were a shopper's sense of being in control and the amount of money they spend.

What the meta-analysis found

Holger Roschk and Masoumeh Hosseinpour published it in the Journal of Marketing in 2020, pooling 71 samples, since a single article can carry more than one, from work published between 1989 and May 2018. Recall, shopping satisfaction, environmental quality, product evaluations, purchase intentions, mood valence, lingering. All positive, all significant, all small. The authors state the exception plainly: scent produced positive effects on customer responses except for mood control and expenditures.

Expenditures carried 52 of those effects. The adjusted correlation came out at roughly five in a hundred, it did not clear the significance threshold, and about three quarters of the variation between studies was left unexplained.

Scent marketing is bought for the one outcome nobody could pool.

Scent marketing is bought for the one outcome nobody could pool.

The three per cent lift that travels in every deck comes from the same paper, and it is not a measured result. It is a prediction from a moderator model with every situational condition at its average. Set the conditions favourably, with a congruent, familiar, single-note scent in a service setting and no clashing music, and the model predicts 23 per cent. Combine the less favourable ones and it predicts a fall of 17 per cent.

The honest version of that figure is a range from plus 23 to minus 17.

Roschk and Hosseinpour also close the loop on themselves, writing that although they attempted to address publication bias, studies that failed to establish ambient scent effects may not have been available, positively biasing their results. And the moderator the trade leans on hardest was pretested in only 31 per cent of the samples they pooled.

What the case for scent rests on

Eric Spangenberg, Ayn Crowley and Pamela Henderson built the founding study in 1996, in a laboratory: a simulated store of sixteen by twenty feet, 298 undergraduates, four scents chosen to have nothing to do with the goods on sale. Evaluations rose almost across the board, the store rated more favourable and more modern, the merchandise better selected, intent to visit up.

Three things in it matter as much. Time in the store did not change. The specific scent mattered far less than its presence, and both hypotheses about which scent to pick failed. And nobody spent anything, because the paper has no sales measure. Its authors write that generalisability is limited because the store was simulated.

Andreas Herrmann and colleagues did it with real money in 2013: a home decoration store in northern Switzerland, eighteen days, 402 purchasing customers, interviewers blind to the hypotheses, spend checked against receipts. A simple orange scent produced a mean basket of 56.83 Swiss francs, against 43.11 for a complex blend and 45.96 for no scent.

Two qualifications travel with it. The complex blend sat below the no-scent control, and the omnibus test is marginal, on one shop over eighteen days. The authors write that the effect of ambient scent on retail purchases is rarely seen in the literature. Their projection of one to 1.4 million dollars a year to a store is extrapolated from the paper's laboratory study, not from the shop.

Two more for the ledger. Maureen Morrin and S. Ratneshwar showed in 2003 that scent improved brand recall a day later, with attention as the mechanism, though their participants were looking at photographs on a screen. And Spangenberg's 2006 clothing store study, the one scent marketing cites most, found gender-congruent scenting doubling money spent against incongruent; those figures reach us through de Groot and colleagues, and the study carried no no-scent control.

What Teller and Dennis measured twice

Christoph Teller and Charles Dennis ran the best-designed field experiment in this literature in 2012, in a real regional shopping mall with forty tenants, over two consecutive weeks in June, scented in the second week by a professional scent marketing agency with a citrus and spice blend. They collected 312 usable questionnaires.

The design is the contribution, and it runs in three steps:

1. Interview every shopper on entry about planned spending, stores to visit, time to spend and fifty-two mall attributes, with the scent diffused or not for a week at a time.
2. Interview the same shopper again on exit, about what they actually did.
3. Use that one dataset to simulate the designs the field had been publishing.

No prior study in this field, they state, had carried that entry measurement. Simulated as a posttest-only design with a control group, which is what the studies they replicated had used, their data behaved as the literature does: the scented group rated five mall attributes higher and reported higher arousal. Simulated as a full pretest and posttest design, the authors write that the results are sobering, since none of the variables differ significantly in favour of the experimental group.

The reason should worry anyone commissioning this research. The randomly drawn groups already differed, before any scent was diffused, on the variables the weaker design then reported as scent effects. Teller and Dennis call that false support for the hypotheses. Their headline sentence covers the whole sample under every design they simulated, and the one significant behavioural result in the paper is an exception to it, confined to a subgroup.

> The most explicit finding is that ambient scent did not affect any observed or surveyed variable of consumer behaviour, no matter which experimental design was simulated.

That sentence is theirs. The exception points away from the trade's claim rather than towards it: among the third of shoppers who said scent was unimportant to them, the scented week went with lower spending and fewer stores visited than planned. Nothing in the paper pointed the other way. And the 2017 atmospherics meta-analysis that pooled thirty years of work on music, scent and colour does not cite it at all.

