How to analyse TripAdvisor reviews and find what to fix

TripAdvisor is often the first thing a traveller reads about a hotel, restaurant or tour, and the last thing the business reads carefully. Most teams skim the newest reviews, reply to the angry ones and watch the bubble rating. That tells you how you are doing. It does not tell you what to change.

This guide sets out a method for reading TripAdvisor reviews as evidence: what guests talk about, how they feel about each thing, and which of those things is getting better or worse.

Why the bubble rating is not enough

The overall rating is an average of averages. A property can hold a steady 4.3 for a year while breakfast quietly gets worse and the new reception team gets steadily better: the two cancel out, and the number never moves.

Three things the rating hides:

• Mixed reviews. "The room was spotless but check-in took forty minutes" is one four-bubble review carrying one strong positive and one strong negative. Averaged into a rating, both disappear.
• Small but growing problems. A complaint that appears in two reviews a month is invisible in the rating until it appears in ten.
• Who is talking. Families, business travellers and couples rate different things. A change that delights one group can annoy another.

The fix is to read the text, not the score, and to read it in a way you can count.

Step 1: Collect every review in one place

Start with the full history, not just the last page. You want at least a few months so you can see change over time, and ideally the same period from the year before so seasonal patterns don't fool you.

For each review keep the date of the stay and of the review, the rating, the title, the full text, the trip type if the guest gave one, and the location if you run more than one property. Your management replies are worth keeping too, so you can later check whether replied-to problems stopped recurring.

If you also collect post-stay surveys or Google reviews, put them in the same place. Guests who write on TripAdvisor are a self-selected group; your own survey replies give you the quieter majority.

Step 2: Group reviews into topics

A topic is a thing guests talk about: breakfast, check-in, room cleanliness, noise, Wi-Fi, staff friendliness, location, value for money.

Two rules make topics useful:

1. One review, many topics. A single review usually mentions three to six things. Tag every mention, not just the main complaint.
2. Topics nest. Breakfast is useful; breakfast → coffee and breakfast → opening hours tell you what to do. Start broad and split a topic when it gets big enough that you can't act on it as one thing.

Doing this by hand works for a few dozen reviews. Past a few hundred it becomes a part-time job, and two people will tag the same review differently. This is where software earns its place: modern language models file mentions into topics consistently and let you rename, merge or split topics as your picture sharpens.

Step 3: Score the tone of each mention, not each review

Attach a tone (positive, negative, or mixed) to each topic mention rather than to the review as a whole. In the check-in example above, room cleanliness is positive and check-in is negative, inside the same four-bubble review.

Once every mention has a tone, you can answer the questions that matter:

Step 4: Watch share of conversation, not raw counts

If you received twice as many reviews in August as in March, every topic will look like it doubled. Raw counts mislead whenever volume changes, and on TripAdvisor volume follows the season.

Use share of conversation instead: the percentage of reviews in a period that mention the topic. If noise goes from 4% of reviews to 11% of reviews, something changed, however many reviews came in. Pair that with the topic's tone: a topic can grow because guests are newly delighted by it, as easily as because something broke.

Step 5: Decide what to fix first

You now have a list of topics, each with a share of conversation, a tone and a direction of travel. Rank the negative ones by three questions:

1. How many guests does it touch? Share of conversation.
2. How much does it hurt? How often it appears in low-rated reviews.
3. Is it getting worse? Its change against the previous period.

A topic that is common, painful and growing goes to the top. A topic that is rare and stable can wait, even if the reviews that mention it are furious.

Then read ten actual reviews for each of your top three topics. The numbers tell you where to look. The reviews tell you what is actually happening at the front desk at 3pm on a Sunday.

Step 6: Close the loop

After you change something, keep watching the topic. If you shortened the check-in queue in May, the share of negative check-in mentions should fall over the next few weeks. If it doesn't, the fix didn't work, or the problem was something else.

Replying to reviews matters too. A short, specific reply that names the fix ("we've added a second person at reception from 2 to 6pm") shows future guests that someone reads them. Our guide to responding to negative reviews covers how.

Doing this with Userforms

Userforms runs this method for you. Connect your TripAdvisor listing, alongside Google, Yelp, Trustpilot and your own feedback forms, and every review is read as it arrives: filed into a topic hierarchy, each mention given its tone, and the whole library compared with the period before.

The dashboard's What changed section shows which topics grew or shrank in share and which turned better or worse, and every chart opens onto the reviews behind it. If you run several properties, you can compare them side by side; see comparing feedback across locations.

You can start for free, or see what each plan includes on the pricing page.