Review Score to RevPAR: The Research Behind the Number

“There’s a Cornell study that shows reviews drive revenue.”

Someone at the table says it, and heads nod. It sounds settled.

Then comes the better question: which study, and what exactly did it measure?

That question is worth asking about any number that travels through a meeting. The research on review scores and revenue is solid enough to take seriously. It’s also specific enough that it gets misquoted all the time.

Where does the review-score-to-revenue link come from?

Mostly from work at Cornell’s Center for Hospitality Research, led by Chris Anderson.

The first major report matched a hotel guest review index with hotel performance data from an industry benchmarking provider. That let the researcher compare how a hotel’s online reputation related to its actual rate, occupancy and revenue per available room (RevPAR), instead of relying on opinion.

The same report also looked at the traveler side. It found that the share of consumers reading reviews before booking a hotel had been rising over time, as had the number of reviews they read.

A later paper by Anderson and a colleague, published in Service Science in 2014, used the same kind of data to test the relationship again and to ask whether it holds equally across different types of hotel.

The research is real. The useful question is what it measured.

What did the research find?

Higher online reputation scores were associated with stronger hotel performance.

That means higher rate, higher occupancy and higher RevPAR. Not rate alone, and that distinction matters. A good reputation doesn’t only let a hotel charge more. It also helps the hotel win more bookings. Guests who read reviews before booking are choosing, not just comparing prices.

The findings were reported as effects of “up to” a certain size. That wording is deliberate, and we’ll come back to it.

We use the word showed here, never “proved.” These are statistical relationships found in historical data across many hotels. They’re strong evidence that reputation and revenue move together. They’re not a guarantee that any one hotel will see the same result.

What did it actually measure?

Three things are worth separating.

• The score. The studies used a review index built from guest reviews across sites. That’s a composite score on a defined scale. It’s not the same thing as “the percentage of mentions that are positive” for a single theme.
• The outcomes. Rate, occupancy and RevPAR, taken from hotel performance data held by a benchmarking provider. These are real operating results, not survey answers.
• The comparison. How changes in the score related to changes in those outcomes across many hotels over time.

Here’s how those translate into what you can and can’t say:

The middle row is the one that gets lost. An average across many hotels is a useful guide. It’s not a forecast for a specific one.

Does it work the same for every hotel?

No, and the researchers say so.

The 2014 paper found that the influence of review scores differs by hotel class. As the class level goes up, the influence appears to go down. In plain terms, an upscale hotel is less sensitive to its review score than a mid-scale one.

That’s intuitive. A traveler choosing a budget or mid-scale hotel leans heavily on reviews. A traveler choosing an upscale or luxury property may be relying more on brand, reputation and past experience.

It also means different studies can give different figures. A later estimate came out lower than the first. Neither contradicts the other. They used different approaches and showed that the answer depends on the type of hotel.

Two careful studies, two different numbers. That’s what a range looks like.

So if someone gives you one confident percentage for “what a better score is worth,” the right response is to ask which study, which hotel class and which score scale. If they can’t answer, the number is borrowed authority.

Why does “up to” matter so much?

Because it changes what kind of claim you’re making.

“Up to” describes the upper end of what the research found. It’s a ceiling to plan against, not a prediction to budget on. Averages across many hotels also hide variation. Some hotels respond more, some less, and the research can’t tell you which one yours is.

There’s one more limit worth stating plainly. These studies describe relationships in observed data. They don’t tell you that fixing a particular problem at your hotel will produce a particular result. That needs your own baseline and your own re-check afterward.

How does SENTEEZ use the research?

As the bridge between guest sentiment and revenue, and nothing more.

The research supplies the link between reputation and revenue. Guest feedback supplies the size of the gap: we measure each recurring complaint theme against a healthy benchmark and weight it by how often guests mention it. Those weighted gaps add up to a hotel-level score. That score is then translated into a revenue range using published hospitality research, and split across four broad pillars, never by single theme.

For our reference property, an anonymized 156-room hotel in the Lake Geneva region, the result is an annual revenue opportunity of CHF 1.13M to 1.25M. The reviews are real. The operating inputs, such as room count, rate and occupancy, are illustrative.

(The step-by-step translation is in From Sentiment Score to P&L: How a Review Theme Becomes a Revenue Figure, and the scoring is in How the Hotel Revenue Opportunity Score (ROS) Works.)

We’re candid about the limits here too.

• The total is a ceiling. It assumes every gap closes at once, which no property does.
• The research describes average behavior, and our estimate inherits that.
• Translating theme-level sentiment into the kind of score the research used is an approximation.

Used honestly, that’s still useful. It sizes the prize and points to where to look. It doesn’t forecast your year. (More on that distinction in Opportunity Isn’t Attribution: Why a Revenue Estimate Isn’t a Forecast.

On pricing a fix, Why We Won’t Put a Price on a Fix From Review Data explains our reasoning.

Knowing that boundary is what keeps a good study from becoming a misused one.

Before your next owners’ meeting

If a study is being quoted, ask four things.

1. Which study is this figure from, and who ran it?
2. What score did it use, and is it the same kind of score we’re looking at?
3. Does the research cover hotels like ours, in class and market?
4. How will we measure our own result against a baseline?

If nobody knows the answer to the first question, the figure isn’t ready to carry a decision.

The bottom line

The research tells you the link is real. Only your own numbers tell you how big it is for you.
A good estimate uses the research honestly: as a guide to the size of the opportunity, with its limits stated. It doesn’t ask the research to promise the outcome.