MarketingMarketing in Travel & HospitalityTravel & Hospitality

A guest who returns once a decade is still worth modelling at full five-figure value

Standard LTV formulas break when your best customers disappear for three years between stays. This article shows how to rebuild the calculation so it reflects what a hotel or resort guest is genuinely worth over a lifetime of irregular, high-value visits.

Ada BrandtBrand & Marketing StrategistSeptember 26, 2026

Listen to the podcast

4 min

Chapters

Key takeaways

Segment irregular high spenders separately before calculating LTV, since blended averages with one-time Groupon guests produce a number that describes nobody.
Model expected visits over a realistic horizon such as 20 years using the segment's real return interval, not a monthly frequency average.
Pull your top 100 spenders, calculate the average gap between their stays, and remove anyone still inside that window from the churn list.
Treat HubSpot's retention cost multiples of 5 to 25 and SEMrush acquisition figures as vendor starting points and cross check against Forrester before quoting them to a board.
Keep contact during quiet years with birthday messages or renovation news rather than treating a non-booking guest as a rejection.

Read the full transcript

Host:This is Leader's Insights. On the table, a guest who returns once a decade is still worth modeling at full five-figure value. If a guest hasn't booked in three years, why is she still on your books as valuable and not written off as dead weight?

Expert:Picture the actual moment. It's March, I'm sitting with the revenue team at a resort group in the Maldives, and someone pulls up a guest who last stayed in 2023. The system has flagged her churned. Churned meaning the software has decided she's gone. She'd spent $41,000 across two visits, and the algorithm wanted to stop marketing to her to save the postage. And you said? I said congratulations. You've just fired your best customer because she didn't behave like she buys coffee. That's the trap, then. The math built for people who buy every week. Exactly. So let's build it from the floor up, because this is where most hotel marketers quietly get it wrong. Lifetime value. LTV. What a customer is worth to you across the whole relationship. The standard formula assumes rhythm. Average purchase. Times how often they buy. Times how many years they stick around. And for a coffee subscription, that's fine. Beautiful. Buys monthly. You can predict it. But a resort guest isn't monthly. She's a comet. She swings past every three or four years, spends a fortune, then vanishes into the dark. If you measure her by frequency, the formula sees long gaps and reads abandoned. It can't tell the difference between someone who left and someone who's simply between orbits. So the fix is what? Just be more patient? Patience isn't a model. The fix is to stop measuring frequency and start measuring the pattern of the gaps. You look at your genuinely high value guests and you ask, what's the normal interval between their stays? If your anniversary couples come back every four years like clockwork, then year three isn't churn. Year three is Tuesday. How do you actually calculate that without kidding yourself? You segment first. Separate the comets from the coffee drinkers, the irregular high spenders from the once and done crowd. For the comets, you model expected visits over a realistic horizon, say 20 years, using their real return interval, not a monthly average. Forester's work on customer valuation makes this point hard. Blended averages destroy value in businesses with lumpy purchasing. When you average a comet with a tourist who came once for a Groupon, you get a number

Host:that describes nobody. Give me the difference in dollars. Make it hurt.

Expert:Same guest, two methods. Standard frequency model saw her going quiet, projected maybe one more visit, valued her around 8,000, rebuild it on a four year interval across a 20 year relationship, and she models at 52,000. That's the swing. That's whether you send her a handwritten note from the

Host:general manager or nothing at all. You mentioned software flags. Her is churned. Who's software and should I trust the churn number it spits out? HubSpot's

Expert:tooling will happily calculate churn and l T V for you, and it's fine. But remember, they sell the customer relationship software, so their default settings assume the frequent buyer world their product was built for. HubSpot has put retention cost multiples in the 5 to 25 range depending on the study, though worth noting they're a vendor. So cross check against Forester before you quote it in a board deck. Same with SEMrush on acquisition costs for travel keywords useful, but they sell the analytics tools. So treat their numbers as a starting point, not scripture. What's the mistake even smart operators make here? They mistake silence for rejection. A guest not booking isn't a guest saying no. She's a guest whose life hasn't lined up with another trip yet marketing to her in the quiet years. A birthday message, a note when a sweet she loved is renovated. That's not waste. That's keeping the runway lit for a plane that lands

Host:every four years. So if a marketer does one thing Monday morning, what

Expert:is it? Pull your top 100 spenders. Calculate the real average gap between their stays. Anyone inside that gap window. Take them off your turn list today. You're not chasing the dead. You're just standing at the door for people who are always going to come back. What we read for this one,

Host:Forester research, SEMrush, vendor, SEO analytics tools, DigiDay, HubSpot, vendor, CRM marketing automation. We'll leave it there. The CMO self assessment at MBA dash training.com will tell you where you really stand.

Guest lifetime value sounds straightforward until you apply it to a property where the average repeat interval is 26 months, the top-spending guests visit once every four to seven years, and a meaningful share of your database has been silent for a decade. At that point, the standard formula, average order value multiplied by purchase frequency multiplied by customer lifespan, produces numbers so low they justify doing almost nothing to retain anyone. That is the wrong conclusion, and it comes from applying a retail-era model to a business whose economics work nothing like retail.

Why perishable rooms break the standard LTV formula

A night's accommodation is one of the most perishable products in any industry. Unlike a handbag or a software subscription, an unsold room generates zero revenue the moment midnight passes, which means acquisition cost is always measured against a narrow window of potential recovery. This creates a margin structure that punishes short-termism: you spend heavily to fill beds tonight, but the genuine return on that spend arrives across a relationship that may span two decades and fewer than a dozen stays.

