Loading the journal
“I want to go to Georgia for ten days.” It is the most natural way to start planning a trip, and it leaves out the single input that most changes the answer. Every seasonal thing a plan knows — the temperature ranges, the rainy-day count, which festivals fall inside the window, what a flight tends to cost — is keyed on a month. Without one, there is no plan to build.
The tempting fix is to default to the current month. It is one line of code, it never fails, and it is wrong most of the time: somebody planning in August is usually not leaving in August, and a Georgian itinerary written for August reads very differently from one written for October.
So a month arrives by one of three routes, and the plan records which one it took.
That third route is where the work is.
First we look for a curated answer. Our city records carry a seasonality note with the months a place is actually good, written by a person; where one exists, it wins, because local knowledge beats a temperature curve. Somewhere can be pleasant in a month everybody avoids for a reason no climate archive records.
Where there is no curated answer, one call to Open-Meteo's historical archive returns the whole year for that location, and each month is scored on the same formula the plan itself displays beside the climate figures. Not a different, hidden ranking — the identical calculation, so the month we chose and the score we show you cannot disagree.
One deliberate thumb on the scale: when two months score close to each other, the sooner one wins. A single point of difference is well inside the noise of a climate average, and it is not a good reason to tell somebody their trip should happen eleven months from now.
The current month is still the last resort
If the destination cannot be resolved well enough to score — nothing curated, nothing in the archive — the plan falls back to the current month. It is then labelled a fallback in the data and said out loud on the page, rather than presented as a recommendation. A guess that announces itself is a different object from a guess that does not.
Each plan stores which route its month came from, and why. That exists so the page can tell you the truth about its own inputs: whether the October in your itinerary is the October you asked for, the October we worked out, or simply the month it happened to be when you typed.
It also matters because long trips are built over several passes, and a plan whose month drifted between passes would contradict itself halfway down. The decision is made once, recorded, and re-read — never recomputed on a whim.
None of this makes the choice authoritative. It is a historical average, and averages are silent about the year you actually travel. What it is, is legible: you can see which month was chosen, on what basis, and what the climate figures behind it were — and change it in one click if you disagree.
Every claim in this post is something you can check on a real plan. One sentence is enough to start one.