U.S. online retail sales rose 31.8% in the second quarter of 2020 from the first quarter, while total retail sales fell 3.9%, the Census Bureau reported on August 18. E-commerce reached $211.5 billion and 16.1% of retail sales, up from an 11.8% share in the first quarter. That 4.3-percentage-point gain combined two movements: more online sales and a smaller retail total. These figures are seasonally adjusted but not adjusted for price changes. [S1]

Holding the retail total at its first-quarter level makes the effect visible. Divide second-quarter online sales by the first-quarter retail total in the August release and the share is about 15.5%, rather than the reported 16.1%. This is an illustration using Census inputs, not another official estimate: it shows how the smaller denominator amplified the online share. E-commerce is already included in total retail sales, so adding it to the retail total would count those sales twice. [S1]

An analyst trying to reproduce the 31.8% growth rate also needs the revised starting point. The May 19 release had put first-quarter e-commerce at $160.333 billion. By August, Census had revised it to $160.414 billion, an $81 million increase. Using the older figure with the August second-quarter figure produces 31.9% growth after rounding, rather than 31.8%. The revision does not overturn the sharp increase, but mixing release vintages prevents a spreadsheet from matching the published comparison. [S1][S2]

The retail total excludes food services. Online travel services, financial brokers and dealers, and ticket sales agencies are also outside both the total and its e-commerce component. The denominator is therefore narrower than all consumer spending online and offline. Census also reported an unadjusted second-quarter online share of 15.1%. That figure and the adjusted 16.1% use different seasonal treatments; they do not describe different groups of shoppers. [S1]

The second-quarter estimate remains preliminary and subject to revision. Census attached a margin of error of 1.2 percentage points to the 31.8% quarterly sales increase, using a 90% confidence interval. That sampling margin is not a bound on every possible error: the report separately identifies nonresponse, incomplete coverage and processing mistakes as sources of nonsampling error. Nor does it measure how much the figure might change in a later release. [S1]

For a warehouse, sales dollars leave another gap: shipment volume. Census counts e-commerce sales where the order is placed, or the price and terms are negotiated, through an online system; payment need not happen online. Its retail universe includes businesses with and without e-commerce sales, not only internet-only firms. Capacity planning still needs order counts and parcel sizes, neither of which this dollar-sales report supplies. [S1]

Source documents

  1. Quarterly Retail E-Commerce Sales, 2nd Quarter 2020 (CB20-120)https://www2.census.gov/retail/releases/historical/ecomm/20q2.pdf
  2. Quarterly Retail E-Commerce Sales, 1st Quarter 2020 (CB20-71)https://www2.census.gov/retail/releases/historical/ecomm/20q1.pdf