U.S. retail e-commerce sales reached an estimated $340.2 billion in the second quarter of 2026 after seasonal adjustment, according to the U.S. Census Bureau’s August 18 release. That was 3.8% higher than the first quarter and 12.2% higher than the second quarter of 2025. Total retail sales grew 2.9% quarter over quarter and 6.7% year over year.
For e-commerce operators, the headline is useful—but it is not a demand forecast for an individual store. The report is a national estimate built from a sample of retail firms. It covers many categories, uses preliminary data for the latest quarter, and is adjusted for seasonal patterns but not for price changes. The best use of the release is not to assume that every brand will grow 12.2%. It is to ask whether the fulfillment plan can handle a wider range of outcomes without sacrificing delivery promises, inventory accuracy, or cost control.
This article explains what the official figures show, how to read the methodology, and which operational tests are worth running before the next peak period.
Key takeaways
- Seasonally adjusted U.S. retail e-commerce sales were $340.2 billion in Q2 2026, up 3.8% from Q1 and 12.2% from Q2 2025.
- Total retail sales were $1,986.5 billion, up 2.9% quarter over quarter and 6.7% year over year.
- E-commerce accounted for 17.1% of total retail sales on the adjusted basis used in the headline release.
- The 12.2% e-commerce growth rate was 5.5 percentage points higher than total retail’s 6.7% rate. That is a calculation from the published figures, not a separate Census Bureau forecast.
- The latest quarter is preliminary, the figures are not adjusted for price changes, and the survey excludes some activities such as online travel services, financial brokers and dealers, and ticket sales agencies.
- Operators should use the data as a stress-testing signal: model order volume, SKU mix, inventory placement, carrier capacity, and returns under multiple scenarios.
The official Q2 2026 numbers
| Measure | Q2 2026 estimate | Quarter-over-quarter change | Year-over-year change |
|---|---|---|---|
| Retail e-commerce sales, seasonally adjusted | $340.2 billion | +3.8% (±0.4%) | +12.2% (±0.9%) |
| Total retail sales, seasonally adjusted | $1,986.5 billion | +2.9% (±0.2%) | +6.7% (±0.5%) |
| E-commerce share of total retail | 17.1% | — | — |
The Census Bureau also publishes figures that are not seasonally adjusted. On that basis, Q2 e-commerce sales were $329.5 billion, up 9.4% from Q1 and 12.4% from Q2 2025. E-commerce represented 16.4% of total retail sales on the not-adjusted basis.
Those two versions are not contradictory. Seasonal adjustment is designed to make changes across periods easier to interpret by accounting for recurring calendar patterns. The adjusted series is normally the better starting point for discussing underlying quarter-to-quarter direction. The not-adjusted series is closer to the raw timing of actual sales, which can be useful for understanding seasonal workload, but quarter-to-quarter comparisons can be distorted by predictable calendar effects.
Three insights—and the limits behind them
1. E-commerce grew faster than total retail
The year-over-year growth gap was 5.5 percentage points: 12.2% for e-commerce minus 6.7% for total retail. The adjusted e-commerce share also reached 17.1%, compared with a revised 16.3% in Q2 2025 in the publication table, an increase of 0.8 percentage points.
This supports a broad conclusion that online sales expanded faster than retail overall during the measured period. It does not reveal whether a specific merchant, product category, or marketplace grew at the same rate. The national estimate includes a wide mix of retailers, and the report does not break the quarterly headline into the exact SKU and customer segments a fulfillment team serves.
2. The latest estimate is preliminary
The Q2 2026 table marks the latest quarter as preliminary and earlier figures can be revised. Census explains that the estimate comes from the Monthly Retail Trade Survey and administrative records. Approximately 10,800 retail firms are selected using a stratified random sample and weighted to represent more than two million employer retail firms.
For any quarter, responding firms account for approximately 68% of the e-commerce sales estimate; missing responses are imputed using similar businesses or historical company performance. The monthly estimates are benchmarked to annual survey estimates and then summed into the quarter.
That is a serious statistical program, but it is still an estimate. A business case should preserve the release date and whether a number was preliminary or revised. It should not treat one reported decimal point as a precise prediction of future parcel demand.
3. Nominal sales are not the same as unit volume
The published estimates are adjusted for seasonal variation but not for price changes. In other words, the 12.2% year-over-year increase is a change in sales value, not necessarily a 12.2% increase in the number of products or parcels shipped. Changes in prices, product mix, basket size, digital services within retail coverage, and order consolidation can all affect revenue without moving parcel volume by the same percentage.
For fulfillment planning, units, orders, lines per order, cartons per order, and returns are more direct workload drivers than sales dollars alone. The Census data should trigger a capacity review, but the review must be completed with the operator’s own order-level history.
What the report covers—and what it excludes
Census defines e-commerce sales as sales of goods and services where the buyer places an order, or the price and terms are negotiated, through an internet, mobile, extranet, Electronic Data Interchange, email, or comparable online system. Payment does not have to occur online.
The estimates include businesses with paid employees. Since the April 2025 benchmark revision, nonemployer firms are no longer included and were removed from the time series to align it with the Annual Integrated Economic Survey. The retail universe includes retailers whether or not they sell online, but online travel services, financial brokers and dealers, and ticket sales agencies are not classified as retail and are excluded from both retail and retail e-commerce estimates.
These definitions matter when comparing this report with marketplace dashboards, private research, or a company’s own gross merchandise value. Different datasets can include different sellers, transactions, and sectors.
