The First Night Effect: Why Guests Sleep Worse in Your Listing, and the Six Fixes That Work

In a lab, one brain hemisphere stays on watch the first night somewhere new. Over 80% of arousals trace to it. Six fixes ranked by evidence per dollar, verified October 6, 2026.

On a first night in an unfamiliar room, one half of the sleeping brain stays partly awake: slow-wave activity drops in the left hemisphere, and more than 80% of that night's arousals trace back to it. By the second session a week later the asymmetry is gone. That is the first night effect, and it is the single biggest structural advantage a hotel has over your listing.

It is also smaller and stranger than the version circulating in hospitality content, and the most-quoted statistic about it has no source at all. What follows is the actual evidence, the places it does not transfer, and six fixes ranked by how much sleep they buy per dollar. The cheapest one is a curtain lining and a roll of electrical tape.

What the first night effect actually is

It is a measured asymmetry between brain hemispheres during the first sleep cycle in a novel environment, and the researchers who found it called it a night watch. It was measured on 11 to 13 young adults in a laboratory, and that qualifier matters more than anything else in this post.

The paper is Tamaki, Bang, Watanabe and Sasaki, "Night Watch in One Brain Hemisphere during Sleep Associated with the First-Night Effect in Humans", Current Biology 26, pages 1190 to 1194, 2016. Three experiments of 11, 13 and 11 participants, two sessions roughly a week apart, instrumented with magnetoencephalography, structural MRI and polysomnography, analysing the first sleep cycle.

• Slow-wave activity in the left default-mode network was significantly smaller than the right on night one (t = 2.59, p = 0.027, d = 0.8), and significantly smaller on night one than on night two (t = 2.69, p = 0.023, d = 0.8).
• The left hemisphere showed an augmented evoked response to deviant sounds on night one (t = 2.92, p = 0.013, d = 0.9). By night two the difference between hemispheres was not significant.
• Participants were aroused more often on night one (p = 0.002), and more than 80% of those arousals came from left-hemisphere trials, significantly above chance (p = 0.001). On night two that dominance was absent (p = 0.751).
• Reaction time to a deviant sound was significantly faster on night one (p = 0.021), which is the detail that makes "night watch" more than a metaphor.

The authors conclude that "one brain hemisphere may work as a night watch during sleep in a novel environment," and that the hemisphere with reduced sleep depth is the one with the enhanced response to external stimuli. Half your guest's brain is listening for your boiler.

The present study has demonstrated that when we are in a novel environment, interhemispheric asymmetry occurs in regional slow-wave activity, vigilance, and responsiveness as a night watch to protect ourselves.

The older literature, and the study that complicates it

The clinical version has been measured since 1966 and the effect sizes are real but modest. Agnew, Webb and Williams studied 43 young adults and found time awake at 3.95% of night one against 0.84% of night two, and latency to first REM at 134.19 minutes on night one against 97.98 minutes by night four. Their own verdict, worth keeping: "Adaptation to laboratory sleep appears to be quite rapid in terms of these particular variables."

A 2016 replication of 24 healthy men found sleep efficiency of 89.15% on night one against 95.01% on night two (p = 0.011), with wake after sleep onset falling from 32.26 minutes to 11.50 minutes (p = 0.006). Sleep latency and REM latency differences were not statistically significant in that sample.

Then the study that should stop you overclaiming. Sprajcer and colleagues ran 124 adults across two consecutive nights of polysomnography with nine-hour sleep opportunities, and found no significant night-one difference in sleep onset latency (p = .295), total sleep time (p = .343), wake after sleep onset (p = .410) or sleep efficiency (p = .342). The only differences were architectural: more stage 1, less stage 2, less stage 3. Their conclusion was that where primary sleep variables are the focus rather than architecture or arousals, an adaptation night may be unnecessary.

So the honest summary is narrow and still useful. The first night effect holds up well for arousals, light sleep and sleep architecture, and weakly or not at all for how long people actually sleep. Across the three studies the one objective measure that moved consistently was wake after sleep onset, 2.8 times higher on night one in the 24-person replication. A guest in your listing probably does not lose an hour of sleep. They probably wake more, sleep lighter, and remember the night as bad. The thing they write a review about is the waking, not the total.

Which is the only part of this you can actually observe. You cannot measure a guest's slow-wave activity, but MagicBNB's Guest Experience dashboard aggregates reviews portfolio-wide with a rating breakdown by category (cleanliness, communication, location, value, accuracy), guest sentiment spotlights and top-praise themes. A door losing a tenth of a point on accuracy while holding cleanliness is telling you something specific went wrong overnight, and that signal is invisible in an overall star average.

