Seasonality means the same swing recurring at regular intervals; NIST describes it as periodic fluctuation. In a single year there is one observation per month, so a repeat cannot be tested: the July peak on the chart has happened once. That summer may also be the one when a new customer arrived, the fleet grew or a route changed.
A hypothetical example shows how counts mislead: 12 deviations in 200 July runs (6 percent) against 2 in 40 January runs (5 percent) is a sixfold gap in raw counts and almost the same rate. The difference may come from more runs rather than from the season, which is why deviations are always read against eligible run counts.
Preparation does not have to wait for year two. Put outside temperature beside the data, follow the same route separately, and write down the date of every change of container, route or crew. When the second year arrives, line up the same months and check whether the peak returns in the same place; for monthly data NIST looks for peaks at lags 12, 24 and 36.
The scale of the decision follows from this. Preparing extra plates for summer or moving a departure earlier can be done on one year of data; buying another cabinet or enlarging the fleet waits for a confirmed repeat. If the note on what the decision rests on (one year, one route, this many runs) is kept, whoever revisits the table a year later does not walk the same road again.