How fulfillment teams can use the data responsibly
Build scenarios, not one forecast
Start with the company’s own weekly orders and construct at least three volume cases. A practical structure is a base case based on recent run rate, a campaign case reflecting planned promotions, and a stress case that combines stronger demand with a delay or capacity constraint. The national 12.2% figure can inform the range, but it should not automatically become the company’s growth assumption.
For each case, translate sales expectations into operational units:
- orders per day and per hour;
- units and order lines per day;
- cartons per order;
- share of single-SKU and multi-SKU orders;
- expected cancellations and address changes;
- return requests and units received back;
- carrier volume by service and destination region.
Test inventory by velocity and location
Revenue growth can hide very different SKU behavior. A few products may create most of the volume, while slower inventory still consumes space. Classify SKUs by order frequency, cube, handling requirements, and margin. Then test whether fast movers are positioned close to packing stations and whether replenishment rules prevent pick-face stockouts during campaign spikes.
For multi-location operations, compare the cost of inventory duplication with the cost and service impact of split shipments or long-zone deliveries. The correct decision depends on order geography, product value, storage cost, and replenishment reliability; there is no universal rule that every SKU should be placed in every location.
Convert parcel growth into labor and cutoff decisions
If volume rises, the bottleneck may appear at receiving, putaway, picking, packing, quality control, manifesting, or carrier handoff. Measure each step separately. An average daily capacity number can be misleading if most orders arrive after a promotion email or marketplace cutoff.
Useful tests include orders per labor hour, pick lines per hour, packing time by carton type, same-day cutoff attainment, dock throughput, and unresolved exceptions at end of shift. Tie staffing decisions to the hourly arrival curve, not only the weekly total.
Stress-test carrier capacity and package economics
Q2 sales data does not reveal how carriers will perform in peak weeks. Operators should still review daily pickup limits, trailer or vehicle capacity, service eligibility, holiday surcharges, dimensional-weight exposure, and backup routing rules. A volume plan is incomplete if it assumes every parcel can move through the preferred service at the modeled rate.
Make exception logic explicit: which orders may move to an alternate service, which cannot be split, which require signature or special handling, and who approves a delivery-promise change. Test labels and manifests before peak rather than waiting for a carrier or integration failure.
Plan the reverse flow
More shipped orders can create more returns even when the return rate is stable. Model return units, not only the percentage. Reserve receiving space and decide how quickly returned stock must be inspected, graded, restocked, quarantined, or disposed of. The customer-facing return promise should match the actual processing capacity.
A practical readiness scorecard
| Planning area | Question to answer | Evidence to use |
|---|---|---|
| Demand | What are the base, campaign, and stress order volumes by day? | Store history, promotion calendar, marketplace events |
| Inventory | Which SKUs drive orders, cube, and replenishment work? | SKU velocity, stockout history, storage map |
| Labor | Where does throughput fail by hour or process step? | Scan timestamps, labor hours, exception backlog |
| Packaging | Which cartons create dimensional or handling exposure? | Packed dimensions, billable weight, damage history |
| Carrier | What happens when the preferred service hits a limit? | Pickup plan, service rules, surcharge schedule, backup routes |
| Returns | Can the reverse flow absorb the projected unit count? | Return units, inspection time, disposition rules |
| Customer promise | Which cutoff and delivery messages remain achievable? | Order-to-ship time, carrier transit data, exception rates |
The scorecard should have an owner, a measurement date, and a threshold that triggers action. “Monitor capacity” is not enough. A better rule is: if forecasted orders exceed tested packing capacity for two consecutive days, activate a defined staffing, cutoff, or routing response.
What the Census release does not tell you
- It does not forecast Q3, Q4, Black Friday, or an individual merchant’s growth.
- It does not measure parcel count, units shipped, or fulfillment labor.
- It does not isolate the effect of price changes because the estimates are not price-adjusted.
- It does not provide category-level results in the quarterly headline sufficient to plan a specific SKU mix.
- It does not guarantee that reported preliminary figures will remain unchanged.
- It does not replace carrier capacity, inventory, labor, or returns data from the operator’s own network.
The next quarterly retail e-commerce release is scheduled for November 19, 2026. Teams can use that update to refresh the external benchmark, but they should monitor their own daily signals much more frequently during peak.
Frequently asked questions
How much did U.S. e-commerce grow in Q2 2026?
Seasonally adjusted retail e-commerce sales increased 12.2% from Q2 2025 and 3.8% from Q1 2026, according to the Census Bureau’s preliminary estimate.
What share of U.S. retail sales was online?
E-commerce accounted for 17.1% of total retail sales on a seasonally adjusted basis in Q2 2026. On a not-adjusted basis, the share was 16.4%.
Does 12.2% growth mean parcel volume grew 12.2%?
No. The report measures sales value and is not adjusted for price changes. Parcel volume also depends on prices, product mix, units per order, split shipments, and other factors.
Should a merchant use 12.2% as its peak forecast?
Not automatically. Use the national figure as external context, then build company-specific scenarios from recent orders, promotions, inventory, conversion, and operational constraints.
Is the Q2 2026 estimate final?
No. The latest quarter is preliminary and subject to revision. The report labels earlier revised figures and publishes updated time series.
Turn a national signal into an operating plan
The value of the Q2 release is not the headline alone. It is the reminder that e-commerce growth and total retail growth can move at different speeds, while sales dollars and fulfillment workload can also diverge. The operational response is to connect external context with internal evidence: orders, units, cartons, labor, inventory, carrier constraints, and returns.
If you are reviewing fulfillment readiness for the next demand cycle, Speedfulfill can discuss the assumptions behind your order profile and help identify the questions that deserve testing. Final decisions should be based on your current operational data, service requirements, and verified carrier terms.