What the till recorded in Nijmegen

Jasper de Groot, Charly Walther and Rob Holland published the study that makes the problem visible in 2022: a second-hand clothing shop in Nijmegen, fully counterbalanced, three conditions, fresh laundry odour against citrus against nothing added, preregistered with data and materials deposited openly. They measured both things at once, 316 questionnaires and 6,781 usable till transactions.

The two measures disagree.

What was measured	Laundry smell	No added smell	Size of the effect
What shoppers said they spent, per trip	17.52 euros	13.60 euros	About a third of a standard deviation
What the till recorded, per transaction	7.95 euros	7.62 euros	About six hundredths of a standard deviation

The units are not the same: the questionnaire asked about a whole trip and the register records one transaction. On the questionnaire the omnibus test was significant. On the transactions it was not, and only a planned contrast reached the threshold at all.

Almost everything else is null: items bought, store impression, store cleanliness, product quality, subjective feeling, energy level, intention to revisit. The preregistered mediation through store evaluation failed on both paths despite adequate power. The effect at the till emerged only in the fourth and fifth hours. Citrus, pleasant and semantically clean but no use to a second-hand jumper, did nothing.

Two details sharpen it. Among the customers who noticed a smell there was no effect to speak of, and the effect lived among those who did not. And de Groot's earlier version of this study, on 57 shoppers, had reported fresh linen at 10.36 euros against 5.67 for no smell. The effect shrank as the sample grew and the measure became objective. A commercial scent supplier provided the scents.

What the only replication returned

Mary McGrath, Peter Aronow and Vivien Shotwell ran the direct test in 2016: thirty-one days in an 800 square foot independent bookshop with an adjoining café in Canada, condition assigned by a coin flip day by day, the chocolate scent diffused out of sight at either end of the shop. Staff and customers were unaware, and data and code were deposited at Yale.

Their result, in their own words: they find no evidence that chocolate scent affects sales. Not on total sales, book sales, café sales or bulk sales, and no estimate reached significance even at the loosest conventional threshold.

What they tested is the most publicised sales result in scent marketing. Lieve Doucé and colleagues had run a randomised ten-day trial in a Belgian chain bookstore in 2013, reporting total sales up 5.07 per cent on chocolate-scented days, congruent book sales up 40 per cent and incongruent up 22 per cent. The replicators note that the original reports no measure of the uncertainty around those estimates. Forbes carried it. Doucé's paper is more careful than its coverage, saying only that sales figures provide some indicative support for the findings.

Indicative support is not a sales effect.

Indicative support is not a sales effect.

The replicators are candid about their limits: twenty-one customers a day against roughly a hundred in the original, a smaller shop, a different diffusion method, and the real possibility that their study was too small to see a real effect. So the honest statement is not that the original is false. It is that it did not replicate in the only attempt anyone has made. [The Psychology You Learned Didn't Replicate](https://thewhyweproject.com/articles/the-psychology-you-learned-didnt-replicate) sets out the wider pattern.

The standard review of this field catalogues fifty-seven categories of atmospheric variable and makes no reference to replication studies.

What the trade sells

Now the numbers that sell scent marketing, chased back as far as a document exists.

The Nike figure, the percentage of shoppers said to be more likely to buy in a scented room, has no paper behind it. The earliest reachable source is a 1991 conference presentation, relayed to marketers by a trade newspaper the same year and never published in a peer-reviewed journal. The 1996 founding paper reports in print that the underlying finding is actually not statistically significant, citing an unpublished working paper.

Where the percentage itself first appeared could not be established. Every instance reachable is a vendor page or a listicle, and they do not agree with one another on the year or on what the number measured. [Why Brands Reach for Blue](https://thewhyweproject.com/articles/why-brands-reach-for-blue) traces a colour statistic the same way.

Two more, briefly. Singapore Airlines is said to have patented a cabin scent; in 2017 its then deputy vice president of public affairs said on the record that the fragrance was not developed for the airline, was not patented by it and is not used to scent cabins. Cinnabon's then president told The Wall Street Journal in 2014 that siting ovens at the back of a test store "significantly" lowered sales. That word is the only quantity on the record.

The most instructive case ended up inside a journal. Roschk and Hosseinpour's introduction relays three industry success stories, and the two sources behind them are a commercial scent vendor's web page and a 2018 Harvard Business Review advice column. That column supports its claims entirely by hyperlink, to magazines, a retail blog and an earlier column, and not one of those links reaches a journal article. Those figures entered the peer-reviewed literature through a trade column and a vendor page, with no study at either end.

