Field notes · 28 July 2026

You Compare Hotel Prices at the Wrong Step

We asked the biggest booking sites the same question about the same hotels in 65 cities. Their answers disagreed by 22 percent at the median. Your deal-hunting habit is aimed at the wrong end of the problem.

Indian travellers are the best last-step price comparers in the world. You know the ritual. Shortlist the hotel, open Booking.com and MakeMyTrip side by side, check the member rate, check the mobile rate, check whether the coupon code from the credit card still works. Squeeze out the last few hundred rupees before paying. It is a good ritual and it works. The gaps it finds on the same room are usually modest, and regulators have spent years litigating exactly that layer: the EU's top court examined Booking.com's price parity clauses in 2024 and declined to treat them as mere side agreements to the main contract.

That ritual guards the smallest gap in the whole transaction, at the last possible moment, after every real decision has already been made. The much bigger gap sits at the first step, in the number you used to set your budget. And almost nobody checks that one.

The question you only ask once

Before you shortlist anything, you answer a quieter question: what does a decent hotel in this city cost per night? You answer it once, early, usually from whichever app you opened first, and every decision after that leans on it. It decides whether the city feels affordable. It decides what your package operator's quote looks like next to "the market." It decides what a fair nightly rate feels like for the rest of the planning.

We maintain price benchmarks for the ScoreMyTrip audit engine, which means we ask that first-step question professionally, in a controlled way: same city, same class of hotel, same dates, same two-adult occupancy, same filters, taxes included, on more than one major booking site. Then we compare the answers.

Across 247 same-question comparisons in 65 cities, checking Booking.com against Hotels.com and Trip.com, the two sites' typical prices for the same tier in the same city disagreed by 22 percent at the median. Only 24 percent of comparisons landed within 10 percent of each other. In 41 percent of them the gap exceeded 30 percent. In 23 percent it exceeded 50 percent.

The pattern gets worse as you go upmarket. Budget-tier answers disagreed by about 20 percent at the median. Mid-tier by 23 percent. Luxury by 32 percent, with more than half of luxury comparisons off by over 30 percent.

Some concrete cases from our current benchmark data, all for one night, two adults, taxes in:

Each row compares the same class of hotel in the same city on the same dates; the sites simply return different medians for it. If you set your Rome budget from one app and your operator priced from the other, you are arguing across a nearly threefold disagreement neither of you knows exists.

Why the answers disagree

There is no scandal here, which is what makes it dangerous. Four ordinary mechanisms do the work.

The tier itself is not a fact. "3-star" sounds like a regulated grade. Mostly it is not. India's official hotel classification is voluntary, and the government's own April 2025 update counts 2,177 hotels classified as 3, 4, or 5 star since 2019, a small fraction of the properties listed online. Europe's Hotelstars Union harmonises stars across 21 countries, but Italy is not a member. Hotels.com states in its own terms that star ratings are "not a representation or promise of any particular feature or amenity." So when two sites bucket "3-4 star, Rome," they are bucketing different pools of properties and calling them the same thing.

The ranking is not neutral. The UK's competition regulator concluded in 2019, after investigating six booking sites, that search results were influenced by the commission hotels paid, and that this was not made clear to users. The first screen of results, the screen your sense of "typical" comes from, is a curated view, and each site curates differently.

The rates come from different pipes. Sites source inventory through direct contracts, wholesalers, and regional partners, and the same property can sit in one site's system at a contracted rate and in another's through a wholesale rate agreed months earlier. Both numbers are real; they answer different commercial arrangements.

The displayed price is assembled differently. Taxes in or out, resort fees, single or double occupancy defaults, member pricing applied or not. The same regulator's 2019 investigation also secured commitments on pressure selling, misleading discount claims, and hidden charges. Display practices have improved since. They have not converged.

Where the anchor gap actually costs you

Three places, in ascending order of expense.

First, the budget you set. Anchor off the high site and a city you could afford reads as out of reach. Anchor off the low one and your trip runs 30 percent over before you have booked a single night.

Second, the package quote you judge. When an operator quotes you a Rome package, you sanity-check it against a quick app search. If your anchor came from the app whose Rome median is ₹13,500 and the operator built from supplier rates closer to the other answer, you will read a fair quote as inflated, or an inflated one as fair, with real conviction either way.

Third, every averaged number downstream. Blog posts, "cost of travel" articles, and AI answers are trained on whatever site their sources happened to check. The anchor gap does not stay on booking sites. It propagates into everything that cites them.

The fix is a habit you already have

You do not need a new skill. You need your existing two-app habit moved from the last step to the first.

Before you set a budget or judge a quote, ask two independent sites the same first-step question and force the comparison to be honest: same dates, two adults, taxes switched to inclusive, sorted by the default ranking rather than lowest price, and read the middle of the first page rather than the cheapest outlier. The whole check takes five minutes. If the two answers land close together, you have a real anchor. If they are 40 percent apart, you have learned the most useful fact about that market before it cost you anything: the "typical price" you were about to trust was one witness, not a verdict.

That is also the principle our audit engine is built on. Every benchmark band ScoreMyTrip uses is sampled with a fixed methodology and cross-checked across independent sites, and where the sites disagree beyond tolerance we triangulate a third source and flag the cell rather than quietly picking a side. When an audit tells you a package's hotel spend is above or below typical for its tier, that comparison is the whole point of the product, so we treat the anchor as something to verify, not assume. If you have a quote in hand, run it through an audit and see what it looks like against a checked anchor instead of a single app's first screen.

Method note: 247 comparisons across 65 cities, July 2026 benchmark data. Each comparison takes one city and hotel tier and compares the median nightly price (per person, twin share, taxes included) sampled on Booking.com against the same sample design on Hotels.com or Trip.com: same dates, same occupancy, relevance sort, top-of-list sample with 10 percent trims. Gap is measured symmetrically as the larger answer over the smaller, minus one. Bhutan is excluded (managed-tariff market). Attributed examples are room-per-night figures for two, rounded, as sampled on the dates in our records; live prices will differ. Sources: CMA hotel booking sites investigation, February 2019; PIB on HRACC classification, April 2025; Hotels.com terms of service; Hotelstars Union; CJEU judgment C-264/23, September 2024.