An estimated average order value of "€62" and an estimated band of "€38 to €95" are not the same kind of information. The first looks like a measurement. The second tells you how little you know.
Thesis
An estimate of revenue or sales velocity should travel with its confidence interval. BoostEcom's model computes one for each of those estimates and stores it next to the figure. Not every screen shows it yet, and this post names the one that does not.
Facts we stand on
- Outside inferences are labelled as estimates. The sales-velocity model stores each monthly revenue and unit estimate with a low bound, a high bound and a confidence score.
- Before calibration the band is wide on purpose: half the figure on either side. It tightens as the model is checked against the real numbers of connected stores that opted into the calibration program.
- The Hub's store list shows the central revenue figure only, with a note that calls it directional. The interval exists in the record; that screen does not print it yet.
- Confidence also gates how loudly a signal surfaces: strong anomalies alert harder than weak ones.
What this does not prove
An interval does not make the estimate true. It makes the uncertainty visible. A wide band is not a failure of the product; it is often the only honest output when the public signal is thin.
Mechanism
Inference from public catalog, pricing, and velocity cues is under-determined. Collapsing that into one point estimate hides the model. Showing the band forces the reader to decide with the same information we had, not with a false sense of precision. A screen that prints only the central figure loses that, even when the band exists underneath.
Implications for operators
Operators act on numbers. A false-but-confident figure produces a false-but-confident decision. Ask every competitive tool for the interval, ours included. If there is none on the screen, ask where the figure comes from. If nobody can answer, you are borrowing someone else's bet.
Action this week
Open any spy or intelligence product you use. For one store, write down the revenue or AOV figure it shows. Next to it, write the interval (or "none"). Keep only the tools that survive that line.
Sources and method
Store Intelligence;/intelligence/transparency.- Verification date: 2026-09-25. Signed by Christopher Lasgi.