Seasonally adjusted vs. raw sales numbers

Did sales really fall in January, or does January always look like this? Seasonally adjusted data exists to answer that question. It removes the usual pattern of the year, such as holiday peaks and post-holiday slumps, so that what remains shows the underlying trend. Raw numbers show what actually sold. Both are useful, and each answers a different question.

The myth is that adjusted numbers are more accurate. They're not more accurate. They're more comparable, and they come with trade-offs.

What is seasonal adjustment?

It's a statistical process that estimates the typical seasonal pattern in a series and removes it. If a category usually sells more in summer and less in winter, the adjustment scales each month so that a normal summer and a normal winter look the same. Any remaining movement reflects something beyond the usual season.

The process usually handles more than weather-driven seasons. It can adjust for the number of trading days in a month and for holidays that move between months. The seasonal adjustment entry explains the main methods.

Government statistical agencies publish both versions. The Census Bureau's monthly retail trade survey releases seasonally adjusted and unadjusted estimates side by side, which is a good example of how the two work together.

When should you use adjusted numbers?

When you want to compare consecutive periods. Month-over-month changes in raw data are dominated by the season in many categories. Adjusted data removes that, so a rise from one month to the next means something beyond the calendar.

When you want to spot turning points. If a category's adjusted trend flattens or reverses, that's a signal worth investigating. In raw data, the same shift might be hidden inside a normal seasonal swing.

And when you're comparing series with different seasonal shapes. A beverage category and a household category have very different annual patterns. Adjusting both lets you compare their underlying trends on the same footing.

When are raw numbers better?

When the question is about actual volume. Ordering, staffing, supply planning and inventory depend on what will actually sell, including the season. A seasonally adjusted forecast of December doesn't tell a distributor how many cases to ship.

When you're comparing with the same period last year. A year-over-year comparison of raw data already cancels most of the season, since both sides share it. Adjustment adds little there, though it can still help with moving holidays.

And when the series is short. Seasonal adjustment needs several years of history to estimate the pattern reliably. With one or two years, the estimate is shaky, and the adjusted numbers can mislead.

Why do adjusted numbers get revised?

Because the seasonal pattern is re-estimated as new data arrives. Each new month adds information about what "normal" looks like for that time of year. When the estimate changes, past adjusted values change with it.

This surprises people. A month's adjusted figure can differ from what was first published, even though no sales were added or removed. The raw number stays the same. Only the estimated seasonal factor moved.

The practical rule is to state which version you're using and when it was calculated. Mixing adjusted and raw figures in one chart, or comparing an early adjusted estimate with a revised one, causes confusion. The July 2026 same-store sales report is one example of how monthly retail comparisons are framed, and more releases are on the NRS Insights site.