Questions to ask before trusting a retail statistic

Where did this number come from? It's the first question in evaluating retail statistics, and it's surprising how often it goes unasked. A figure appears in a presentation, a news story or an industry newsletter, and it gets repeated because it sounds authoritative. A few plain questions, asked every time, separate the figures worth using from the ones worth setting aside.

None of these questions requires technical training. They require the habit of asking.

What questions should you ask about any retail figure?

A short checklist covers most of what matters.

1. Who produced it? A data provider, a trade group, a survey firm, a brand? Each has different methods and, sometimes, different interests.
2. What does it measure? Dollars or units? Sales or shipments? Same-store or all stores? A figure described only as "sales grew" leaves most of the meaning out.
3. Which stores or shoppers does it cover? A national figure, a regional one, a single channel? Independent-store data and chain data describe different things.
4. What period does it cover, and what is it compared with? Month-over-month and year-over-year comparisons can point in opposite directions.
5. How large is the sample? A survey of a few hundred people and a dataset covering many stores carry very different levels of uncertainty.
6. Has it been revised, or might it be? Early estimates often change.
7. Does the conclusion go further than the data? A figure from one channel or region applied to all retail is a stretch.

If you can't answer most of these, treat the figure as unverified.

Why does the producer matter?

Because method and motive both shape a number. A data provider's store-level figures are a record of what sold in its stores. A survey's figures depend on who responded and how the questions were worded. A brand's figures may reflect a particular definition that flatters its results.

None of this makes any source bad. It means each figure should be read with its source in mind. Methodology notes, where they exist, are worth reading.

What are the most common problems?

Missing comparisons. "Sales reached a record high" means little without the prior period, the store basis and whether prices rose.

Blurred definitions. A category figure that includes different items from another source's category figure can't be compared.

Small samples presented as large conclusions. A result from a handful of stores or respondents can be real and still highly uncertain. Concepts such as statistical significance help judge whether a difference could be noise.

And quality problems upstream. A figure built from incomplete or duplicated data can be wrong in ways the author didn't notice. General principles of data quality apply to every statistic you encounter.

What should you do with a figure that doesn't hold up?

Don't repeat it, or repeat it with its limits. "One survey suggested X, based on a small sample" is honest. Stating X as fact isn't.

Look for a better source. Often a more careful figure exists that answers the same question with clearer methods. If it doesn't, say that the evidence is thin.

And hold your own figures to the same standard. Every report you publish should answer these questions for its readers. The NRS Insights monthly reports describe their basis, such as same-store comparisons across the NRS network of independent retailers, and the July 2026 same-store sales report is a recent example.