The meta-analysis is not endorsing the figures, and says so: success stories inevitably portray the positive side.

One disclosure belongs in the article and this is the place for it. Your Brain on Art, by Susan Magsamen and Ivy Ross, is where this article started, and no finding here is taken from it. Magsamen founded and directs the International Arts and Mind Lab at Johns Hopkins, and Ross is a serving Vice President of Design at Google.

What the field's own reviewers say

Davide Giacalone, Bartlomiej Pieranski and Barbara Borusiak searched in 2021 for research on real customer behaviour in real shops, not laboratories, and found twenty articles. Their verdict is that neither of the field's two theoretical frameworks, stimulus-organism-response and thematic congruency, is consistently supported by the data. No study measured the concentration of scent in the air, the airflow or the air quality, and only three named the diffuser model.

> Taken collectively, such omissions seriously limit the reproducibility of such research and may explain inconsistencies in the results.

Congruence, the trade's main operating rule, does not behave consistently either way. It has support in the clothing store, in the Belgian bookstore, where approach and buying behaviour rose towards congruent books and fell towards incongruent ones, and in a moderator in the 2020 synthesis.

Now the ones that run the other way. Morrin and Ratneshwar's memory effect appeared whether or not the scent matched the category. Hendrik Schifferstein and S. T. Blok, as relayed by the replicators, found neither a congruent gain nor an incongruent loss on magazine sales in three bookshops. Doucé's 2016 field experiment found the gender-incongruent perfume beating both no perfume and the congruent one, its authors writing that the results are in contrast with existing literature. And the founding paper found presence mattered more than choice.

Combinations can run the other way entirely. Roschk and Hosseinpour explain their clashing-music moderator by reference to a 2005 study in which expenditures dropped by an average of 40 per cent when slow music met a citrus scent. That characterisation is the meta-analysis's, not the original authors'.

### Where the question actually sits

De Groot's shop comes closest to the design this question needs, and it still does not quite reach it. It has all three conditions and a register, but its two scents differ in how useful they are to the goods rather than in thematic congruence as this literature defines that term. So the rule the trade sells by has not yet met a till in a single design, and what scent marketing does to spending stays an open question.

Be clear what that does and does not concede. On ratings, memory and stated intentions the evidence is consistent and we accept it: ambient scent changes how a shop is judged and how it is remembered. [The white cup](https://thewhyweproject.com/articles/why-your-coffee-tastes-different-in-a-white-cup) is a crossmodal effect that does hold. On money there is no reliable evidence either way, and we hold that as a position, not as a shrug.

Which leaves two possible worlds. In one, scent marketing is an atmosphere purchase, like lighting or a playlist, bought to make a shop feel a particular way. In the other it is a sales instrument with a measurable return, which is how it is sold. Those are different products.

Nobody in this market prices them differently, and a buyer paying for the second product is paying for the first one.

Common questions

Does scent marketing increase sales?

The evidence does not settle it. The largest synthesis of ambient scent could not pool a reliable effect on money spent, and the best-controlled field experiment found no effect on any behaviour it observed, across every design it simulated. One Swiss home decoration store produced a larger basket on a marginal test, and one preregistered Dutch study found a small effect at the till beside a much larger one on questionnaires.

Does scent congruence matter in scent marketing?

Inconsistently, on this evidence. A clothing store result, the Belgian bookstore's approach and buying measures, and a moderator in the 2020 synthesis support the rule. A brand-memory study found its effect whether or not the scent matched the category, a magazine study found neither a congruent gain nor an incongruent loss, and one field experiment found the incongruent perfume winning.

Where does the Nike scent figure come from?

A 1991 conference presentation, relayed to marketers by a trade newspaper and never published in a peer-reviewed journal. The 1996 founding paper of this literature reports in print that the underlying finding was not statistically significant. The widely quoted percentage itself could not be traced to any document: the instances reachable are vendor pages and listicles that disagree with one another.

What does ambient scent reliably change?

Judgements and memory, not behaviour. Across the pooled literature, scent moves recall, shopping satisfaction, product and environment evaluations, purchase intentions and mood valence, all positively and all by small amounts. The founding laboratory study raised almost every rating it measured and found no change in time in store.

