Hotel Mystery Shopping: Secret Shop Your Resort in Two Hours

Hotel mystery shopping doesn't need a firm. To test your own resort's guest response, have someone your team won't recognize send one four-question inquiry by email, web form, text, website chat and phone: at 2pm on a weekday, in the hour after check-in time, and after 10pm. Then score each contact 72 hours later. That's eight contacts over three days and about two hours of work. It's a test a general manager (GM) runs on their own property, and it shows what a guest actually gets.

If you have traveled with small kids, you know the 9pm text. The crib isn't in the room, or the flight is late and you need someone to know you're still coming. You send it, and then you watch the screen. Nobody at the property ever sees that part. A mystery shop is how you see it.

The Guest Response Gap, the first ResortConcierge AI Research report, gathers the published evidence on what that silence costs. The headline: in J.D. Power's 2025 study of 39,219 branded hotel guests, satisfaction fell 217 points on a 1,000 point scale when a problem came up during the stay, and 12% of stays had one ^[1]. The study covers branded hotels, not independents, and doesn't separate the problem from the wait for an answer. Anyone who has sent the 9pm text has a guess about which part stings.

What is hotel mystery shopping, and can a GM run it alone?

Hotel mystery shopping tests service from the guest's side: someone poses as an ordinary guest, makes contact and writes down what happens. Professional programs use outside evaluators who vary the days, times and shifts they test, as SIS International Research, a research vendor, described in 2025 ^[2]. Most are built around stays and phone calls.

A full stay audit is a job for a firm. The response layer isn't. Does anyone answer, how fast, and with what next step? A GM can test that alone, cheaply, every month, with a borrowed phone and an email address the desk has never seen.

Two limits, up front. A self-test is a small sample, not a census: eight contacts tell you where to look, not how every night goes. And it measures the system, not a person. If the 10:30pm text goes unanswered, the finding is that nobody owns texts at 10:30pm.

If you're mapping tools to channels, start with how guest communication software fits together at a resort.

What did published secret shops find?

Published shops and call data agree on one uncomfortable thing: a large share of guest contacts get no answer, or a passive one.

In Revinate's 2026 secret shop of 135 luxury and upscale hotels, 26% of after-hours reservation calls dropped straight into a standard voicemail box [3]^[4]. Across the 308 calls, 56% included no discovery questions and only 22% included a direct ask for the reservation ^[3]. When the front desk answered, 74% of those calls had no discovery questions, and fewer than 2 in 100 asked for the booking ^[3]. That pair matters most at an independent resort, where the front desk often is the reservations team.

Invoca's 2026 platform data says 64% of calls to travel businesses (hotels and resorts, plus cruise lines) are answered by a person ^[5]. Email has been tested too. In 2023, Stampede sent a mystery email to 250 hospitality venues across the UK and Ireland: 22% replied within the first hour, and a third never replied at all [6]^[7]. Those were hospitality venues, not specifically hotels.

All of it is vendor data, none of it from independent resorts. Read it as a direction, not a benchmark. For the money side, see what slow replies cost a resort.

When should you send each test contact?

Timing matters more than volume. Three slots test three different desks.

The weekday afternoon is your baseline. The desk is staffed, arrivals haven't started, and anything that goes wrong at 2pm will go worse later. The second slot is the hour after check-in time on your busiest arrival day, when the person who answers the phone is also handing keys to a line of tired families. The third is after 10pm.

Night is where volume and urgency meet. Akia's July 2026 analysis of 1.4 million guest messages, vendor platform data, found about 4 in 10 arrived outside 9 to 5, property local time, and lockouts peak between 11pm and 2am ^[8]. The check-in slot rests on practice, not data. No published figure I found singles it out. It's simply the hour the desk has the least attention to spare.

The two-hour response check in the report uses six contacts, at 2pm and after 10pm. This version keeps those and adds the check-in hour, for eight in all.

If the night slot is where your results fall apart, covering the desk after hours without a second night hire walks through the options.

What should a mystery shop message say?

One message, the same on every channel, so the results compare. Ask four things in it: availability for two nights about three weeks out, whether a child can share the room or there's a crib, the rate, and how to book.

An illustrative version, for text: "Hi, we're looking at two nights about three weeks from now, two adults and a toddler. Can our little one share the room, or do you have a crib? What would the rate be, and how do we book?"

In a 2004 test published in 2006, Oliveira and colleagues sent the same four questions by email to 116 lodging properties in southern Brazil. Within 13 days, 66% replied, and only 23% of the replies, 18 of 77, answered all four ^[9]. It's a seminal study and two decades old, so don't compare your numbers to it. Borrow the design. Counting answered questions gives you a score that goes past "did they reply."

