Sentiment Drift: The Competitor Metric Nobody's Tracking

Key takeaways

• A star rating is a snapshot. Guest sentiment is a trend line.
• A long-window average moves slowly, so a sliding theme can hide for months.
• Not every change is drift. When the sample changes, a benchmark moves without any guest changing their mind.
• Tracking drift takes four things: a theme, a window, a minimum volume and a baseline.
• Public reviews show perception, not cause.

Every competitor analysis you've read is a photograph

Most hotel competitor analysis works the same way. You pull a rival's rating, compare it with yours, maybe break it into themes, and conclude that they're stronger on breakfast and weaker on maintenance. It's useful, and it's also a photograph.

A photograph can't tell you whether that rival is getting better, getting worse or standing still. Yet that is often the question that matters. A weakness that is closing is a very different competitive fact from one that is widening.

That gap is what sentiment drift measures, and almost nobody tracks it.

Why a long-window average hides direction

A hotel's overall rating is an average across every review the platform has counted, often over several years. That makes it slow to react by design, and this is simple arithmetic rather than a criticism of any platform.

Here is a hypothetical example. A hotel has 1,000 reviews averaging 4.2. Then a run of 30 recent guests, all unhappy about the same problem, leave an average of 3.0. The overall rating moves from 4.20 to roughly 4.17. Almost nothing on the surface changes, but a view that isolated those 30 reviews by theme would show something close to a collapse.

This is the same reason we keep saying that a strong average isn't a finding, it's a hiding place. Averaged across a long window, a real and recent problem can sit unnoticed for months. Anyone benchmarking a competitor by star rating alone has the same blind spot, in both directions. They can miss a rival that is slipping, and they can miss one that is quietly fixing its weak spot.

What sentiment drift actually is

Sentiment drift is the change in a theme's positive share over a fixed window, compared with a baseline.

That definition has four parts, and leaving any of them out makes the number unreliable.

Theme. Drift is measured at theme level (Maintenance & Upkeep, Breakfast Quality, Value for Money), not for the hotel as a whole. A rating can hold steady while one theme falls apart underneath it.

Window. The fixed period you measure over. A quarter is a workable default: long enough to collect meaningful volume, short enough to catch a turn.

Minimum volume. The number of reviews you need behind a figure before you trust it. A common working minimum is around 30 reviews per hotel and theme. Below that, the number should be labeled preliminary, not treated as a finding.

Baseline. What you compare the window against: the hotel's own prior window, or the wider market over the same period.

Two kinds of change: real drift versus the sample moving

This is the part most benchmarking gets wrong, so we'll use real data to show it.

In our Lake Geneva dataset, we tracked Food & Beverage sentiment across the Montreux hotels. With five hotels in the dataset, the review-weighted positive share for that theme was 75.9%, across 708 reviews. We then added a sixth Montreux hotel. The same measure became 78.4%, across 887 reviews. That is a move of about 2.5 points.

No guest changed their mind. Nobody's dinner got better. The benchmark moved because the sample changed: we added a hotel with strong F&B feedback, and the average moved with it.

This is composition change, and it produces false trends. A tracker that can't separate it from real drift will report improvement or decline that never happened. Two more confounders are worth naming.

Seasonality. Breakfast complaints cluster in busy summer months, noise complaints in different ones. A hotel's summer-quarter figure isn't comparable with its winter-quarter figure.

Volume shifts. A quarter with far fewer reviews is a noisier quarter. A swing there may be nothing more than small numbers.

The rule that follows is simple. Before you call something drift, check that the comparison set stayed the same, the seasons are comparable and the volume is comparable.

Why a rival's drift matters more than its snapshot

A snapshot tells you where to look. Drift tells you when.

Take a theme we see across the region. In the hotels where guests discuss Maintenance & Upkeep heavily, positive sentiment is consistently weak in all four Lake Geneva towns we've analyzed, roughly 18% to 33% depending on the town. As a snapshot, that says everyone struggles with maintenance, so it tells you nothing about who is pulling ahead.

Drift changes the reading.

A weak theme that is shrinking means a rival is fixing something. The opening you saw last year is closing.

A weak theme that is growing means an opening is forming, and it may be worth positioning against.

A strong theme that is eroding means a rival's advantage is fading, whatever the overall rating still says.

This matters most around renovation and repositioning. A rival mid-refurbishment can look weak in a snapshot and fixed in the trend, because recent guests are already reporting a different experience than the long-window average shows. If you decide where to compete based only on the snapshot, you may be planning against a competitor that no longer exists. For how to read a competitor's weak themes at a single point in time, see. For how those themes should inform a renovation decision, see.

How to track sentiment drift yourself

You don't need special software to start. You need discipline about the method.

1. Pick 3-4 themes that matter in your market. Not all nine. Fewer themes make it easier to see a pattern.

2. Fix a window and stick to it. A quarter works well. Log each theme's positive share for each window.

3. Record a figure only with enough volume behind it. Around 30 reviews per hotel and theme is a workable minimum. Mark anything below that as preliminary instead of dropping it in silently.

4. Keep the comparison set constant. Don't add or drop rivals partway through a series. If you must change the set, note it and restart the baseline, or you'll report the composition effect as a trend.

5. Compare against the market average. If your rival's breakfast score falls in Q3 and the whole market's does too, that's seasonality, not drift.

Done consistently for a few quarters, this gives you something a single benchmark never can: a direction.

What a single snapshot can't tell you

We'll be direct about our own data. The Lake Geneva figures in this article, and across Guest Experience Insights, are a single snapshot. They support comparison between hotels and towns at one point in time, and that is why we use them here to illustrate the measurement problem rather than to claim a trend in any hotel.

Measuring drift properly needs the review dates and repeated measurement over time. That is a core capability of SENTEEZ: we track reviews with their dates and follow theme-level sentiment across periods, so the same four-part method (theme, window, minimum volume, baseline) can be applied to a hotel and to its competitive set.

One limit applies either way. SENTEEZ works only with publicly available reviews, so what it shows is guest perception. A theme sliding at a hotel tells you that guests are describing that theme less favorably. It doesn't tell you why, and it doesn't prove what is happening inside the property.

Read your own drift first

Your own profile drifts too, and a theme sliding at your hotel is a bigger problem than one sliding at a rival. Most operators can recite their current rating. Far fewer can say which theme has weakened over the last two quarters, or whether last year's fix actually held.

Tracking your own trend before your competitors' gives you the baseline the whole method depends on. It also shows you the drift a guest would notice before you do. That is the logic behind an Owner Report Card: a hotel's own theme-level picture, kept up to date, rather than a rating checked occasionally.

Where to go next

If you want to see how theme-level comparison works on real data, the Guest Experience Insights platform covers 53 Lake Geneva hotels across four towns, free to browse:

The Guest Experience Insights platform covers 53 Lake Geneva hotels across four towns.