Sources

Roschk, H., & Hosseinpour, M. (2020). 'Pleasant Ambient Scents: A Meta-Analysis of Customer Responses and Situational Contingencies.' Journal of Marketing, 84(1), 125-145. doi:10.1177/0022242919881137. [FULL TEXT READ, open access at SAGE. Roschk first and corresponding, Hosseinpour second, both University of Klagenfurt]: the spine. 71 samples from 64 articles published 1989 to May 2018, 15,447 respondents, 671 extracted effects. Table 3 gives adjusted correlations by customer response: recall .177, shopping satisfaction .144, environmental quality .122, product evaluations .121, time elusiveness .111, purchase intentions .105, intention to recommend .098, mood activation .097, mood valence .088, lingering .078. EXPENDITURES: 52 effects, 8,150 respondents, adjusted r = .051, carrying no asterisk, which the table note defines as p greater than .05, with I squared of 76 per cent. The authors state that scent produced significant and positive effects on customer responses 'except for mood control and expenditures'. The quoted three per cent is NOT that pooled effect: it is a sensitivity calculation on Model 3, three per cent with all contingencies at their mean, ten per cent with a congruent, unidimensional and familiar scent, 23 per cent adding a service setting, no incongruent music and a female-dominant sample, and 'if the respective less favourable conditions were combined, the presence of ambient scent caused a reduction of 17 per cent, compared with its absence'. Congruency significantly raises effect sizes, b = .053, and yet 'even congruency was pretested in only 31 per cent of samples'. Authors' own limitation: 'though we attempted to address publication bias, studies that failed to establish ambient scent effects may not have been available, positively biasing our summarized results.'

Teller, C., & Dennis, C. (2012). 'The effect of ambient scent on consumers' perception, emotions and behaviour: A critical review.' Journal of Marketing Management, 28(1-2), 14-36. doi:10.1080/0267257X.2011.560719. [FULL TEXT READ, accepted manuscript at Surrey Open Research. Teller first, Dennis second]: the best-designed field experiment in this literature. A real regional shopping mall in a central European city, 30,000 square metres, 40 tenants, two consecutive weeks in June, week two scented by a professional scent marketing agency with a blend of orange, grapefruit, bergamot, cinnamon, cardamom, ginger and pimento. 136 usable control questionnaires and 176 experimental, 312 in all. Every respondent was interviewed on entry about planned spending, stores to be visited, time to be spent, emotional state and 52 mall attributes, and again on exit about what they actually did, which the authors state no prior study in this field had done. The one dataset was then used to simulate three designs. Under the posttest-only-with-control-group design, the design the replicated studies used, the scented group rated five mall attributes higher and reported higher arousal. Under the full pretest-posttest-with-control-group design, 'the results are sobering since none of the variables are significantly different in favour of the experimental group', and all three hypotheses were rejected. The randomly drawn groups already differed before any scent was diffused, on precisely the variables the weaker design showed as scent effects, which the authors call 'false support for the hypotheses'. Headline, verbatim: 'The most explicit finding is that ambient scent did not affect any observed or surveyed variable of consumer behaviour, no matter which experimental design was simulated.' The only significant behavioural result was negative, the scented group visiting fewer stores than planned. A FACT ABOUT THE FIELD, verified by reading the 2017 meta-analysis's reference list in full: Roschk, Loureiro and Breitsohl 2017 does not cite this paper anywhere.

de Groot, J. H. B., Walther, C., & Holland, R. W. (2022). 'A Fresh Look on Old Clothes: Laundry Smell Boosts Second-Hand Store Sales.' Brain Sciences, 12(11), 1526. PMCID PMC9688905. [FULL TEXT READ, open access. de Groot first, Behavioural Science Institute, Radboud University]: the preregistered study that measured the questionnaire and the cash register in the same experiment. A fully counterbalanced field experiment in a second-hand clothing store in Nijmegen, three conditions, fresh laundry odour against citrus against no added smell; 316 questionnaires and 6,781 usable till transactions; preregistration, data and materials at osf.io/ar8td. The two results do not agree. Self-reported spend: F(2,313) = 3.81, p = 0.023, laundry EUR 17.52 against no smell EUR 13.60, d = 0.33. Registered transactions: omnibus NOT significant, F(2,6860) = 2.34, p = 0.096; a planned contrast gave laundry EUR 7.95 against citrus EUR 7.61 and no smell EUR 7.62, p = 0.031, d = 0.06, and with outliers included only laundry against no smell survived. The units differ, self-report being trip spend and the register per transaction. The run of nulls matters as much: items bought, general store impression, store cleanliness, product quality, product cleanliness, subjective feeling, energy level and intention to revisit were all non-significant, and the preregistered mediation through store evaluation failed on both paths despite adequate power. The effect emerged only in the fourth and fifth hours. Citrus did nothing. Among the 269 customers unaware of a smell the effect held; among the 39 who noticed, F was less than 1. The study replicates the first author's earlier second-hand store study at N = 57, which reported fresh linen at EUR 10.36 against no smell EUR 5.67. DISCLOSURE RECORDED: the acknowledgements thank a named commercial scent supplier, Iscent, for the scents and technical support.