The statistic everyone quotes has no source

The claim that more than half of people sleep worse on their first night in a new place is not supported by any peer-reviewed or government source we could find. The figures that circulate come from surveys commissioned by hotel and mattress companies, and the best-known one reports the opposite of what it is used to prove.

The Westin Hotels sleep study of August 2009, roughly 1,000 respondents per country across twelve countries, found that 57% said a great hotel bed provides a better sleep experience than at home, with 59% preferring hotel mattresses and 54% preferring hotel pillows. A hotel-commissioned survey finding people sleep better in hotels is not evidence of a first-night penalty. It is marketing, and it points the wrong way for the people who cite it.

The peer-reviewed prevalence data runs the same unhelpful direction. Xiong, Fan and Qi, Frontiers in Psychology volume 11, article 724, 2020, surveyed 505 hotel guests across 28 hotels and found a home insomnia rate of 24.2% against a travel insomnia rate of 15.4%. People reported sleeping better away than at home. New travel-induced insomnia affected 7.5% of the sample, while 16.2% recovered from home insomnia while travelling. The strongest risk factor for travel insomnia was not the room at all: business travel carried an odds ratio of 16.99.

Take the correction seriously, because it changes the brief. You are not fixing a problem that afflicts most guests. You are fixing one that afflicts a minority badly, concentrated in morning chronotypes, short sleepers, people highly sensitive to their sleep environment, and business travellers. Those are also the guests most likely to write about it.

And nobody has ever run the laboratory experiment in a hotel or a rental. Not once. So the hemispheric night watch is a plausible mechanism for a feeling everyone recognises, not a measured property of your spare room. An instrumented lab night also confounds novelty with electrodes, wires and being watched, which your guest does not have.

The six fixes, ranked by evidence per dollar

Light is the strongest evidence and the cheapest fix, mattresses are the weakest evidence and the most expensive, and the obvious nightlight fix does not do what you think. In that order.

1. Light. Strongest evidence, lowest cost

Roughly 100 lux, which is ordinary dim room light, delivers about half the melatonin suppression of 9,000 lux bright light. That is the finding, and once you know what 100 lux looks like you will find it in every listing you own.

Zeitzer and colleagues, Journal of Physiology volume 526, pages 695 to 702, 2000, studied 23 healthy volunteers over a nine-day protocol. The abstract is unusually blunt about the surprise: "Contrary to expectations, we found that half of the maximal phase-delaying response achieved in response to a single episode of evening bright light (approximately 9000 lux) can be obtained with just over 1% of this light (dim room light of approximately 100 lx)."

A 2022 study in the Proceedings of the National Academy of Sciences took it further, comparing one night of 100 lux during sleep against a dimly lit environment below 3 lux in healthy adults. Light during sleep increased nighttime heart rate, decreased heart rate variability, and increased next-morning insulin resistance. One night.

• A hallway light left on, a bathroom light under a door, a bright streetlamp through unlined curtains and a cluster of appliance standby LEDs all land in the territory where the measured harm starts.
• The fix is blackout lining rather than blackout-coloured fabric, which are different products and only one of them blocks light.
• Electrical tape over every standby LED in the bedroom costs almost nothing and removes a light source that runs all night, every night, in every booking.
• If you provide a night light, provide a dim one low to the floor rather than an overhead, for reasons the fourth fix explains.
• There is no official lux guidance for bedrooms. We checked WHO, CDC and ASHRAE. The thresholds above come from chronobiology research, not from a standards body, so treat them as evidence rather than a code requirement.

2. Noise. Strong evidence, moderate cost, and the fix is the gap not the wall

The World Health Organization recommends night noise below 45 decibels for road traffic, 44 for rail and 40 for aircraft, and at 40 decibels roughly 3% to 4% of a population still report being highly sleep disturbed. Those are outdoor population limits for policymakers, not a specification you can install.

The guidelines are the Environmental Noise Guidelines for the European Region, published by WHO's Regional Office for Europe in 2018, and all three night-time values are strong recommendations set specifically against "adverse effects on sleep." The decision rule the guideline group used was that the absolute risk at the chosen value should not exceed 3% of the population highly sleep disturbed.

The number you can actually act on sits in building code. Section 1206.4 of the 2025 California Building Code requires that interior noise from exterior sources "shall not exceed 45 dB in any habitable room," and section 1206.2 sets a minimum Sound Transmission Class of 50 between sleeping units. Both are construction requirements most operators cannot change, which is why the practical work is elsewhere.

