On October 15 we told you September's online sales were close to flat. Census revised that number on Tuesday. It now shows a 1.1 percent gain.
The Census Bureau released its advance retail report for October at 8:30 a.m. on Tuesday, November 17. Nonstore retailers, the category that carries most of ecommerce, sold $128.2 billion, seasonally adjusted. That is up 1.2 percent from September and 11.8 percent from October 2025. Total retail and food services sales came to $748.2 billion, up 0.2 percent on the month and 6.8 percent on the year. (Census Bureau release CB26-181, November 17, 2026)
Census says the 0.2 percent headline is statistically indistinguishable from zero. The online gain clears that bar, using the standard error Census publishes for the category.
This is the last monthly retail read before peak. November's numbers come out December 16, after Cyber Week has shipped.
September was stronger than the first estimate
The first September estimate put nonstore sales at $124.3 billion, up 0.3 percent on the month and 8.2 percent on the year. (Census release CB26-165, October 15, 2026) Tuesday's release moved September to $126.7 billion. That is a 1.1 percent monthly gain and 10.4 percent on the year.
A 0.8 point swing is big for this line. Over the last 12 months, the average size of a revision to the online monthly change has been 0.3 points, according to the release's own table.
The upgrade went beyond online. Core retail sales for September were revised from a 0.1 percent gain to 0.5 percent, and October added 0.6 percent, Reuters reported. (Reuters via Yahoo Finance, November 17, 2026)
The three months through October now run 10.4 percent above the same stretch of 2025. In our September retail sales post, we said a plan built on double-digit online growth was running about two points hot. On the latest three months, it is close to right.
Online is pulling away from stores
These figures are dollars. Census does not adjust them for price changes, so inflation lifts every line. The distance between the lines tells you more than any one of them.
| Census category | October 2026 vs. October 2025 |
|---|---|
| Nonstore retailers | +11.8% |
| All retail and food services | +6.8% |
| All retail and food services, excluding gas stations | +4.9% |
| Clothing and accessories stores | +3.1% |
| Department stores | -1.6% |
| Gas stations | +30.3% |
Gas stations account for almost two points of the 6.8 percent headline.
Strip cars and fuel out of store-based retail, and what is left grew about 3.4 percent. Online grew 11.8 percent. Prices lift both lines, so most of that eight-point gap is share moving from shelves to parcels. Nonstore sales were 17.1 percent of all retail and food services sales in October, up from 16.4 percent a year earlier. (Both calculations are ours, from Census Table 1.)
The holiday forecasts used the softer numbers
The National Retail Federation released its holiday forecast on October 15, the same day Census published the September estimate that has since moved up. NRF expects November and December retail sales to grow 4.2 percent over 2025. (NRF, October 15, 2026) Run October through NRF's core retail definition and sales were up a bit over 5.5 percent on the year, by our math.
One month is not a season. Reuters said the October online jump likely reflected Amazon's Prime Big Deal Days promotion. September came before that event, though, and it was revised up anyway.
What this means for your brand
Plan the next six weeks for the high side. If you trimmed your peak plan after the first September print, put the volume back.
On October 17 we suggested building December on unit growth of 2 to 3 percent. That figure described all of retail, stores included. For a brand that sells online, it now looks low.
At 1,000 to 25,000 orders a month, the bigger risk in a high-side peak is running out of stock on your best sellers. A labor plan can flex in a week. A purchase order cannot. Check now which reorders can still land before Cyber Monday, November 30.
Spend this week on inventory. Recalculate weeks of cover on the SKUs that drive most of your orders, using October sell-through instead of September's. Ask your 3PL for its receiving cutoff for stock that must be live for Cyber Week, and book the dock date now. Then set ship-by dates with a buffer, since UPS is already warning of extra ground transit time for a limited number of packages from November 16 to January 8. (Darrigo Consulting, updated November 17, 2026)
The spike itself is the warehouse's job. We scale labor for holiday peaks, and orders are processed within 24 hours. From Canal Winchester, ground shipping reaches 90 percent of customers in two days. If the upside arrives as extra inventory, seasonal overflow storage holds it with no long-term commitment. How we handle spikes is laid out on our ecommerce fulfillment page.
If your peak plan still runs on September's first numbers, send it to us before Thanksgiving week. We will check it against what the warehouse needs.
FAQ
Was October's online jump just Amazon's Prime event?
Partly. Reuters said the 1.2 percent gain likely reflected Prime Big Deal Days. September came before that event and still printed 1.1 percent after revision. The three months through October ran 10.4 percent above last year, and one promotion does not explain a quarter.
Should I change my Cyber Week plan this late?
Change inventory and ship-by dates. Leave the offer alone. A new promotion in the middle of peak is hard to read in the data, and the next monthly Census read does not arrive until December 16.