Level 4 · November is always up

Two growth rates

Month over month (MoM) compares a month with the one before it. Year over year (YoY) compares it with the same month a year earlier. In a business without seasons they tell the same story. Harbor & Pine has strong seasons: November and December carry Black Friday, Cyber Monday and holiday gifts, and sell about twice a normal month.

month     net sales    MoM      YoY
2025-10   184,241.76   +26.6%   +19.0%
2025-11   305,172.51   +65.6%    -8.0%
2025-12   293,358.44    -3.9%    -1.2%
2025-01   156,093.59   -47.4%   +39.9%

November’s +65.6% MoM is the season, and it happens every year. Against last November, sales are down 8%. January’s −47.4% looks like a crash and is in fact the best YoY growth of the year.

Seasonality

A pattern that repeats every year at the same time is seasonality. When it is strong, MoM growth mostly measures the calendar. YoY compares like with like: same season, same holidays, same weather. That is why retailers report “same period last year” and why a store’s best November MoM can sit next to its worst YoY.

YoY has its own catches: a holiday that moves (Black Friday falls on a different date each year), a promotion that ran last year and not this year, or a different number of weekends in the month. Name them in a footnote when they matter.

LAG with a look-back

LAG(net_sales, 12) OVER (ORDER BY month) reaches twelve rows back. That only works when every month has a row, and when the rows from last year are still there when the window runs. SQL runs WHERE before window functions, so filter to 2025 in an outer query, after the LAG has seen 2024.

Common mistakes

• Headline MoM growth in a seasonal month. It is the calendar, not performance.
• Filtering before LAG. January ends up with NULL growth, and every YoY is NULL.
• Missing months. If a month has no row, LAG(…, 12) silently compares with the wrong month. Build a full month list first when data can have gaps.