Industries
Hotel reviews on your website: turning OTA scores into direct bookings
Hotel reviews on your website have exactly one job: to make the direct rate feel as safe as the OTA rate the guest was reading ninety seconds earlier. Pulling the figures from our live data on 25 September 2026, the 23 lodging businesses we host keep a typical 137 guest reviews on Booking.com or Airbnb and a typical 5 on Google, so for a property the OTA channel is the reputation and Maps is a rounding error. The twist worth knowing before you choose a widget is that those OTA scores are the harshest figures we hold anywhere in the system, and their harshness is precisely what wins the direct booking. Below are the figures, then the three decisions they settle.
How much guest feedback a property really keeps
Adding up every review we hold for businesses running a Booking.com or Airbnb profile gives 5,072 records over 23 lodging businesses, running from 13 February 2014 through to 24 September 2026. The range is unusually broad: 2 records at the bottom of the set, 1,216 at the top, a midpoint of 137 and an arithmetic average of 220. Build the page against the midpoint, since three of the 23 hold half of everything between them, at 1,216, 721 and 617 records.
None of these are chain hotels. The profile URLs we hold read as apartments, villas, cottages, guesthouses, a boardinghouse and a single recognisable chain property. Worth saying early, because a four-apartment operator and a 200-room hotel face the identical design question here with very different volumes of material to answer it.
The believability table, ordered by how imperfect each source is
Every platform a property is likely to plug in, ranked by the share of ratings that hit five stars, least flattering at the top. That ranking is the argument of this article: on a hotel page, the least flattering platform is the most persuasive one.
| Platform, all customers, 25 Sep 2026 | Reviews held | Five stars | 4 to 4.9 | 3 to 3.9 | Under 3 | Average |
|---|---|---|---|---|---|---|
| Booking.com | 3,276 | 52.4% | 40.3% | 5.6% | 1.7% | 4.58 |
| 391,212 | 85.8% | 7.4% | 2.1% | 4.7% | 4.71 | |
| Airbnb | 1,796 | 90.4% | 7.9% | 1.6% | 0.1% | 4.89 |
| TripAdvisor | 12,820 | 91.1% | 5.7% | 1.5% | 1.6% | 4.85 |
Booking.com is the only platform in our whole system where five stars are a minority. Nothing else we import gets near it, and the reason is procedural rather than cultural: Booking asks after check-out, it asks everybody, and it asks for a number out of ten instead of a sentiment. Ten-point scales generate sevens and eights. Five-star prompts generate fives.
A guest who reaches your site from an OTA has just spent ten minutes reading 8.6 and 9.2. Hand them a flawless run of five-star quotes and you have argued against the page they arrived from. Hand them the 4.58 they already saw, with the Booking.com mark on each card, and the direct rate stops looking like the gamble. Why perfect ratings read as invented is covered in our star distribution analysis; for rooms it is not a nuance, it is the entire case for embedding the OTA feed instead of a quote you wrote yourself.
The half-star problem nobody warns you about
Scoring out of ten means the map onto five stars produces fractions. In our data 1,027 of 3,276 Booking reviews, 31.3%, carry a fractional rating: 849 sit at 4.5, 134 at 3.5, 22 at 2.5, 10 at 1.5, 8 at 0.5, plus a handful at values like 4.8 and 4.4. Close to a third of a property’s evidence cannot be drawn with whole stars.
This constrains widget choice and it is easy to discover far too late. A widget that draws only whole stars has two options and both are bad: round 4.5 up and inflate every second guest, or round down and understate the property. Either way the figure on your site drifts from the figure on Booking, and the person comparing both tabs is the one who notices. Check half stars before you check anything else about a hotel widget.
TripAdvisor is not the hotel platform you assume
TripAdvisor carries the biggest pile of reviews in this article, 12,820 over 23 customers, and almost none of it belongs to hotels. Sorting our 26 TripAdvisor profiles by the kind of page each one points at:
| What our TripAdvisor profiles actually point at | Profiles |
|---|---|
| Attractions, tours and experiences | 17 |
| Restaurants | 5 |
| Hotels | 3 |
| Other (member profile pages) | 1 |
Seventeen of 26 are tours, walking guides, kayak hire, a zoo, an escape room. Three point at a hotel, and those three keep 42, 42 and 35 reviews. Two of them also run Booking.com, where the very same businesses keep 258 and 617. For those two, in the same period, the Booking profile carried six and eighteen times the evidence of the TripAdvisor profile respectively.
So: if you sell tours, meals or an attraction, TripAdvisor is your platform and the TripAdvisor widget guide is your page. If you sell rooms, starting with TripAdvisor is how you finish with a widget displaying thirty-five reviews while several hundred sit idle on Booking.