McGrath, M. C., Aronow, P. M., & Shotwell, V. (2016). 'Chocolate scents and product sales: a randomized controlled trial in a Canadian bookstore and cafe.' SpringerPlus, 5, article 670. doi:10.1186/s40064-016-2303-5, open access. [FULL TEXT READ. McGrath first and corresponding, Aronow second, Shotwell third; all contributed equally; no competing interests declared]: the direct replication, and the article's hinge. A 31-day trial in an 800 square foot independent bookstore with an adjoining cafe in Canada; condition assigned by Bernoulli coin flip at probability .5 independently across days; chocolate essential oil through an electric diffuser at one end and melted dark chocolate over a low heat at the other, both out of customers' line of sight; staff and customers unaware; no other change to atmosphere, personnel, layout or promotions; replication data and code deposited at the Yale Institution for Social and Policy Studies. Manipulation check on fifteen customers put detectable ambient scent at .93. RESULT: 'We find no evidence that chocolate scent affects sales.' No effect on total, book, cafe or bulk sales; 'none of our estimates of treatment effects are statistically significant at the two-tailed p < .10 level'; cafe sales directionally negative, a fall of 8.9 Canadian dollars from a control mean of 336.5. On the state of the field: with Turley and Milliman listing 57 categories of atmospheric variable, 'there is always a new atmospheric effect to investigate, and replication studies are rare'; that review 'makes no reference to replication studies'; and Teller and Dennis 2012 is 'to our knowledge the only published replication study on ambient scent atmospherics'. The replicators are candid about their limits: 21 customers a day against roughly 100 in the original, a store a third the size, a different diffusion method, and the possibility that their study was underpowered. A FACT ABOUT THE FIELD, from the Crossref item record: this replication has been cited once.

Spangenberg, E. R., Crowley, A. E., & Henderson, P. W. (1996). 'Improving the Store Environment: Do Olfactory Cues Affect Evaluations and Behaviors?' Journal of Marketing, 60(2), 67-80. doi:10.2307/1251931. [FULL TEXT READ from a JSTOR copy; the paper states the authors contributed equally]: the founding study, and it is a laboratory with no sales measure. 308 undergraduates from business classes at a large northwest university across thirteen conditions over twenty weeks, ten dropped for mentioning scent, final sample 298. A simulated store of 16 by 20 feet, deliberately stocked with unscented goods, four scents unrelated to anything on sale: lavender, ginger, spearmint and orange. Evaluations rose almost across the board, the store rated more favourable, good, positive, liked and modern, the merchandise more up to date, better selected and higher quality, and intent to visit up from 4.20 to 4.94. What it did not find: actual time in store unchanged, 577.71 seconds scented against 585.61 unscented; product tags examined unchanged; only one of three products gained in purchase intention; and both hypotheses about which scent to use failed. NOBODY SPENT ANY MONEY: there is no sales measure in the paper. The authors' conclusion: 'the specific scent used did not matter as much as the presence of the scent.' Their own limitation: 'the generalizability of this study is limited, because it was conducted in a simulated store.' The paper also supplies the primary trace for the Nike and casino trade figures.

Herrmann, A., Zidansek, M., Sprott, D. E., & Spangenberg, E. R. (2013). 'The Power of Simplicity: Processing Fluency and the Effects of Olfactory Cues on Retail Sales.' Journal of Retailing, 89(1), 30-43. doi:10.1016/j.jretai.2012.08.002. [FULL TEXT READ, open copy at Washington State University. Herrmann first]: the cleanest positive field result on real money, and it is one shop. Study 1 in a home decoration store in a large city in northern Switzerland, 10am to 7pm on weekdays across eighteen days, simple orange scent against a complex basil-orange-and-green-tea blend against no scent, conditions randomly assigned to days with at least one blank day between them. 402 purchasing customers: 103 simple, 102 complex, 198 control. Data taken only from customers who shopped at least five minutes and made a purchase, so the measure is basket size among purchasers. Interviewers blind to the hypotheses, wearing no fragrance; self-reported spend checked against receipts; advertising, pricing and availability held constant. Result: simple 56.83 Swiss francs against complex 43.11 and no scent 45.96; F(2,400) = 3.36, p = .04; Tukey put simple above both others and 'other comparisons were not significant'. The complex scent sat below the no-scent control. The authors write that 'the effect of ambient scent on retail purchases is rarely seen in the literature'. Their closing projection of one to 1.4 million dollars a year is extrapolated from Study 3, a laboratory study, not from the field data.