How do you mystery shop your own hotel without tipping off the team?

Use a friend, or a manager from outside the property, with a personal number and an email address your team has never seen. Your own phone is out; someone at the desk has it saved.

Tell your staff a response test will happen this quarter, without saying when. SIS recommends telling staff the program exists ^[2]. Most people can live with being checked. What they can't live with is feeling set up. A public rule with private timing keeps it a test, not a trap.

The shopper's checklist:

1. Use a personal number and email the team won't recognize.
2. Send the same four questions everywhere.
3. Never book or hold a room.
4. Don't record calls.
5. Close the thread politely.

On recording, Kilpatrick Townsend's 2024 summary of US wiretap laws lists 13 states that require the consent of everyone on a call: California, Connecticut, Delaware, Florida, Illinois, Maryland, Massachusetts, Montana, Nevada, New Hampshire, Oregon, Pennsylvania and Washington ^[10]. The other 37 states and the District of Columbia require one party's consent ^[10]. Notes are enough for this test. This is general information, not legal advice.

If you test a competitor too, thank them and close the thread, so nobody spends an evening chasing a booking that isn't coming.

How do you score a hotel mystery shop?

Score each contact 72 hours after you send it, as meets, close or gap against the clearest published line for its channel. Anything between the two lines is close. The bands follow the report's self-check.

No one body sets a standard for every channel, so each line is borrowed. AAA's lodging guidelines, in an edition posted in 2019, expect the service phone extension answered within three rings and requests acknowledged with an expected fulfillment time ^[11]. That was written for the in-house extension, and here it's borrowed for the main line. For texts, 57% of 1,074 consumers in EZ Texting's 2025 vendor survey, across all industries and not hotel guests, expect businesses to respond within 15 minutes ^[12]. For email and web forms, Airbnb's help center counts responses after 24 hours as late ^[13]. For website chat, about half of consumers expect an instant reply, in a 2021 Leadferno vendor survey that is still the newest figure I found ^[14].

Time is only half the score. The write-down column matters as much: questions answered, a next step with a time, and on calls, whether anyone asked your dates or offered to book. Those last two are what most calls in the Revinate shop skipped ^[3].

Whether a text reply was any good is a separate question, and what a good text reply to a guest sounds like covers it.

What do you do with the results?

Fix the gaps in a set order: no reply at all, then silence after 10pm, then answers with no next step, then answers with no ask. The order follows reach. No reply ranks first because an unanswered problem stays a problem, and in J.D. Power's 2025 study a problem during the stay cost 217 points ^[1]. After-hours silence ranks second because it's common: in Revinate's 2026 shop, 26% of after-hours reservation calls hit a standard voicemail ^[4].

Read every gap as a channel nobody owns at that hour, not a person failing. Whoever ignores a 4pm text with a line at the desk is triaging the person in front of them. That's a design problem, and it needs a design fix.

If the miss landed in the check-in hour, look first at the 3pm handover where requests go missing.

Share the results with the team as a finding about the system, then move on to turning results into response targets your team can hit. Run the test again in 30 days with a new shopper, and once more before peak season.

Where ResortConcierge helps

Many night gaps are routine questions with a known answer: parking, pool hours, checkout time. ResortConcierge's AI concierge answers those by text where the channel is connected, from entries the property's own team writes and edits. The work is in a knowledge base your AI concierge answers from, and it's mostly writing down what the desk already knows.

Some messages shouldn't get an automatic answer at all: the lockout, the complaint, the guest who is already upset. A conversation that needs a person goes to ResortConcierge's escalation queue, which shows the guest, the room, the sentiment and the reason, so whoever picks it up doesn't start cold.

If texts and emails live in different places, ResortConcierge puts WhatsApp, SMS and email threads in one list for the team. It doesn't answer phone calls today; voice is planned, so the three call rows on your sheet stay a staffing question.

When is a do-it-yourself test not enough?

After one round, your shopper isn't anonymous. A good desk team remembers a voice. One run is one sample, and an owner may want a read from outside the building.

The report's two-hour check and PDF scoring log are free to download. For an outside read, you can request a private response check: over three days, an email, a web form, a text or WhatsApp message, a website chat and two calls, one of them after 10pm, with no booking made or held and no call recorded. You get a one-page readout against the published benchmarks and a 20 minute walkthrough, and the results go only to you.

Either way, start this week. Pick your three times, write the four questions, and find out what the 10:30pm guest gets. I'd rather a GM feel that wait once, on purpose, than read about it in a review.