Should a 10/10 from 12 reviews ever beat a 9.1 from 4,000?

We launch tomorrow, and this is the decision I keep second-guessing.

HotelRoost scores 245,901 hotels with a single formula. The hard part was never the maths — it was deciding how much to trust a small sample. A brand-new guesthouse with twelve perfect reviews genuinely might be excellent. Or it might be twelve friends.

What we settled on: pull every rating toward its country's mean in proportion to how little evidence stands behind it. Fifty reviews is the point where a hotel's own rating carries as much weight as its country baseline; below that the score is deliberately conservative. Then re-base 70/30 against the country, because an 8.6 in Japan and an 8.6 in Egypt are not describing the same hotel.

The consequence bothers me. Small excellent places can almost never reach the top of a city list. It protects against manipulation and it punishes genuinely good new hotels, and I don't know how to have one without the other.

So, two questions for anyone who has built a ranking. Where would you put that threshold — fifty feels defensible but it is arbitrary. And is there a better signal than review count for how much you should trust a rating?

The whole formula is published at hotelroost.com/methodology, deliberately: you cannot argue with a ranking you are not allowed to see. Pick holes in it.