Giacalone, D., Pieranski, B., & Borusiak, B. (2021). 'Aromachology and Customer Behavior in Retail Stores: A Systematic Review.' Applied Sciences, 11(13), 6195. doi:10.3390/app11136195, open access. [FULL TEXT READ at MDPI. Giacalone first and corresponding, University of Southern Denmark]: the published criticism of the field's methods. Searching specifically for research on actual customer behaviour in actual retail environments, the review found 'only 20 articles meeting the inclusion criteria'. Results are indicative of positive effects on emotional states, in-store behaviour and attitudes, but 'neither the S-O-R nor the thematic congruency frameworks are consistently supported by the available data, suggesting that current theoretical explanations on the effects of scents on customer behavior may need updating'. Three critical issues named: nearly all studies analyse aggregate means despite large documented individual differences in olfaction; nearly all vary scent while holding other atmospherics fixed; and reporting is poor, with most papers giving only a commercial or lay name for the scent, no study measuring scent concentration, airflow or air quality, and only three naming the diffuser model. Verdict, verbatim: 'Taken collectively, such omissions seriously limit the reproducibility of such research and may explain inconsistencies in the results.'

Morrin, M., & Ratneshwar, S. (2003). 'Does it Make Sense to Use Scents to Enhance Brand Memory?' Journal of Marketing Research, 40(1), 10-25. doi:10.1509/jmkr.40.1.10.19128. [ABSTRACT READ IN FULL from the OpenAlex deposited abstract; full text paywalled and not retrieved. Morrin first, Ratneshwar last]: the memory result, and it is a computer screen in a room rather than a shop. Subjects evaluated familiar and unfamiliar brands while viewing digital photographs of products on a screen, with viewing times measured covertly and ambient scent diffused into the room; recall and recognition were assessed 24 hours later. Ambient scent improved both, for familiar and unfamiliar brands. Three findings the trade never quotes, all in the authors' abstract: 'This pattern emerges whether or not the scent is congruent with the product category'; the enhancement is 'due to the presence of ambient scent during encoding rather than retrieval'; and 'although apparently ambient scent did not alter subjects' self-assessed mood or arousal levels, it increased their attention in terms of longer stimulus viewing times'. The outcome is memory. Nobody bought anything.

Douce, L., Poels, K., Janssens, W., & De Backer, C. (2013). 'Smelling the books: The effect of chocolate scent on purchase-related behavior in a bookstore.' Journal of Environmental Psychology, 36, 65-69. doi:10.1016/j.jenvp.2013.07.006. [ABSTRACT READ IN FULL at the Hasselt University repository; full text restricted and not read. Douce first and corresponding]: the original of the failed replication, and the source of the most publicised sales number in this literature. A field study with 201 participants in a real bookstore chosen to have no cafe and no nearby scent-associated shops. Chocolate scent positively influenced general approach behaviour and negatively influenced goal-directed behaviour; with gender controlled it improved approach and buying behaviour toward thematically congruent books and decreased both toward incongruent books. The authors' own hedge, which the press coverage dropped: 'Sales figures also provide some indicative support for the findings.' The 5.07 per cent total-sales figure, the 40 per cent congruent and 22 per cent incongruent figures, and the absence of any measure of uncertainty around them, reach this file through McGrath, Aronow and Shotwell 2016 and are attributed to that relay. Also used: Douce, L., Janssens, W., Leroi-Werelds, S., & Streukens, S. (2016), Journal of Consumer Behaviour 15(3), 271-280, doi:10.1002/cb.1567 [ABSTRACT READ IN FULL], a field experiment at 182 participants in which a pleasant gender-INCONGRUENT perfume beat both no perfume and the gender-congruent perfume on several dimensions of customer value, the authors writing that 'These results are in contrast with existing literature'; the outcomes are customer value, not sales.

Spangenberg, E. R., Sprott, D. E., Grohmann, B., & Tracy, D. L. (2006). 'Gender-congruent ambient scent influences on approach and avoidance behaviors in a retail store.' Journal of Business Research, 59(12), 1281-1287. doi:10.1016/j.jbusres.2006.08.006. [NOT READ. Paywalled at Elsevier with no repository copy found. Bibliographic record and author order confirmed at Crossref: Spangenberg first]: the famous real-store congruence result, reached only second hand. Vanilla, pretested as the most feminine scent, and rose maroc, pretested as the most masculine, were put into the respective women's and men's sections of a clothing store. As relayed inside a peer-reviewed paper, de Groot, Walther and Holland 2022, congruency against incongruency 'doubled the amount of money spent (USD 55.14 vs. USD 23.01) and number of items bought (1.71 vs. 0.91)'. THAT IS A SECONDARY REPORT AND IS CITED AS ONE. The comparison is congruent against incongruent, and a later field experiment notes the study carried no no-scent control group, so the doubling contrasts two scented conditions rather than scenting against not scenting. NOT VERIFIED at source.

