Most programmatic SEO scales by adding rows. More products, more cities, more listings.
Booking.com scaled by adding adjectives. On 20 August 2026 we read its robots.txt and found 354 sitemap index files — and 312 of them are the same page type, one per hotel theme: all-inclusive, capsule hotel, glamping, casino, farmhouse.
We then measured what that buys, with our own tools: 1.4 million referring domains, and a site audit that came back at 100 out of 100.
Below: how the themes multiply, the one number that shows the discipline behind them, and what you cannot copy.
Results at a glance
Booking publishes a page for a theme in a city in a language. There are 312 themes and 45 languages, so the theme sitemaps alone come to about 14,000 files.
The discipline is in the size of each one: casino has 310 English city pages, farmhouses has 11,775. Each theme is published only where the hotels exist.
What Booking actually sells
Booking.com sells rooms it does not own. Its inventory is other people's hotels, and every property carries attributes: it has a pool, it is a farmhouse, it takes dogs.
Those attributes are the raw material. A hotel is one row, but "hotels with a casino, in this city, in Portuguese" is a page — and the same hotel supports dozens of them.
So the catalogue does not grow when Booking signs a hotel. It grows when Booking names a new way to want one.
Many sitemaps, one repeated shape
The first surprise is in robots.txt: 354 Sitemap: lines, every one an index of indexes. Sorting them by name shows how lopsided the system is.
Your version: sort a competitor's sitemap names alphabetically and count the families. The one that repeats is the bet they actually made, and it is rarely the one their homepage talks about.
Themes multiplied by languages
Open one theme and the multiplication shows up. The all-inclusive index holds 45 child files, one per language: Arabic, Bulgarian, Catalan, Czech, through to Thai, Turkish and both Chinese scripts.
The URLs read like a formula: /all-inclusive/city/fr/belle-plagne.en-gb.html. Theme, then country, then city, then language.
So the themed sitemaps alone come to roughly 14,000 files — 312 themes times 45 languages — before counting a single city inside them.
Some themes are not adjectives at all. bestwestern and 7days_inn are hotel brands. alberghi and gites are the local words for lodging in Italian and French, given their own sitemaps.
Your version: your themes are the filters your customers already use. If your product has facets, you have a theme list; the work is deciding which facets are worth a URL, not inventing them.
Glamping is a real market
A theme only earns its 11,000 pages if people search for it. We checked one with the free Keyword Research tool.
That is the test each of the 312 themes has to pass. "Glamping" is a word people use, so it gets a sitemap and 11,317 English city pages.
Your version: check the demand before the template, not after. A theme with no search volume produces pages nobody asked for, however cleanly they render.
Casinos lose, farmhouses win
This is the number that shows the discipline. We counted the English city pages inside six themes.
| Theme | English city pages | What that means |
|---|---|---|
| farmhouses | 11,775 | rural inventory almost everywhere |
| glamping | 11,317 | a real category, not a niche |
| all-inclusive | 4,903 | concentrated in resort countries |
| bestwestern | 719 | a brand, so only its own hotels |
| capsule hotel | 458 | mostly Japan |
| casino | 310 | few cities have one |
A 38-fold spread between the smallest and largest, in the same language, from the same template.
That is the part most copies get wrong. The lazy version of this system publishes "casino hotels in every city" and produces thousands of empty pages. Booking publishes 310, because 310 is how many cities have a casino hotel.
Your version: bind every generated page to a row that exists. If the query returns nothing, the page should not be built — and your page count should be allowed to come out uneven.
What all those pages earned
We ran booking.com through our free Backlink Analyzer.
Compare it with the Canva profile we measured last week: 624,000 referring domains at 93% dofollow. Booking has more than twice the domains and a far weaker follow ratio.
That fits how travel links get made. Booking URLs are pasted into forums, itineraries, comparison widgets and aggregators — contexts that nofollow by default.
Your version: do not benchmark your dofollow share against another industry. A 56% profile is excellent for travel and would be a warning sign for a design tool.
A perfect audit score
We pointed our free Website Audit at booking.com and got the first clean sheet in this series.
Across the series the same audit returned 79 for Wise, 90 for RTINGS, 91 for Canva. Booking is the only 100.
One caveat worth stating: the detailed technical list still flags structured data as missing or invalid on the homepage, even though the summary counts zero issues. A perfect score is a summary, not a promise.
Your version: a site this size scoring 100 is not luck, it is a team treating the crawl surface as a product. Read your own group scores, and read the detail rows even when the headline says zero.
What you can copy
- Make a theme list from your existing filters, not from a keyword brainstorm.
- Bind every page to real inventory. Uneven page counts per theme are the sign you did it right.
- Treat language as a dimension, deliberately, rather than translating whatever exists.
- Benchmark your dofollow share within your own industry.
- Read a competitor's sitemap names before their content. The repeated one is the strategy.
What you cannot copy
You cannot copy the inventory. Every theme page works because thousands of real hotels sit behind it, with prices and reviews. Without that, the same template is a directory of empty pages.
You cannot copy 1.4 million referring domains. They accumulate because people share where they are going, over twenty years.
