The analysis
Describe enough operators the same way and the market starts explaining itself.
Once you have a few dozen operators described the same way, patterns show up that no single brochure would tell you. These are observations about the market, drawn from reading the catalogues side by side — not recommendations, and not about any particular traveller.
The corpus doesn't behave like one industry. It falls into three worlds that barely compete with each other, and knowing which one you're shopping in explains most of the confusion.
Self-drive and tailor-made. An operator arranges the car, the route, the hotels and the ferries, then gets out of the way. You drive. This world is remarkably consistent on comfort — accommodation is almost uniformly private and en-suite — because the whole proposition is "arranged, but independent". It is also where the deepest regional catalogues live: the specialists here have often done one region for decades and know it far better than a generalist ever will.
Small-group tours. A guide, a fixed departure, a set of strangers who become companions. The content is frequently excellent and the price per day lower, but this is the world where the tiers matter most: the same operator will run the same itinerary at two or three comfort levels, and the gap between them is real. Shopping the brand rather than the tier is the single most common way to be disappointed here.
Premium and editorial. Bespoke operators who quote on request rather than publish prices, expedition outfits reaching places no road goes, and magazines that book nothing at all but tell you where to look. They serve a different purpose from the first two worlds: not a trip you compare, but a trip you commission — or an idea you follow up elsewhere.
Price per day sorts more honestly than price. A fourteen-day trip is not "expensive" next to an eight-day one. Reduced to a daily rate, the field compresses dramatically: most tailor-made operators sit in a narrow band, and the genuine outliers — expedition and bespoke — sit an order of magnitude above it, for reasons that are visible in what they include.
Comfort tracks the travel model, not the price. This is the finding that surprised us most. Shared facilities show up in two completely different situations, and they mean opposite things. On a mountain hut traverse or a desert camp, shared facilities are the experience — there is no other way to be there. On a cheaper tier of a mainstream group tour, they are a cost decision, and an en-suite version of the same trip usually exists a tier up. Same field on a spreadsheet, entirely different judgement.
Season is a harder constraint than most catalogues admit. A brochure is always in season. The reality is that any given month is right for perhaps a third of the world and wrong for the rest — monsoon, heat, hurricane season, or a polar window that closes in weeks. An operator's willingness to say "not then" is one of the more reliable signals of how they think about their customers.
Review scores need their denominator. A 4.8 average over seventy reviews and a 4.7 over five thousand are not the same claim. Neither are two scores from different platforms: the same operator can sit near the top on one aggregator and mid-table on another, because the populations differ. We record both the score and the count, and where a source's public record is thin we say so rather than quietly averaging it away.
This is a point-in-time reading, and it ages. Prices and availability drift within months; catalogues get restructured; an operator that was quiet for two years starts publishing again. Some of the underlying figures are approximations — bespoke operators publish no prices at all, and a few aggregate scores come from what the operator reports rather than a source we could verify directly. Where that's true of a specific source, it's noted on that source's entry.
None of this is a ranking. Sources here are grouped by what they're for, not ordered by score — a sorted list would imply a precision the evidence doesn't support, and would quietly turn a considered judgement back into the thing this whole method exists to avoid.