• Guest-room sound leaks through gaps, not through walls. A door undercut and an unsealed perimeter will defeat a wall built to STC 50.
• A door sweep and perimeter seals on the bedroom and bathroom doors cost less than one night's revenue and are the highest-yield acoustic work available to a leaseholder.
• Soft-close hardware on cupboards and the bathroom door removes the specific category of noise, a sharp transient, that the Tamaki study suggests a first-night brain is primed to catch.
• A fan or white-noise machine the guest controls is the only intervention here that works on noise you do not own, such as traffic and neighbours.
• No ANSI, ASTM or ISO standard recommends an STC rating for hotel guest rooms. The "STC 50 to 55 recommended for hotels" figure appears only on acoustics vendor pages with no standards citation. Do not let a contractor quote it to you as code.

3. Temperature. Good evidence, and the popular numbers are about the wrong thing

Sleep was most efficient between 20 and 25 degrees Celsius, which is 68 to 77 Fahrenheit, with a 5% to 10% drop in sleep efficiency as temperature rose from 25 to 30 degrees. The widely repeated figures of 29 and 30 degrees are not thermostat settings and using them as such will make your listing uncomfortably hot.

The range comes from Baniassadi and colleagues in Science of the Total Environment, volume 899, article 165623, 2023, studying community-dwelling older adults with wearables and environmental sensors in their own homes. The authors note "substantial between-subject variations" and stress personalised adjustment, so read it as a band rather than a setpoint. It is also a familiar-environment study in an older cohort, which is a different population from your average guest.

Here is the correction that matters. A 2012 review in the Journal of Physiological Anthropology reports a thermoneutral temperature of 29 degrees Celsius, and that figure is for semi-nude subjects with no bedding. The same review found that with normal bedding and clothing, air temperatures from 13 to 23 degrees produced no significant difference in sleep. Figures around 30 to 32 degrees describe the bed microclimate under the covers, not the room. Conflating the two is the most common error in consumer sleep-temperature writing, and it is why so many listings arrive stifling.

ASHRAE publishes no sleeping-temperature setpoint. Its Standard 55 product page does not mention bedrooms or sleeping occupants at all, and the standard is sold rather than published. No CDC or WHO bedroom-temperature guidance exists either. If a cleaning checklist tells you to set the thermostat to a number, that number came from somebody's opinion.

The actionable version is unglamorous: a thermostat the guest can reach, read and understand, pre-set to arrive at the low twenties Celsius, with instructions that do not require the app.

4. Wayfinding in the dark. Good evidence, and it cuts against the obvious fix

A guest who gets up at 3am is measurably less steady than they are in daylight, and adding room lighting does not fix it. That is the finding, and it is the opposite of what every hospitality checklist recommends.

McBean and colleagues, Journal of Clinical Sleep Medicine volume 12, pages 1477 to 1486, 2016, tested 21 healthy adults of mean age 64.7 across four overnight laboratory stays under varying illuminance. They found that "over one-third of the tested gait and balance parameters were significantly worse at the midsleep awakening as compared to baseline," with nearly a quarter showing medium to large effect sizes. The conclusion carries the sting: "This impairment is not ameliorated by exposure to room lighting, when compared to dim lights."

The impairment is sleep inertia, not darkness, and light alone does not reverse it. Which means the right intervention is removing the hazard rather than illuminating it. It matters most for older guests: 27.6% of US adults aged 65 and over reported a fall in 2020.

• Clear the bed-to-bathroom route completely. No loose rugs, no trailing cables, no low furniture corner at shin height.
• Contrast-mark any step or level change on that route, because low-contrast visual acuity falls sharply as illuminance drops.
• Provide a grab point where there is a level change, which is the one structural fix worth paying for in a property that takes older guests.
• A very dim path light low to the floor is still worth having, because obstacle detection needs strikingly little light, and because a dim one costs you far less melatonin than an overhead.
• No CDC or NIH source publishes a figure for the share of falls that happen at night, in the dark, or in unfamiliar places, and none publishes an effectiveness estimate for night lights. If you see such a number, it was invented.

5. Bedding and mattress. Weakest evidence, highest cost

The convergent answer is medium-firm, and that is about as much as the literature will support. This is the section to be most honest in, which is easy because it is also the most expensive fix.

A 2025 randomised crossover in Nature and Science of Sleep put 12 young adults of normal BMI through three foam mattresses. Medium firmness beat soft on sleep latency, 7.71 minutes against 12.42 (p < 0.05), and produced fewer stage transitions than firm. But total sleep time did not differ significantly (p = 0.28), and neither did subjective discomfort ratings. A 2021 narrative review of 39 articles in the Journal of Orthopaedics and Traumatology concluded that "a medium-firm mattress promotes comfort, sleep quality and rachis alignment," with no pooled effect size and an explicit limitation that standardisation of volunteers was low.

