Demo hotel. The Rivenhall and its competitors are invented, and so is every number and quote. This is the real product.About VerbatimSee it on your hotel
The Rivenhall
data through 6 Nov 2026
Report ↓

How this works

Everything in this portal is built from public guest reviews — processed, scored and benchmarked the same way for every hotel in your market.

1
We read every public guest review

For The Rivenhall, that is 5,812 verified guest reviews across Booking.com (2,270) · Google (1,560) · TripAdvisor (1,320) · Agoda (662), refreshed on an ongoing basis. Nothing is sampled — every review is processed.

2
AI breaks each review into specific observations

A review like "beautiful pool but check-in took forever" becomes two separate observations — one positive about the pool, one negative about check-in — each tagged to a topic taxonomy covering rooms, dining, staff, facilities and more. 24,130 observations underpin what you see here.

3
Observations become a sentiment score

Each observation carries a sentiment strength and a confidence weight. Your sentiment score (out of 10) is the confidence-weighted balance of praise and complaint in a period: 10 is all praise, 0 all complaint, 5 an even balance. One furious outlier can't swing it, but a pattern will.

4
Sentiment is tied back to the ratings guests gave

Every review also carries the guest's own star rating. We measure, on your own reviews, how much lower reviews that complain about each topic rate you. That turns a complaint into rating points, and rating points into room revenue using published hospitality research.

5
Every hotel is measured identically

Your competitive set is processed by exactly the same pipeline, the same taxonomy, and the same scoring — same sources, same rules. When you see a gap versus a competitor, it is an apples-to-apples gap.

6
Numbers always come with evidence

Behind every score movement are the guests' own words. We surface representative quotes throughout, so you can verify any number against what guests actually wrote.

7
Alerts favour precision over noise

We only flag a topic when its recent sentiment moves well outside its own historical range with sufficient volume behind it. If this portal raises an alert, it is worth your attention.

Fair questions, straight answers

Why is my sentiment score different from my Booking.com score?

They measure different things. The guest rating (8.4 out of 10 over the last 90 days) is the average of the stars guests gave, the number closest to what booking sites show. The sentiment score (6.5) reads what they wrote, where 5 is an even balance of praise and complaint. Guests who give 8 out of 10 often still write about something that went wrong, so the sentiment score usually sits below the rating. What matters is that the two move together, which the Trends page shows.

How do you put a dollar figure on a complaint?

On your 4,960 rated reviews since 8 Nov 2023, we measure how many rating points reviews that complain about each topic give up, with every topic measured together so nothing is counted twice. Multiplied by how often guests complain, that is the topic's cost in points off your average. Published research (Anderson (Cornell CHR, 2012); Ye, Law and Gu (2009)) links each rating point to 4–8% of annual room revenue, which gives the dollar range. These are associations, not guarantees: fixing a problem removes the complaint, and the rating follows over the months after.

How sure are these numbers?

We test the model on reviews it has not seen. Fitted only on reviews up to 8 Aug 2026, it predicted that the 429 reviews since would average 8.5 out of 10, from what those guests wrote; they actually gave 8.4. Last year's average alone would have said 8.6. Dollar figures are shown as a range because the research behind them is a range.

What does the AI do, and what do you calculate?

An AI model (Claude Opus 5.5) reads your reviews: every complaint review behind your costliest complaints, the complaints behind an alert, and the last month's reviews. It names the causes guests describe and files each review under them; we count the filings. It then drafts the actions, the “biggest problem” grouping, the month's explanation and the reply drafts, and we check that every number it writes comes from what it was given and that every quote is copied word for word. Patterns such as weekend against weekday stays are measured directly on your reviews, every dollar figure is calculated by us, never written by the model, and we never estimate what an action would cost you. Reply drafts are only drafts: nothing is posted for you.

Is it fair to compare us against different hotels?

The comparison is relative, not absolute: identical sources, identical processing. A hotel with more reviews doesn't score higher — volume affects confidence, not the score itself.

Can competitors see our data?

Other hotels see only aggregate scores that are derived from public reviews — exactly what you see about them. Your access link, your intervention log and your detail views are private to you.

How current is this?

Data through 6 Nov 2026. Reviews are collected and re-scored on a weekly cycle. If a refresh is missed for more than 14 days, every page says so at the top.

What about fake or malicious reviews?

We only ingest platform-verified reviews, weight by confidence, and require volume before flagging anything — a single review, genuine or not, cannot move your numbers meaningfully.

Questions about methodology? Write to info@verbatimfeedback.com.