THE NIKE STORE SCENT FIGURE, TRACED. Circulated as eighty-four per cent of shoppers more likely to buy in a scented room, often with a price premium attached and dated variously 1990, 1991 or 1993. The earliest reachable source is a conference presentation, not a paper: Hirsch, A. R., & Gay, S. E. (1991), 'The Effect of Ambient Olfactory Stimuli on the Evaluation of a Common Consumer Product', presented at the Thirteenth Annual Meeting of the Association for Chemoreception Sciences, relayed to marketers by Miller, C. (1991), 'Research Reveals How Marketers Can Win by a Nose', Marketing News, 25 (4 February), 1-2. Both references read off the reference list of Spangenberg, Crowley and Henderson 1996. IT WAS NEVER PEER-REVIEWED: a Crossref title and author search returned nothing, and every OpenAlex-indexed work by Alan R. Hirsch from 1988 to 1996 was retrieved, fourteen items, and the Nike study is not among them. AND THE RESULT WAS REPORTED NON-SIGNIFICANT, in the 1996 paper's own words: 'a highly publicized finding that Nike athletic shoes were evaluated more positively in a scented than in an unscented environment (Hirsch and Gay 1991; Miller 1991) is actually not statistically significant (Bone and Ellen 1994).' That critique is an unpublished working paper by Paula Fitzgerald Bone and Pam Scholder Ellen, West Virginia University; its published successor is Bone & Ellen (1999), Journal of Retailing 75(2), 243-262, confirmed at Crossref and NOT READ. WHERE EIGHTY-FOUR PER CENT FIRST APPEARS COULD NOT BE ESTABLISHED: every instance reached is a vendor page or a listicle, contradicting one another on the year, on what was measured and on the premium. NOT VERIFIED. NOT READ: Marketing News for 1991 and 1993, print only; the 1991 conference abstract; the Bone and Ellen working paper and its 1999 successor.

Hirsch, A. R. (1995). 'Effects of ambient odors on slot-machine usage in a Las Vegas casino.' Psychology & Marketing, 12(7), 585-594. doi:10.1002/mar.4220120703. [ABSTRACT READ IN FULL from the OpenAlex deposited abstract; FULL TEXT NOT READ, paywalled at Wiley, so the methods, the identity of the odorants and how the three areas were matched are unknown. Sole author]: the one trade-famous scent figure that reached a peer-reviewed journal. For one weekend two slot-machine areas in a Las Vegas casino were odorised with different aromas and a third served as unodorised control, with amounts gambled compared across the odorisation weekend and the weekends before and after. Money gambled around Odorant No. 1 exceeded the surrounding weekends 'by an average of 45.11 per cent', with a larger rise on the Saturday when concentration was higher. And the sentence the trade drops: 'The area surrounding Odorant No. 2 did not change significantly compared to the previous weekend and following weekend of the experiment.' THERE IS NO HUMAN SAMPLE: the unit of analysis is money taken by slot machines in three floor areas of one casino over three weekends. Spangenberg, Crowley and Henderson 1996 cite the same figure to an unpublished Smell & Taste Treatment and Research Foundation report and describe it as proprietary research whose significance 'is unclear because the research has not been conducted in controlled environments nor has it been submitted to rigorous statistical analyses'.

SINGAPORE AIRLINES AND 'STEFAN FLORIDIAN WATERS', TRACED. THE AIRLINE DENIES IT, ON THE RECORD. Nicholas Ionides, then deputy vice president of public affairs, quoted in Runway Girl Network, 'Smelling a good idea, Zodiac unveils cabin scent diffusing system', by Neelam Mathews, 16 January 2017, FULL PAGE READ: 'We do purchase a fragrance called Floridian Waters, but it wasn't developed specifically for us and wasn't patented by us' and 'we do not use it to scent our aircraft cabins, as some have reported over the years. It has become associated with us over time as we use it to scent our wet towels, but we wouldn't consider it a 'signature scent' because it is used in a subtle manner and we don't have any specific rights over it.' REGISTRY CHECK: a full read of the Justia Trademarks owner listing for Singapore Airlines Limited returned fourteen marks and no Floridian Waters and no scent mark of any kind. CAVEAT RECORDED: the USPTO's own search returned empty bodies, and the Singapore register and all patent databases were NOT READ, so the trademark negative rests on a single third-party mirror in one jurisdiction. CINNABON, TRACED. Kat Cole, then president of Cinnabon, spoke to The Wall Street Journal on 21 May 2014; THE WSJ PAGE RETURNED HTTP 403 AND IS RECORDED AS NOT READ, and the quotations reach this file as block quotes relayed by Fast Company of the same date, which linked to it: the bakeries are intentionally located in malls or airports so smells can linger, aroma is a huge part of the formula, and putting ovens in the back of stores at a test location 'significantly' lowered sales. 'Significantly' is the only quantity she gave: no figure, no number of stores, no period. Any percentage attached to Cinnabon's scent is unsupported.