Google is not the default here, and that catches people out
In most trades we say start with Google. Rooms are the exception. Of the 23 lodging businesses, ten run no Google profile with us at all, and taken together the typical Google figure across all 23 is 5 reviews. On the 13 that do run Google, the OTA side is the bigger of the two for 7 of them, so it is nearer a coin flip than a landslide.
The failure mode is the mirror image of the salon case. A property that wires up Google alone usually finishes with single digits on the page, because guests were prompted by whoever took the booking, not by Maps. A property that wires up the OTA alone loses the walk-in layer. Volume favours the OTA, sequencing says do both, and the per-industry benchmark puts a property next to other trades.
Recent arrivals confirm which side moves. Between 25 June and 25 September 2026 our lodging customers picked up 135 fresh Booking.com reviews and 31 fresh Airbnb reviews. One Croatian apartment operator produced 96 of the Booking figure by itself, roughly 32 a month, which is what a healthy OTA feed looks like and what stops a widget going stale.
Languages: the complication unique to rooms
A property takes guests from everywhere and the feed shows it. Our Booking.com reviews run across 32 distinct languages over 3,276 reviews, about one language for every 102 reviews. Our Google reviews run across 140 languages over 391,212, one for every 2,795. Per review, an OTA feed is an order of magnitude more multilingual than a Google feed.
At customer level it sharpens. That Croatian apartment operator keeps 721 Booking reviews in 24 languages, a 4.42 average and only 42.2% at five stars: a small business with the review profile of a mid-sized hotel. Drop that unfiltered onto an English page and most cards are unreadable to the visitor. Restricting the widget to the language of the page is the fix, and the wider European picture sits in our language mix study.
What to do this week
The order the figures support: wire up Booking.com or Airbnb first because that is where the evidence lives, add Google second for the walk-in layer, keep the platform mark on every card so the imperfect numbers read as verified, and put the widget beside the rate and the booking form rather than in the footer. Our Booking.com reviews widget handles the first step from your public property URL, draws half stars properly, and refreshes every 24 hours so the figure on your site tracks the figure on the OTA. You can open an account and have guest reviews live on the property site today.
Be clear about the plan, since this is where hoteliers get surprised. Our free tier carries Google and Facebook only, and the Plus plan at EUR 9 covers nine platforms that do not include the OTAs. Booking.com and Airbnb belong to the Ultimate plan at EUR 39 a month, which also brings 20,000 held reviews, review curation and several profiles per platform, that last item being what a property with separate apartment listings genuinely needs. Language filtering arrives a step earlier, on Pro. The full ladder is on the pricing page and paid plans open with a 7-day trial.
The honest caveats, in the order they bite. An OTA profile is wired to a public URL, so a property that changes its Booking slug has to redo it. Booking reviews arrive without guest photographs and no widget invents them, so pick a card that does not reserve room for a face. Six of our 23 lodging customers run a single platform and nothing else, so they display a narrower slice of their own standing than they already own. And none of it moves a booking until the snippet sits on the page next to the price.
FAQ
Which review source should a hotel put on its own website first?
Whichever booking channel holds the guest feedback, which for our customers means Booking.com or Airbnb rather than TripAdvisor. As of 25 September 2026, the 23 lodging businesses we host keep a typical 137 reviews on those two platforms and a typical 5 on Google, and ten of the 23 run no Google profile with us at all. TripAdvisor is weaker for rooms than its reputation suggests: of our 26 TripAdvisor profiles, only three belong to a hotel.
Do mixed scores hurt a hotel page?
The mixed scores are why the page works. Just 52.4% of the Booking.com reviews we hold are five stars, against 85.8% on Google, 90.4% on Airbnb and 91.1% on TripAdvisor. A guest arriving from an OTA has been reading scores like 8.6 and 9.2, so a hotel page showing forty flawless quotes contradicts the tab they came from. The same 4.58 average they already saw does not.
Why do Booking.com ratings show up as half stars?
Because Booking scores out of ten and we map that onto five stars, so 8.6 becomes 4.3. In our data 1,027 of 3,276 Booking reviews, 31.3%, carry a fractional value such as 4.5 or 3.5. A widget on a hotel site has to draw half stars or it will quietly round a 7.0 property up to four, and the guest comparing your page with the OTA tab is the person who finds that.
How many languages will a hotel widget have to cope with?
More than a local business. Our Booking.com reviews run across 32 distinct languages over 3,276 reviews, about one language per 102 reviews, while our Google reviews run across 140 languages over 391,212, one per 2,795. A single Croatian apartment operator in the set keeps 721 Booking reviews in 24 languages. Restricting a widget to the language of the page is a Pro plan feature and it matters here more than in any other trade.