Marriott's Bonvoy database contains roughly 210 million enrolled members as of 2026. The majority of those members have not stayed in over 12 months. Under a conventional RFM (recency, frequency, monetary) framework, most of them look worthless. Under a correctly built LTV model for the travel sector, a meaningful proportion of those dormant members represent confirmed future demand: people who have already demonstrated willingness to pay, are geographically mobile, and whose next trip simply has not happened yet.

The same logic applies at boutique level. A single guest who stays at a Maldives overwater villa every three years, brings a partner, adds excursions and spa, and refers one colleague booking a corporate retreat may generate 40,000 to 60,000 euros in total net revenue across a 15-year relationship. That number never appears in any 12-month performance report, which is exactly why it needs a dedicated model.

How do you calculate guest LTV with multi-year gaps?

The standard formula needs three adjustments for long-interval hospitality.

Survival probability replaces frequency. Rather than dividing total stays by years on file, you model the probability that a guest returns at all within each rolling 36-month window. Hilton's analytics teams, like most major chains, use Pareto/NBD (negative binomial distribution) models borrowed from direct marketing to estimate individual-level return probability. For a property without that statistical infrastructure, a simpler segmented approach works: classify guests by their longest observed inter-stay gap, then apply historical return rates for each segment. A guest who has returned twice after gaps of more than 18 months has already proved multi-year loyalty; their survival probability for a third return is materially higher than a first-time guest.

Average booking value is adjusted for category drift. Guests often trade up over time. A couple who honeymooned in a standard room returns for a 10th anniversary in a suite and adds a champagne package. Ignoring this trajectory understates LTV. Use cohort data: segment guests by their first-stay room category and map average spend at second and third stay. The upgrade curve is a real and quantifiable asset.

Discount rate selection is different from e-commerce. A standard LTV calculation discounts future cash flows at 8-10% annually to reflect time value of money. With multi-year gaps, a guest whose next stay is probabilistically three years away has their future value discounted heavily. Use a lower discount rate for verified loyal segments (where the stay will happen, the question is only when) and apply the standard rate only to lower-confidence segments. This distinction alone can shift a dormant guest's calculated LTV by 30% or more, which has direct implications for how much you spend on re-engagement.

A concrete example: a guest at a Four Seasons property stayed in 2019 and again in 2022. Their inter-stay gap is 36 months. Historical cohort data for 36-month returners shows a 58% probability of a third stay within four years. Average second-stay revenue for that guest type is 4,200 euros. Applying a 6% discount rate over four years, the expected present value of that third stay is roughly 2,070 euros. Add referral value (conservatively 15% of one additional booking from this guest's social network) and their remaining LTV contribution is around 2,500 euros. That number justifies a re-engagement spend well above most email marketing budgets would allocate to a "dormant" contact.

Understanding the truecost of acquiring a guest in the first place makes these LTV numbers land harder: when a single booking through a major OTA costs 18-22% in commission, the imperative to model and defend long-term guest value becomes a budget conversation, not a theoretical one.

Where hospitality LTV models hold up and where they fail

This approach is most reliable for properties with three or more years of clean stay history per guest, consistent product (a resort that has not changed category), and a customer base with stable demographics. It works well for luxury independents, branded urban hotels with corporate accounts, and destination resorts where the guest journey is a considered annual or biennial decision.

It breaks down in two situations. First, transient city hotels where the majority of guests are travelling for reasons that disappear (a conference venue that moves, a job that ends, a relative who relocates) have survival probabilities that are structurally lower and harder to estimate from past behaviour. Second, any property that has undergone a significant repositioning or price change will have historical cohort data that no longer predicts future behaviour.

The model also has nothing to say about guests acquired through aloyalty program that drives repeat stays at the program level rather than the property level: a Bonvoy member who is loyal to Marriott as a brand but distributes stays across 40 different properties is measured by the chain, not by any individual general manager. CMOs working at brand rather than property level need to run this calculation across the portfolio, not per asset.

One further honest caveat: LTV models of this kind require data governance that many hospitality operators still do not have. GDPR and its national equivalents across the EU, as well as the evolving privacy landscape in the US and Southeast Asia, constrain how long guest data can be retained and how it can be used for profiling. A model built on a database that your legal team is about to trim is a model built on sand.

The practical output of this work is not a single number for a board presentation. It is a tiered re-engagement budget: how much per guest segment is defensible to spend on win-back, based on their calculated remaining value. Most hospitality CMOs who run this exercise find that they are underinvesting in guests they had written off and overinvesting in broad acquisition that brings in low-probability returners.

The full course on this sector:Marketing in Travel & Hospitality.

Frequently asked questions

Is it worth spending money on a guest who has not stayed in years?

Often yes. A Four Seasons guest with a 36-month inter-stay gap sits in a cohort with a 58% probability of a third stay within four years, giving a remaining LTV of roughly 2,500 euros once referral value is added. That justifies a re-engagement spend far above what most email budgets allocate to a dormant contact.

What discount rate should a hotel use in its guest LTV model?

Use two rates rather than one. The standard 8-10% annual discount applies to lower-confidence segments, while verified loyal guests (where the stay will happen and only the timing is uncertain) warrant a lower rate, around 6%. That single distinction can move a dormant guest's calculated LTV by 30% or more.

How do OTA commissions change the guest lifetime value conversation?

They turn it into a budget argument. A booking through a major OTA costs 18-22% in commission, so every acquisition repeated through that channel erodes the margin a long guest relationship is supposed to produce. Modelling and defending long-term guest value then competes directly with commission spend.

Does guest LTV modelling work for loyalty program members?

Not at property level. A Bonvoy member loyal to Marriott as a brand may spread stays across 40 different properties, so their value is captured by the chain, not by any individual general manager. CMOs working at brand level need to run the calculation across the portfolio rather than per asset.

Go deeper

The lessons that take this article further, free to read.

Sources

Finished reading?

Validate your read to earn XP and feed your radar.

Marketing articles Back to Blog