A dollar gain is still a gain
Census estimates September retail and food-service sales at $745.0 billion. The 0.8% monthly rise has a sampling-error margin of ±0.4 percentage points at the 90% confidence level. That interval does not include zero; uncertainty is not a reason to call the reported rise imaginary. Other errors, including nonresponse and reporting mistakes, are not captured by that margin. These are advance estimates, subject to revision. (Sources 1)
August was revised, not repeated
The same release revises August's monthly increase from 1.1% to 1.3%. September's 0.8% is a smaller reported increase, not a decline in the dollar level of sales. Both changes use the release's seasonal and calendar adjustments; neither removes price changes. (Sources 1)
The price index uses a different basket
BLS reports a seasonally adjusted 0.6% September increase in consumer prices. That is useful context, not a ready-made deflator for this retail total: the CPI includes shelter and other spending outside retail and food services. Subtracting 0.6 from 0.8 would not establish real retail growth or what a household could afford. (Sources 1–2)
Ask what the increase measures
The release establishes an increase in estimated dollar sales, not how purchases were distributed among households or how much more they received. A seller's receipts and a family's buying power are related questions, not the same measurement. (Sources 1)
What this record cannot settle
These advance dollar-sales estimates do not measure inflation-adjusted quantities, every household's spending or welfare, or the effect of a particular policy. The sampling margin does not cover every source of error. No investment or individual financial recommendation is made.