What a moving annual total is
A moving annual total, almost always written MAT, is the sum of the most recent twelve months of a metric, recalculated each period by adding the newest month and dropping the one that fell out of the window. In syndicated CPG reporting the same idea usually arrives as the latest 52 weeks, which is the same window expressed in retail weeks rather than calendar months.
The reason it exists is seasonality. A salsa brand that does a third of its year between Memorial Day and Labor Day will show a June that looks like a triumph and a February that looks like a collapse, and neither is information. A full trailing year always contains exactly one of every season, so what is left when you compare two of them is the trend.
How a moving annual total is calculated
The formula is a subtraction and an addition, not an average:
MAT this period = MAT last period + newest month - the month it replaced
Here is Northfield Foods, an invented salsa brand, rolling through spring 2026.
| MAT as of | Month added | Month dropped | Moving annual total | Change from prior |
|---|---|---|---|---|
| Mar 2026 | Mar 2026 | Mar 2025 | $14,820,000 | |
| Apr 2026 | $1,340,000 | $1,180,000 | $14,980,000 | +$160,000 |
| May 2026 | $1,290,000 | $1,410,000 | $14,860,000 | -$120,000 |
| Jun 2026 | $1,520,000 | $1,250,000 | $15,130,000 | +$270,000 |
May is the row worth staring at. The brand sold $1,290,000 in May 2026, a perfectly healthy month, and the moving annual total fell $120,000. It fell because May 2026 replaced a $1,410,000 May 2025. A MAT never moves on the newest month alone. It moves on the difference between the two ends of the window, which is why a MAT can decline in a month you grew and rise in a month you shrank.
That property is the source of most confusion about the metric, and it is also the point of it: a falling MAT in a good month means last year's comparable was better, which is exactly what you want a trend measure to tell you.
MAT vs YTD vs latest 52 weeks
Three reporting periods get quoted in the same meeting as though they were interchangeable. Read against Northfield's data at the end of June 2026:
| Period | Window | Dollars | Prior-year comparable | Growth |
|---|---|---|---|---|
| MAT | Jul 2025 to Jun 2026 | $15,130,000 | $13,760,000 | +10.0% |
| YTD | Jan to Jun 2026 | $7,940,000 | $7,180,000 | +10.6% |
| Latest 52 weeks | 52 weeks ending 27 Jun 26 | $15,090,000 |
Same brand, same day, three answers. MAT growth is 15,130,000 / 13,760,000 minus 1, which is 10.0%. YTD growth is 7,940,000 / 7,180,000 minus 1, which is 10.6%. The half-point spread is real and it means something: the back half of 2025 grew more slowly than the first half of 2026, and the MAT is still carrying it while the YTD is not.
The latest-52-weeks number differs from the MAT by $40,000 for a duller reason. It stops on a Saturday and the MAT stops on the last day of June, so the two windows contain slightly different days. That gap is not an error to chase. It is a reason never to mix the two in one table.
YTD has one property the others do not: it resets. In January a YTD read is one month of data being asked to carry a year's worth of narrative, and it stays noisy through about April. A moving annual total has the opposite problem, which is that it is slow. Twelve months of history means a genuine inflection takes two or three months to become visible, so a MAT will never be the metric that tells you a promotion worked. For that you want a period short enough to isolate the event, which is the whole argument in choosing a baseline period.
What a 53rd retail week does to a rolling annual total
Most retail calendars are 4-5-4: months of four, five and four weeks so that comparable months hold the same number of weekends. That structure covers 364 days, so the extra day accumulates, and the National Retail Federation adds a 53rd week when four or more days of January are left over after the 52-week calendar is laid out. It says this happens roughly every five to six years, and names 2006, 2012, 2017 and 2023 as 53-week years.
Here is the part that matters for a moving annual total, and it is the opposite of what people expect. A MAT is immune to the 53rd week, because a MAT is defined as 52 weeks and always holds exactly 52. What breaks is everything you compare it against:
- A fiscal-year total is not immune. In a 53-week year the reported year holds one more selling week than the MAT does. Northfield averages $290,962 a week against a $15,130,000 annual total, so that extra week is worth about 1.9% of the year. Put a 53-week fiscal total next to a 52-week MAT and you have manufactured 1.9% of growth out of the calendar.
- The year-ago comparable shifts. NRF restates a 53-week year for comparability by pushing each week back one week and ignoring the first week of the fiscal year, which keeps the holidays aligned. Some companies instead simply ignore the 53rd week. Those two conventions produce different year-ago numbers, so a MAT growth rate is only meaningful once you know which one the source used.
- Quad-week and monthly aggregations drift. Syndicated databases are usually built on rolling weeks and quad-weeks rather than calendar months, so a 53-week year pushes one quad-week's contents into a different position than the calendar month it was standing in for.
The practical rule: never compare a rolling 52-week window to a fiscal-year figure without asking whether either side contains 53 weeks, and never accept a year-over-year growth rate across a 53-week boundary without asking how the base year was restated.
Choosing a reporting period
Syndicated providers hand you a menu. Nielsen and Circana databases carry individual weeks plus aggregates of 12, 13, 24, 26 and 52 weeks and year to date, with quad-weeks as the month substitute, and the standard subscription holds two rolling years of that history, which is 104 weeks or 26 quad-weeks. Which one to reach for:
| Question you are answering | Period to use |
|---|---|
| Is the brand structurally growing? | MAT or latest 52 weeks |
| How is the current fiscal year tracking? | YTD, against prior YTD |
| Did a specific promotion work? | Promo weeks vs a matched baseline |
| Has a recent trend broken? | Latest 12 or 13 weeks |
| Is a new item taking hold? | Weekly, since first scan |
One warning that costs people credibility in category reviews: do not pick the period after you see the numbers. Decide which period answers your question, write it in the header, and keep it there when it stops flattering you.
Where Scout fits
Every period on that menu is the same transactions summed over a different window, and the work is keeping the windows honest across retailers whose calendars do not agree. Scout reads your retailer POS and syndicated feeds and computes MAT, YTD, latest-52 and custom windows off one dataset, holding the calendar convention constant so a MAT pulled in June and a MAT pulled in July are the same measurement. It reports the period you asked for and shows the window behind it. Choosing which period belongs on the slide is still your call.
The short version
- A moving annual total is the trailing twelve months of a metric, rerun each period by adding the newest month and dropping the one it replaced.
- A MAT moves on the difference between the two ends of its window, so it can fall in a month you grew. Northfield's MAT dropped $120,000 in a month that sold $1,290,000.
- MAT and YTD answer different questions and will not match: +10.0% and +10.6% on the same brand on the same day.
- A MAT always holds 52 weeks, so the 53rd retail week never enters it. It enters the fiscal-year total you are comparing it to, worth about 1.9% of an annual figure, and NRF restates the prior year to compensate.
Sources: CPG Data Tip Sheet, "Timing is Everything: Which Time Periods Should You Get on Your Database?"; National Retail Federation, "4-5-4 Calendar".