Twelve people, one night per condition, one material, and the objective sleep measure that matters most showed nothing. Buy medium-firm, replace what is visibly worn, and then stop. 12 participants is not a basis for a 4-figure purchase. Every dollar beyond that buys less sleep than the same dollar spent on blackout lining and a door sweep, and you now have the citations to know why.

6. The one hotels actually do: have a specification at all

No major hotel brand publishes its guest-room standards, and the competitive point is not that they hit a number you are missing. It is that they have a number, enforced across every room, and most operators running 2 to 20 doors have none.

We checked. Marriott's development site publishes no temperature setpoints, blackout specification, mattress specification or acoustic specification. Its own Franchise Disclosure Document, a genuinely public filing dated March 31, 2024, confirms a system of "design criteria and specifications for your applicable brand" and reserves the right to "unilaterally add to, merge, discontinue, or otherwise modify components of the system at any time," while keeping the manuals themselves undisclosed. Hyatt and IHG publish nothing comparable. The manuals that circulate on document-sharing sites are unauthorised copies of confidential franchisee material and we did not open them.

So write your own. One page, applied to every door, checked on every turnover: curtain lining present and closing fully, no visible standby LEDs, door sweeps intact, thermostat at the arrival setting, bed-to-bathroom route clear. That list is free and it is the actual thing a hotel is selling.

Finding which door is quietly failing it is the harder half. MagicBNB's Discovery spotlights are AI-generated insight cards that name patterns rather than metrics: "same revenue different ops", "silent winner", "fast decliner", "cleaning burden", "issue magnet." A door flagged as an issue magnet while its neighbours are not is usually an environmental problem rather than a guest-quality problem, and it is the door where this checklist pays for itself first.

What a better night is actually worth

A two standard deviation improvement in rating raised expected occupancy from 56% to 63% in a study of 16,144 London properties, a 12.5% relative increase, with price held constant. On a six-door portfolio at a $240 average rate that is roughly $36,792 of additional annual revenue, computed from their published result rather than measured on your doors.

The study is Sahadev, Seiler and Scarf, Journal of Vacation Marketing 31(2), pages 351 to 365, 2025, covering 16,144 peer-to-peer properties in London between June 2017 and May 2018. Its methodological contribution is exactly the thing most review research omits: "we control for price variation when estimating these review-occupancy effects." Rating carried a coefficient of 0.083 and sentiment 0.012, both significant. Two standard deviations is a large move, so do not translate it into one more star.

What guests complain about is the other half of the equation, and sleep is not at the top. A 2021 analysis of 59,766 negative reviews across 27,980 Airbnb listings in twelve cities ranked dirtiness and sanitary conditions as the strongest driver of dissatisfaction at a coefficient of minus 0.457, ahead of property issues at minus 0.318. Sleep disturbance appears as a topic but does not lead. Fix cleanliness first. This post is about what to fix second.

The ranking link runs through Airbnb's own documentation, and every step is quotable. Airbnb's search page states that the algorithm evaluates "ratings and reviews from guests," and that for homes it "considers the Superhost criteria," which include ratings and reviews, when ordering results. Superhost requires a "4.8 or higher overall rating" maintained, a cancellation rate under 1%, a 90% response rate within 24 hours, and at least 10 reservations or 3 totalling 100 nights.

One correction while you are here, because it is in almost every host blog. Airbnb does not publish a 4.9 rating threshold for Guest Favorite. Its help page describes a percentile system, highlighting the "Top 1%, 5%, and 10% of homes," with a minimum of five reviews in the past four years. No numeric rating cutoff appears anywhere on that page. If a guide tells you to hit 4.9 for the badge, it invented the number.

The leading indicator is not the rating, it is whether guests bother to write at all. MagicBNB's Retention Leader card surfaces the property with the highest review-submission rate among doors with five or more reservations, which is the operational version of the question this whole post asks. A guest who slept badly and left quietly costs you a review you will never see; a door where people consistently take the time to write is a door where something is going right overnight.

Frequently asked questions

What is the first night effect?

It is a measured asymmetry in how the two brain hemispheres sleep during a first night in an unfamiliar place. In a 2016 Current Biology study, slow-wave activity in the left default-mode network was significantly reduced on night one, the left hemisphere showed an enhanced response to deviant sounds, and more than 80% of that night's arousals traced to left-hemisphere trials. By a second session a week later the asymmetry was gone. The authors describe it as one hemisphere acting as a night watch.

