Nathan Zhao
简繁EN

From the week

Two Fridays are not automatically comparable

Looking at a booking curve this week, I switched from the comparison chart to a calendar.

We began a booking-curve discussion this week with flights departing on the same weekday. The two lines looked clean together and both dates were Fridays. Before moving on to pickup, I noticed that one set fell before a holiday and the other did not.

I closed the overlay and opened a calendar. Rather than immediately replacing the sample with dates that looked closer, I laid out the days around departure too. The matching weekday was one shared property; the reasons people travelled could still differ.

A slower discussion

One question concerned final load factor; another was whether later bookings would strengthen. My chart could not yet answer either. It had compressed departures before a holiday, during a holiday and in an ordinary working week. Averaging made the differences look conveniently smooth.

I regrouped departures by their position relative to the holiday and compared bookings at the same lead time. That reduced the sample, and some dates had no sufficiently close reference. I left those gaps outside the chart rather than adding unsuitable flights to complete a line.

Nor did the holiday explain everything. Schedule, frequency, groups and the opening of sales still needed separate attention. The holiday label helped decide which observations belonged in the discussion. Selecting that reference eventually took longer than looking at the curve.

Back at my desk I placed a small calendar beside the comparison, marking only the departure dates actually used. I did not insert the entire holiday schedule. The next person opening the file could at least see why those dates had been selected.

The filename stayed the same, but the smoothest line was gone. I left it out. There was now enough space on the screen for the whole calendar.