Transparency

Our methodology

This page publishes the Hoprat score calculation exactly as it runs in the site's code: the shares below are read from the scale at the moment the page is built, they are not copied out by hand. No affiliate commission enters the score, at any point.

What the Hoprat score measures

The Hoprat score is a mark out of 100, recalculated for every search. There is no universal ranking of holiday parks: the same park can be worth 82 to a family with young children and 61 to a couple without children, because those two people are not looking for the same thing.

Each of the ten dimensions is marked out of 100, then the ten marks are combined into a weighted average: that is what gives the score out of 100. The shares below are the starting weights brought back to a hundred; after rounding to one decimal place their sum does not land exactly on 100, so we do not manufacture any total.

The share of each dimension

Starting scale, before any profile adjustment. Total: around 100%.

Distance: 14.4%; driving time from the departure city you have chosen, excluding traffic
Price: 11.5%; our estimate of the accommodation budget, placed against that of every other park in the database
Facilities: 9.6%; indoor pool, wave pool, spa, sauna, children's club, bowling, bike hire, mini-market
Surroundings: 9.6%; forest, sea, lake, mountains, countryside or a town nearby, and nature outings around
Family: 9.6%; children's facilities on site and suitable activities within 40 minutes
Dining: 7.7%; best sourced public rating among the restaurants listed, how close they are, and the park's own restaurants
Things to do: 9.6%; number, variety and proximity of the activities listed around the park
Rainy day: 9.6%; everything that is still possible under cover, on site and nearby
Fine weather: 6.7%; outdoors, water, cycling, outdoor pool
Reviews: 11.5%; the park's average public rating, recorded on an identified platform

The weights change with the profile — and you are the one who chooses it

The shares above are those of a search with no profile. As soon as you say who you are travelling with, it is your choice that shifts the weights, not ours: a baby pushes up “Distance” and “Family” and brings down “Dining”; a family with young children pushes up “Family” and “Rainy day”; a large family pushes up “Price”; a couple pushes up “Dining” and “Surroundings” and almost erases “Family”; seniors push up “Distance”, “Surroundings” and “Dining”; a stay with the dog pushes up “Surroundings”.

A few preferences shift a weight too: “complete quiet” and “romantic & spa” lower “Family”, “children's paradise” raises it, “subtropical pool” raises “Rainy day”. These adjustments are the same for every park in the database: they change your ranking, never the ranking of one operator in particular.

Two anchored scales

Two dimensions are converted onto fixed scales, written down once and for all. They are not derived from the parks currently in the database: a boundary calculated on the panel would move with every park added, and every published score would change in silence.

The journey

The “Distance” dimension depends on the journey alone: a hundred up to 1h, zero beyond 5h, and a linear decline between the two.

The search's “maximum driving time” slider filters parks, it does not mark them: moving that slider changes no park's distance mark.

The public rating

The “Reviews” dimension converts the park's public rating onto an anchored scale: 5.5/10 is zero, 9/10 is a hundred, linear between the two, and capped beyond.

The scale is set on the ratings actually observed on holiday parks, where almost everyone sits between 7 and 8.5. A scale running from zero to ten would separate nothing; a scale whose zero point sat too high would punish a park for having a public rating at all, which is the same lie in the other direction.

Missing data is removed from the calculation, never counted as zero

It is the project's absolute rule, applied everywhere: It is never counted as a zero. The dimension drops out of the calculation, its weight is redistributed proportionally across the remaining dimensions, and the weighted average stays out of 100 — so it remains comparable from one park to another.

The consequence is visible on the park pages: a park for which no public rating has been recorded shows no “Reviews” row, and its score is calculated on the other nine dimensions. Not knowing is not the same as knowing it is bad.

The same rule applies inside a dimension: the facilities share is calculated only on the facilities actually verified, and the quality of the restaurants counts only if at least one public rating is sourced.

No commission enters the score

The ranking is not for sale. The score knows nothing of affiliate networks, commission rates or partnerships: that information simply does not exist in the function that calculates it. An operator therefore cannot buy a position, nor pay to enter the database.

Booking buttons may become affiliate links, flagged as such: that is how the site will be funded, and it comes after the calculation, never inside it. The order of the results is the order of the score, and nothing else.

No operator proofreads these pages before publication, and none has any say over the data of a competing park. The site's advertisements are identified and kept separate from the content: they are never mixed in with the search results.

Where the data comes from

The score currently covers 33 parks, all described in the same data file the calculation reads.

Park listings: facilities, settings, accommodation, activities and restaurants nearby, compiled by hand from the operators' public websites and regional tourist information, then proofread. An unverified facility is marked as such rather than declared absent.
Driving time: measured with OSRM on OpenStreetMap road data, from town centre to park, excluding traffic. These are measured times, not as-the-crow-flies estimates.
Public ratings: public averages recorded on a given date, kept only when the platform they came from is known. A rating whose origin we do not know is deleted rather than kept.
Amounts: unless marked “price recorded on … on the park's website”, the amount is a hoprat estimate, calculated to compare parks with each other. It is not an operator rate.

The full workings are shown on every park page, dimension by dimension, with the sentence that explains the mark: see Center Parcs Les Ardennes for example.

What the score does not tell you

The score compares what can be measured. It replaces neither a visit nor your own judgement: a good score is not a promise of a good stay.

It judges neither the cleanliness, nor the state of the accommodation, nor the welcome: those are things you see for yourself on site, and they vary from one cottage to the next.
The amounts are estimates meant for comparison, not operator rates: the real price depends on your dates, the accommodation and whatever promotions are running.
Driving times are calculated excluding traffic: on a Friday evening at the start of the holidays, allow appreciably more.
Opening hours, restaurants and activities nearby change, close or move: check before you set off.
The score only compares the parks in our database: a park that is absent is not a poorly rated park, it is a park we have not described yet.

A figure looks wrong to you, a detail has changed, a park is missing: write to us. We correct it, and the correction shows up in the score at the next deployment.

To go further

To go further: the list of parks gives the full database, the About page says who edits the site and how it is funded, and the guide explains the rest of the calculation — total budget, programme, weather.