How We Rate Booking Platforms
One rating system per sector, not one for everything
At Flight4, every star rating and platform review starts with real, hands-on research, not a formula borrowed from somewhere else, and never a paid placement. We visit each platform directly, test the process end-to-end, and cross-check what we find against independent third-party sources. Our own historical data, built up since 2006, feeds into the picture too. For certain research tasks (cross-referencing third-party ratings across many sources, or researching detailed coverage facts that would take considerable time to verify by hand) we also use AI as part of the process, working alongside our own direct testing rather than instead of it.
The metrics we choose to measure, the judgement calls behind every score, and the final ratings are all Flight4’s own. AI helps us analyse data at scale and present our findings clearly and consistently; it supports the research, it doesn’t replace the judgement behind it. We think how a rating is made matters as much as the rating itself, which is why we’re upfront about all of this.
- 50 Platform ratings published
- 6 Separate rating systems
- 9–18 Criteria per platform, by sector
- 0 Paid placements in our ratings
Six rating systems, not one
We rate 40 booking platforms across six sectors, and each sector has its own set of criteria rather than one checklist stretched to fit everything. That is deliberate. What makes a car hire broker good (fuel policies, excess cover, what happens at the counter) has almost nothing in common with what makes an attraction ticket seller good. A single shared checklist would have to be so general that it stopped measuring anything useful, or so specific that most of it read N/A on most platforms.
So the number of criteria differs by sector, from 9 to 18, and so does the scale they are scored on. Hotels, car hire, transfers and city breaks score each criterion out of 10, while flights and attraction tickets use a 0–5 scale, which suits a smaller, more focused set of criteria. Each sector’s own methodology page states which it uses. What does not differ is the part that decides a rating: every criterion within a sector counts equally, the star rating is a plain average with no weighting and no bonus, the criteria are published before the scores, and the research is disclosed the same way throughout.
| Sector | Platforms | Criteria | Its methodology |
|---|---|---|---|
| Flights Booking sites and metasearch engines for flight tickets. | 11 | 18 | How we rate flights |
| Hotels Accommodation booking platforms, from global OTAs to specialists. | 10 | 17 | How we rate hotels |
| Car rental Brokers and rental companies, where the small print matters most. | 10 | 18 | How we rate car rental |
| Airport transfers Shuttles, private cars and managed driver networks, where the flight-tracking policy matters most. | 6 | 14 | How we rate airport transfers |
| Attraction tickets Ticketing and tour marketplaces for things to do. | 4 | 9 | How we rate attraction tickets |
| City break providers Tour operators and OTAs selling flight-plus-hotel breaks, rated against what each sets out to do. | 9 | 9 | How we rate city break providers |
How the star rating is calculated
Every criterion is scored on its sector’s scale, and the star rating is the mean of the criteria that apply to that platform. Where a sector scores out of 10, that mean is halved to give a rating out of 5; where it already scores out of 5, the mean is the rating. Either way the figure is rounded to one decimal place, and nothing is weighted.
Worked example
A platform in a sector scored out of 10, averaging 8.4 across its applicable criteria, is rated 4.2 out of 5.
There is no weighting and no bonus. Every criterion in a sector counts the same as every other, which means a strong score cannot be bought with one standout feature, and a platform that is quietly good at everything will out-rank one that is spectacular at a single thing.
Why some criteria are marked N/A
In sectors whose criteria include functions some platforms are not built to offer, a criterion that sits genuinely outside what a platform sets out to do is marked N/A and excluded from that platform’s average rather than counted as a failure. A metasearch engine is not marked down for not selling packages, any more than a library would be marked down for not selling books.
Where it applies, it matters more than it sounds. Scoring an absent function as zero pushes every specialist below every generalist automatically, regardless of how well either does its actual job, and a table built that way looks more objective than it is.
Not every sector needs the distinction. Where the criteria describe things any platform in that sector could reasonably offer, each one is scored for all of them and the sector’s own methodology page says so, along with what that choice does to the results. The rule is that a platform is never penalised for being a different kind of business than its competitors; N/A is one way of honouring that, and stating the effect plainly is the other. Where a whole sector needs that distinction spelled out, the comparison page carries a scope label beside each rating saying what kind of company it is, so the rating and the remit are read together.
How independent reviews factor in
We look at independent review scores, but we do not simply repeat them. They inform our assessment of customer service alongside each platform’s documented support model: whether there is a helpline at all, whether you can reach a person, what the published help documentation actually commits to. Where independent sentiment and documented provision disagree, we say so on the platform’s own page rather than quietly averaging them out.
Where we cite a third-party figure we cite the published number and name its source. We do not reproduce a review platform’s own summary of what its reviewers said, and no score of ours is a copy of anyone else’s rescaled. Review volumes also differ enormously between companies (hundreds of thousands against a few thousand), so two identical scores are rarely equivalent evidence, and we do not treat them as such.
What these ratings do not tell you
Not a safety or financial-protection rating
ATOL cover, ABTA membership and trust-account arrangements differ between platforms and between products from the same platform. Always check what your specific booking carries.
Not a price guarantee
A high rating means a platform does its job well, not that it will be cheapest for your trip.
Not permanent
Websites, inclusions, promotions and support models all change. Every rating carries the date its research was completed, and the figures behind it are a snapshot taken on that date.
Not whole-of-market
We rate the platforms we have researched. Others exist that we have not assessed, and a platform being absent is not a judgement on it. We add platforms as we research them, and we do not accept payment for inclusion or for a higher position.
Why we publish this
Editorial “best of” rankings are common in travel, and they are useful, but only when it is clear whose opinion they represent and how it was formed. These are ours, based on our own research, published alongside the reasoning and the remit so you can judge whether our assessment matches what you actually need.
This page is about how we rate booking platforms. We rate destinations separately and by a different method. See how we rate destinations.
All 6 comparison indexes
Compare booking platforms
All six comparison indexes in one place.
Start a search
Flights, hotels, car hire and tickets, in one place.