Do most people really sleep worse the first night somewhere new?

There is no primary source for the popular claim that more than half of people do. The peer-reviewed prevalence data points the other way: a survey of 505 hotel guests found travel insomnia at 15.4% against a home insomnia rate of 24.2% in the same sample. The strongest lab effects are on arousals and sleep architecture rather than total sleep time, and the largest study, 124 adults, found no significant night-one difference in total sleep time or sleep efficiency at all.

What temperature should I set for guests?

The best available evidence puts the band at 20 to 25 degrees Celsius, or 68 to 77 Fahrenheit, with sleep efficiency falling 5% to 10% as temperature rises from 25 to 30. Be careful with figures around 29 to 32 degrees that circulate as ideal: those describe the bed microclimate under the covers or semi-nude subjects without bedding, not room air. ASHRAE publishes no sleeping setpoint and no government agency publishes bedroom temperature guidance, so anyone quoting a precise number as a standard is quoting an opinion.

Is a night light a good idea in a short-term rental?

A dim one low to the floor, yes. A bright one, no, and for a reason most checklists miss. A 2016 study of 21 older adults found over a third of gait and balance measures significantly worse during a mid-sleep awakening, and concluded that "this impairment is not ameliorated by exposure to room lighting, when compared to dim lights." Brightness does not restore steadiness, and 100 lux is already enough to cost your guest melatonin. Clear the route instead, and keep the light dim.

Does better sleep actually improve my bookings?

Indirectly, through reviews, and the effect is measurable. A study of 16,144 London short-term rental properties found that a two standard deviation increase in rating raised expected occupancy from 56% to 63%, a 12.5% relative increase, with price controlled. Airbnb's own documentation confirms ratings and reviews feed search ranking directly and through Superhost status. Note that cleanliness outranks everything else as a driver of negative reviews, so fix that first.

What do hotels do that I cannot?

They enforce a written room specification across every room, and they do not publish it. Marriott's own Franchise Disclosure Document confirms a system of "design criteria and specifications" that franchisees must meet, without disclosing the contents. No major brand publishes thermostat setpoints, blackout specifications, mattress specifications or acoustic specifications. The replicable part is not any particular number, it is having a one-page standard applied identically at every turnover, which costs nothing.

Key takeaways

• In a 2016 laboratory study, slow-wave activity dropped in the left hemisphere on a first night in a novel room and more than 80% of that night's arousals traced to it, with the asymmetry gone by a second session a week later.
• Nobody has ever run that experiment in a hotel or a rental, and the largest sleep study on the question, 124 adults, found no night-one difference in total sleep time or sleep efficiency, only in sleep architecture.
• The claim that more than half of people sleep worse on a first night has no primary source. A survey of 505 hotel guests found travel insomnia at 15.4% against 24.2% at home in the same sample.
• Light is the strongest evidence and the cheapest fix: about 100 lux of ordinary dim room light delivers half the melatonin suppression of 9,000 lux, and one night at 100 lux versus under 3 lux raised next-morning insulin resistance.
• The ideal sleeping temperature band is 20 to 25 degrees Celsius with a 5% to 10% efficiency penalty by 30 degrees, and the commonly quoted 29 to 32 degree figures describe the bed microclimate under the covers rather than room air.
• A two standard deviation rating improvement raised expected occupancy from 56% to 63% across 16,144 London properties with price controlled, which on six doors at a $240 average rate works out to roughly $36,792 a year.

To know whether a seven-point occupancy gain would even be visible on your doors, start from Airbnb occupancy rate benchmarks. And for the full picture of what a hotel is selling against you, read Airbnb vs hotel in 2026.

A seven-point occupancy swing is worth roughly $36,792 a year on six doors, and it arrives through reviews you have to be able to read by category. See review categories by door in MagicBNB →

About MagicBNB

MagicBNB is portfolio analytics for operators running 2 to 20 short-term rental doors. It connects your property management system and your bank accounts and reports what each door actually kept. The 5-Star Champ card keeps your reputation flagship visible, with its five-star count, the share of reviews at five stars and its average rating, so the playbook worth copying is identifiable rather than assumed. Today Pulse merges every booking, check-in, check-out and cancellation happening today into one timeline-ordered feed, which is what turns a turnover checklist into something that actually gets done on the right day. Milo, the built-in analyst, carries a glossary of more than 60 metrics and injects the right definition when it detects one in your question, so RevPAN means the same thing in March as it does in September. MagicBNB does not make beds. It tells you which bed is costing you a review.