The Census Bureau published advance retail sales for August at 8:30 this morning. Nonstore retailers, the line that carries most of ecommerce, came in at $141.3 billion seasonally adjusted. That is up 2.6 percent from July and up 9.9 percent from August 2025. (Census Bureau release CB26-153, September 16, 2026)
Total retail and food services sales were $773.9 billion, up 1.2 percent on the month and 6.0 percent on the year.
Read like that, it looks like demand accelerating straight into peak. Look at July before you staff on it.
The July dip is doing most of the work
July nonstore sales fell 1.7 percent from June. That drop is in the same release, two lines above the one everyone quotes.
Run the dollars instead of the percentages. June nonstore sales were $140.2 billion. July fell to $137.8 billion. August recovered to $141.3 billion. Across two months, online sales grew 0.8 percent.
That is the honest run rate going into Q4. The 2.6 percent print is one month climbing back out of the prior month's hole, and a forecast built on it will overstate your November by a wide margin.
The advance number is going to move
Advance estimates come from a subsample of roughly 4,800 firms, not the full survey. Census publishes how much that costs you in accuracy. For nonstore retailers, the median absolute revision to the month-over-month change is 0.4 percentage points.
So 2.6 percent is really somewhere around 2.2 to 3.0 percent, and the fuller number lands October 15. If your peak staffing plan changes depending on which end of that range is true, the plan is too tight.
Strip out gasoline and the gap widens
The 6.0 percent headline for all retail is flattered by one category. Sales at gasoline stations were up 21.0 percent from a year ago. Excluding gasoline stations, total retail grew 4.9 percent.
Against that 4.9 percent, ecommerce at 9.9 percent is growing at roughly twice the rate of everything else. The categories underneath tell you where:
- Miscellaneous store retailers, up 14.0 percent year over year
- Sporting goods, hobby, musical instrument and book stores, up 10.7 percent
- Electronics and appliance stores, up 7.8 percent
- Clothing and accessories, up 4.3 percent
- Department stores, up 1.9 percent
- Food and beverage stores, up 0.5 percent
One more caveat, and it matters. None of these figures are adjusted for price changes. August CPI ran 3.4 percent over 12 months. (BLS, September 11, 2026) Real growth in online sales is closer to 6 percent than to 10. Six percent more boxes is still six percent more boxes, but it is not the double-digit year the headline implies.
What this means for your brand
Nominal nonstore growth of 9.9 percent is running above the 7.5 to 8.4 percent that Deloitte forecast for holiday ecommerce last week. August is beating the season forecast. That is a reason to check your ceiling, not a reason to celebrate.
Here is the practical work for a brand shipping 1,000 to 25,000 orders a month.
Build your Q4 volume forecast off the June through August trend, not off August alone. Census puts nonstore sales for that three-month window up 10.2 percent against the same window last year, and that is a steadier base than any single print. Apply your own growth rate on top of the category rate. Then take the result and ask what happens if you are wrong by 30 percent in either direction.
Underbuilding is the more expensive error. A brand that under-forecasts eats late shipments, refund requests, and a January full of support tickets. A brand that over-forecasts pays for capacity it did not use, which stings but does not cost you customers.
That asymmetry is the argument for variable capacity. Our ecommerce fulfillment is built to absorb the swing, because orders get processed within 24 hours whether the day brings 200 or 2,000. You are not hiring seasonal pickers in October or signing a lease on space you will be sitting on in February.
Location does some of the work too. Canal Winchester puts 60 percent of the US and Canadian population inside 600 miles, and 90 percent of customers get their order in two days by ground. Short zones cost less per parcel, and they hold up better when a carrier network gets tight in December.
If you are not sure your current setup survives a month that runs 30 percent above plan, send us your last 12 months of order data and we will tell you where it breaks.
FAQ
Is "nonstore retailers" the same thing as ecommerce?
Close, but not identical. Nonstore retailers is NAICS category 454. It covers electronic shopping and mail-order houses, vending machine operators, and direct selling. Online sales are the large majority of it, which makes it the best monthly proxy available. Census publishes a cleaner ecommerce-only figure, but only once a quarter.
Should I change my Q4 plan because of one monthly report?
No. One advance print with a 0.4 point revision band is not a planning input on its own. Use it as a check against the forecast you already built. If August came in far above or below what your model expected, that gap is worth investigating. If it landed close, you learned your model works and you can stop refreshing the spreadsheet. Our FAQ covers how we handle capacity planning with clients heading into peak.