The cheapest upgrade in travel is the calendar

Take any destination and any hotel. Compare the second week of August with the last week of September. The room is the same room, the sea is warmer in September, the town is emptier, the staff are less exhausted, and the price is frequently forty per cent lower. Nothing else available to a traveller produces that combination of effects, and it costs nothing but flexibility.

Why the saving is so large

Hotel pricing is driven by occupancy, and occupancy is driven by school holidays far more than by weather. The demand curve therefore has cliffs in it that have nothing to do with whether the destination is pleasant. The week after the schools go back is a different market from the week before, at the same latitude, in the same weather.

Airlines behave the same way with more volatility, which is why the fare can halve across a single weekend in early September. The pattern is so consistent that it can simply be planned around rather than hunted for.

The two windows worth knowing

• Late spring, roughly mid May to mid June in the northern hemisphere. Everything is open, the landscape is at its best, the sea is still cold, and walking is at its most pleasant.
• The month after peak, roughly the second week of September through October around the Mediterranean, later further south. The sea is at its warmest, the towns are emptying, and the light is better.
• Deep winter in a summer destination, which is a different proposition entirely: very cheap, very quiet, and half the town closed.

The sea lags the air by about six weeks. That single fact makes September the best month on most of the Mediterranean and June one of the most overrated.

What to check before committing

Shoulder season has one genuine failure mode, and it is seasonal closure. In small resort towns, a substantial share of restaurants, shops and boats shut within days of the season ending, and the date is local, unpublished and consistent year to year. A town that is delightful on 20 September can be shuttered by 5 October.

The reliable check is not a guidebook. It is the opening hours listed by three or four specific restaurants you would want to eat at, read for the actual dates you are going, and the ferry or boat schedule, which switches to winter timetable on a fixed date and is published.

Weather, stated honestly

Shoulder months carry more variance. A week in late September on a Mediterranean coast has a high probability of being superb and a real probability of two days of rain. A week in August has almost no rain and almost no availability. What you are buying with the saving is a slightly wider distribution of outcomes, which for most trips is an excellent trade and for a wedding is not.

The mitigation is choosing a destination with something to do indoors. A city in shoulder season is nearly risk free. A single beach with one road is not.

The other calendar effect

Within a week, the day of arrival moves prices too. Sunday to Thursday stays are consistently cheaper than Friday to Monday in city hotels and consistently more expensive in resort ones, because the two markets have opposite demand patterns. Shifting a city break by two days is worth a category of hotel on its own.

Spotted an error? Tell the desk