What we could not measure, stated plainly. We have no traffic, revenue or conversion data of our own for Booking, and we used no third-party traffic estimate.
We measured six themes in one language out of 312 themes across 45 languages, so the 310-to-11,775 range is a sample, not a full census. We did not count every URL Booking publishes and we do not claim a total.
We also cannot see how many theme-and-city combinations Booking chose not to publish, only the sizes of the ones it did. And the audit reflects the homepage, not the millions of themed pages behind it.
Run this census yourself
- Read
/robots.txtfirst. Count theSitemap:lines and sort them by name. - Group the names into families. The family that repeats is the bet.
- Open one index and see what the children are split by — language, country, or letter.
- Count occurrences, not lines. These files are often a single line, so
grep -creturns 1. Many are gzipped, so pipe throughgunzip -cfirst. - Measure two or three members of the big family and compare their sizes. An even spread means pages were generated; an uneven one means they were earned.
- Pull the link profile and judge the dofollow share against the industry, not against a number you read somewhere.
The same Backlink Analyzer report shown above, on any domain you like. You get total backlinks, referring domains, the dofollow and nofollow split, referring IPs and subnets, and the top referring pages with their anchors — which is how you tell links that were earned from links that were placed.
- The same check that returned 1.4 million referring domains for Booking
- Dofollow split and subnet spread, not just a headline total
- 3 checks a day, no signup, no card
Frequently asked questions
How many sitemaps does Booking.com have?
Its robots.txt declared 354 Sitemap: lines when we read it on 20 August 2026, and every one of them is an index pointing at further files. 312 of the 354 are themed city pages; the remaining 42 cover flights, cars, attractions, cities, countries, regions, districts, landmarks, airports, beaches, hotel reviews and editorial.
What is a themed city page?
A page for one hotel attribute in one city in one language, at URLs like /all-inclusive/city/fr/belle-plagne.en-gb.html. The themes include property types (farmhouses, capsule hotels, glamping), facilities (casino, all-inclusive), hotel brands such as Best Western, and local words for lodging like the Italian "alberghi".
How many pages is that in total?
We did not count them all and we are not going to guess. What we measured: 312 themes times 45 languages is roughly 14,000 sitemap files before any city is counted, and a single theme in a single language ranged from 310 pages (casino) to 11,775 (farmhouses). Any total you see quoted, including one from us, would be an extrapolation.
Why do the themes have such different page counts?
Because each is bound to real inventory. Casino gets 310 English city pages because roughly that many cities have a hotel with a casino; farmhouses gets 11,775 because rural stays are almost everywhere. The uneven spread is the discipline — the failure mode of this pattern is publishing every theme in every city and producing empty pages.
How strong is Booking.com's backlink profile?
Our Backlink Analyzer returned Domain Rank 97 with 563,616,610 backlinks from 1,379,527 referring domains on 20 August 2026, split 315,990,249 dofollow to 267,854,223 nofollow — about 56% dofollow — across 178,487 referring IPs and 71,075 subnets.
Is 56% dofollow good or bad?
For travel it is normal and strong. Booking links get pasted into forums, itineraries and aggregators, which nofollow by default.
For comparison we measured Canva at 93% dofollow across 624,000 referring domains — a different industry with different linking habits. Compare within your category, not across.
Did booking.com really score 100 on your audit?
Yes: 100 out of 100 with zero critical issues, zero warnings and zero notices across 123 checks, and perfect group scores for on-page, technical, speed and mobile. It is the only 100 in this series — Wise scored 79, RTINGS 90, Canva 91. The detailed technical list does still flag structured data on the homepage, so read the rows as well as the score.
How was this case study measured?
Every figure is ours, measured on 20 August 2026. The sitemap taxonomy came from fetching booking.com's robots.txt and its sitemap indexes directly and counting <loc> occurrences after gunzipping.
The link profile, keyword data and technical figures are single free-tier runs of our own Backlink Analyzer, Keyword Research and Website Audit. No third-party traffic estimate was used.
Decide your unit of scale
Every site in this series answered the same question differently: what is the thing we add more of? RTINGS adds measurements, Wise adds currencies, Zapier adds app pairs and refuses to publish most of them, Canva adds templates its users made.
Booking adds ways of wanting a hotel — and then only builds the ones with rooms behind them.
So the question for your own build is not how many pages you can generate. It is what your unit is, and whether every unit you ship has something real underneath it.
Start where we finished — check a backlink profile free — and you will see in about a minute whether your links were earned or placed.
Measured on 20 August 2026. The taxonomy comes from booking.com's robots.txt and sitemap indexes, fetched directly and counted by <loc> occurrence after gunzipping.
Link profile, keyword and technical figures are single free-tier runs of our own tools. Exact values were 1,379,527 referring domains and 563,616,610 backlinks; the body rounds them for readability.
Theme page counts are a six-theme sample in one language out of 312 themes across 45 languages, and no site-wide total is claimed. Sitemap URLs are not indexed pages and not traffic.
No traffic or revenue data for Booking was available to us and none is claimed. Third-party metrics change; re-run anything here before you rely on it.