What Is Hotel Mapping? A Practical Guide for OTAs, Bed Banks and Travel Tech Teams

If you aggregate hotel inventory from more than one supplier, you have already met the problem: the same hotel shows up three, four, five times on your results page, each with a slightly different name, a different address format and, of course, a different ID. Hotel mapping is the discipline of fixing that - once, centrally, and for every supplier at the same time.

This guide explains what hotel mapping is, why it is harder than it looks, how matching engines actually decide that two records are the same property, and what a good mapping layer should give your product and engineering teams.

Why the same hotel arrives five different ways

Every bed bank, wholesaler and channel manager keeps its own hotel database. There is no global registry they all agree on, so each one assigns its own identifier and writes its own content. Pull the same property from five suppliers and you typically get:

Field	Supplier A	Supplier B	Supplier C
Name	Hilton Dubai Jumeirah	HILTON JUMEIRAH BEACH	Hilton Dubai Jumeirah Resort
Address	Jumeirah Beach Rd	Jumeirah Beach Road, Dubai	P.O. Box 26878, Dubai
City	Dubai	DUBAI	Jumeirah
Coordinates	25.2048, 55.2708	25.2051, 55.2701	missing
Star rating	5	5.0	Five Star

None of these fields is reliable on its own. Names change after rebrands. Addresses are formatted differently in every market. Coordinates are sometimes the city centre rather than the building. And the city itself may be a district, a suburb or a tourist region depending on which supplier you ask.

The downstream cost is real: duplicate results that confuse travellers, broken price comparison (you cannot show the cheapest rate for a hotel if you do not know it is the same hotel), fragmented reviews and content, and analytics that count one property as five.

How hotel mapping actually works

A mapping engine takes a supplier record and answers one question: which master hotel is this? In practice that is a pipeline, not a single lookup.

1. Normalise the input

Before any comparison, every record is cleaned: casing and accents are normalised, chain prefixes and legal suffixes (Hotel, Resort, Ltd, S.A.) are stripped for comparison purposes, addresses are tokenised, and country and city are resolved against a master geography. This step alone removes a surprising share of false negatives.

2. Narrow the candidates

Comparing every supplier hotel against a million masters is wasteful. Candidate masters are selected by geography first: same resolved city, and within a small radius when coordinates exist. Only those candidates go on to detailed scoring.

3. Score the match

Each candidate is scored on several independent signals:

• Name similarity - token-based fuzzy matching that tolerates word order, abbreviations and missing words
• Geo-distance - metres between coordinates, with sensible handling for missing or city-centre coordinates
• Address overlap - street tokens, postal code, PO box
• Structural hints - star rating, chain or brand, property type

A weighted combination produces a confidence score. High-confidence matches are linked automatically; medium-confidence ones go to a review queue; low-confidence records create a new master.

4. Assign the master ID and keep it stable

Once linked, the supplier ID is permanently mapped to the master ID. Your product keys everything - search index, cache, reviews, images, analytics - off the master ID. When a supplier renames or re-codes a property, the mapping is updated and your IDs do not move.

5. Deduplicate and monitor

Mapping is never finished. New hotels open, suppliers re-import catalogues, and two masters occasionally turn out to be the same building. A mature mapping layer runs continuous duplicate detection and exposes a change feed so consumers can sync without full reloads.

Build vs. buy

Teams usually start by building a matcher in-house. The first version takes a fortnight and handles the obvious cases. The next eighteen months are spent on everything else: multi-language names, properties that moved, suppliers that merged two hotels into one code, review tooling, and the operational load of keeping it all current.

A dedicated hotel mapping service turns that into an API call. At Mappin the master dataset currently covers 1,012,000+ active hotels in 146,000+ cities across 258 countries, with 1.23 million supplier mappings across Hotelbeds, Dida, RezLive, Restel and Travellanda. Over 141,000 master hotels are linked to two or more suppliers - exactly the properties where price comparison and deduplication matter most.

What to look for in a hotel mapping provider

Whether you build or buy, these are the capabilities that matter in production:

1. A stable master ID that never changes when supplier codes do
2. Supplier-to-master lookup in both directions - given a supplier code, get the master; given a master, get every supplier code (Mappin's /api/provider-map/ does this in batches of 100)
3. A change feed so you can sync deltas instead of re-pulling a million rows (/api/changes/?since=...)
4. Geography that is mapped too - cities and countries need master IDs just as much as hotels
5. Transparent confidence and a review process for the small percentage of matches that need a human
6. Predictable limits and pricing - per-day quotas, rate limits you can design around, and a free tier to evaluate with

Getting started

The fastest way to understand your own duplicate problem is to measure it. Take one busy city, pull hotels from each of your suppliers, and run them through a mapping API. The number of supplier records that collapse into one master is your duplicate rate - and the amount of comparison and content you were silently losing.

You can do that today with a free Mappin account: create an API key, open the Playground, and query /api/city/?city=Dubai followed by /api/city/?city_id=... to see master hotels with every supplier code attached. The full reference is in the API documentation.

Frequently asked questions

What is hotel mapping?

Hotel mapping is the process of identifying when hotel records from different suppliers (Hotelbeds, Dida, RezLive, Restel, Travellanda and others) refer to the same physical property, and linking them to one master hotel ID.

Why can't I just match hotels by name?

Names differ between suppliers (Hilton Dubai Jumeirah vs Hilton Jumeirah Beach), include chain prefixes or legal suffixes, and change after rebrands. Reliable matching combines name similarity with geo-distance, address, city, star rating and chain signals.

What is a master hotel ID?

A master hotel ID is a single, supplier-independent identifier for a property. Each supplier's own ID is mapped to it, so your search, caching, reviews and analytics key off one ID instead of five.

How many hotels does Mappin map?

Mappin maintains over one million active master hotel records across 146,000+ cities and 258 countries, with 1.2 million supplier mappings across five connected suppliers.