THE CIRCULATING SALES-LIFT NUMBERS, TRACED. Roschk and Hosseinpour's own introduction, read at source, states: 'Ambient scent has created various success stories: Hyatt Place enhanced its brand memorability, Novotel increased its breakfast sales, and Samsung found that customers underestimated shopping time by 26 per cent (Minskey, Fahey, and Fabrigas 2018; Scent Australia 2019).' SCENT AUSTRALIA IS A COMMERCIAL SCENT VENDOR, confirmed by fetching its own site; the cited page is dead and the Wayback snapshot could not be opened, so it is RECORDED AS NOT READ. 'Minskey, Fahey, and Fabrigas 2018' is a Harvard Business Review digital article of 11 April 2018, 'Inside the Invisible but Influential World of Scent Branding', and the surname is misspelled twice in the Journal of Marketing: the author is Laurence Minsky. ITS BODY IS PAYWALLED AND NOT READ; the summary, first paragraph, link targets and author bios were read. Its visible first paragraph supports its claims entirely by hyperlink, to Fast Company, an earlier HBR piece, The Independent, a Shopify retail blog and Inc., and not one link goes to a journal article. The author bios carry the relevant interest: 'Caroline Fabrigas is CEO of Scent Marketing Inc. ... her clients include Hyatt Place & Hyatt House Hotels.' THE META-ANALYSIS IS NOT ENDORSING THESE FIGURES and says so: it calls them success stories, notes that 'the success stories inevitably portray the positive side', and writes that industry reports such as Novotel's breakfast sales lie within its predictions 'whereas claims of an 80 per cent higher purchase intent (Nike) do not'. The eleven per cent sales figure could not be traced to any document at any depth. NOT VERIFIED. None of the Samsung, Hyatt Place or Novotel figures carries a sample, a method, a period or a test anywhere, and none is used in the article.

SECONDARY REPORTS AND RELAYS USED IN THE ARTICLE. Schifferstein, H. N. J., & Blok, S. T. (2002), 'The signal function of thematically (in)congruent ambient scents in a retail environment', Chemical Senses, 27(6), 539-549, doi:10.1093/chemse/27.6.539 [NOT READ DIRECTLY; design, result and wording taken from McGrath, Aronow and Shotwell 2016, which quotes it]: testing countervailing effects on magazine sales in three bookstores, 'Schifferstein and Blok found no evidence of countervailing effects: ambient scents neither increased sales of magazines deemed congruent with the scent nor decreased sales of magazines deemed incongruent.' Recorded and cited as a secondary report. THE MORRIN AND CHEBAT NEGATIVE RESULT, as relayed by the meta-analysis: Roschk and Hosseinpour 2020 explain their incongruent-music moderator by reference to 'why Morrin and Chebat (2005) observed negative sentiments, shoppers' expenditures dropped by an average of 40 per cent, when slow-tempo music was combined with a citrus (i.e., activating) scent compared with when either stimulus or both were absent'. The underlying paper is Morrin, M., & Chebat, J.-C. (2005), Journal of Service Research 8(2), 181-191, doi:10.1177/1094670505279420 [FULL TEXT NOT READ, abstract only]. THE FORTY PER CENT FIGURE IS THE META-ANALYSIS'S CHARACTERISATION AND IS ATTRIBUTED TO ROSCHK AND HOSSEINPOUR IN THE ARTICLE. Roschk and Hosseinpour also concede that 'research has also presented opposite findings, showing that the presence (vs. absence) of a pleasant ambient scent increases or decreases expenditures by up to 60 per cent', citing Madzharov, Block and Morrin 2015 and Morrin and Chebat 2005 and naming neither pole; neither full text was read and THE ARTICLE DOES NOT ASSIGN THE SIGNS. Roschk, H., Loureiro, S. M. C., & Breitsohl, J. (2017), Journal of Retailing 93(2), 228-240 [FULL TEXT READ at the Kent Academic Repository] is used in the article for one fact only: it pooled 66 studies and 135 effects across music, scent and colour on four outcomes, folded actual expenditure into behavioural intentions rather than modelling it separately, and does not cite Teller and Dennis 2012 anywhere in its reference list.

Related concepts

• The Replication Crisis
• Publication Bias
• Conflict of Interest Disclosure