Methodology
How we measure hotel freshness
Trust is our product. Every number on HotelChrono can be traced back to a source — and one rule governs everything: no quote, no data.
The Chrono Score formula
The Chrono Score is a 1–10 measure of freshness only, kept deliberately separate from star ratings. It is not a black box: it is a single, transparent function of the most recent year we can source. You can recompute any hotel's score yourself from the sourced date on its page.
The exact rule
Freshness fades as time passes since the last completed renovation (−3.5 / year); a building simply ages more slowly (−1.5 / year). The slope is tuned to a typical ~14-year renovation cycle — a hotel due for a refresh lands around the middle of the scale, not the floor. A hotel with no sourced year is left unscored — we never fill the gap with a default number.
Worked examples
The score is a deterministic function of a sourced year — the same inputs always give the same number, and nothing is hidden. What varies is confidence: the quality of the source behind that year is shown separately as a tier (below). A weak source lowers the confidence tier and is labelled as such — it never silently moves the number. Hotels marked under renovationare flagged separately and are never shown as “fresh”. Future versions may add factors; each will be published with its own version number.
Score bands
Fresh
8.0–10
Ageing
5.0–7.9
Tired
0–4.9
Where the data comes from
Every fact carries the weight of its source. When two sources disagree, the higher weight wins; if the conflict is too close to call, a human decides.
A confidence tier on every fact
We don't pretend every date is equally certain. Each build and renovation year carries a visible tier, so you always know how strong the evidence is:
Backed by a strong source — an official register, editorial or trade publication — and re-checked by a capable model.
Cited, but from a weaker or aggregator source. Plausible, lower certainty — the autopilot keeps hunting for a stronger one.
No source on record yet. Shown for context, clearly marked — never counted as a verified fact.
We also continuously re-verify suspect records with a capable model and remove any claim a source can't back — so confidence only goes up over time.
Renovation states — announced is not completed
A renovation that is only announced or under way is not a finished one. We classify each renovation by state, and only a completed renovation raises the freshness score — a hotel mid-refurbishment is shown as “under renovation,” never as “fresh.”
Completed
A renovation that has finished — only this raises the freshness score.
Under renovation
Work announced or under way; the hotel is not yet reopened. Shown as 'under renovation', never as 'fresh'.
Announced
A future renovation is planned but not started.
The rule: no quote, no data
For every year we record, the AI must return the exact sentence it was extracted from. If there is no verbatim quote, the data is discarded — full stop. This is our strongest guard against hallucination, and the reason every hotel profile shows its sources in full.
“Following an 18-month, full-property renovation, the hotel reopened in March 2023.” → built into the timeline, with the source attached.
A living, self-maintaining pipeline
Seed
Import of open data (OpenStreetMap, Wikidata, public accommodation registries).
Enrich
An AI worker reads official sources and extracts structured facts with a cited quote for every date.
Monitor
A daily job scans industry media and chain press rooms for new renovations — the database breathes.
Confidence
Each fact is weighted by its source quality. A weak or wiki-style source alone can never be 'high confidence'.
QA wall
Before a score is shown: is the renovation completed (not just announced)? does it have a source? is the name correct? Issues are flagged openly.
Re-verify
An autonomous loop re-checks records over time — so 'under renovation' becomes 'renovated' once a hotel reopens, and stale data refreshes itself.
Publish
Pages regenerate the moment a fact changes — always fresh, always fast.
Sources & licensing
We build only on legitimately obtained data: OpenStreetMap (ODbL), Wikidata (CC0), public registries and permits, official hotel communications, and industry news. Photographs are used only with a clear licence (Wikimedia Commons, hotel-provided, or user uploads).
We never scrape content or photos from booking platforms. We respect robots.txt and rate limits. Attribution is given where licences require it.
See it in action.
Open any hotel profile and you'll find its timeline, Chrono Score and every source behind it.
View a sample profile →