A store dashboard mixes, on the same screen, numbers from very different origins. Some were counted. Some were estimated by a model. Some are projections. Some are hypothetical scenarios. When everything carries the same visual weight, the decision comes out wrong.
Afluxion's rule is simple: every indicator carries a seal that says where it came from.
Measured
Computed directly from observed data: entries counted at a line marked at the entrance, dwell by department, density on the floor plan. It has a stated time window, analysed population and calculation rule. When coverage falls below the minimum, it becomes partial.
Example: "Between 5 pm and 7 pm, the main produce camera measured 412 customer visits."
Inferred
The output of a model over the data, with stated confidence: the customer or employee classification, the bottleneck score combining density, speed and standing time, an anomaly against the expected pattern.
Example: "The model inferred congestion in the central aisle from 6:10 pm to 6:40 pm."
Predicted
A projection for a future period, compared with a recorded baseline and with a known error.
Example: "For Saturday, the model predicts a flow peak between 10 am and noon."
Simulated
A layout scenario applied to the Digital Twin, with assumptions and model version. It is never a measurement.
Example: "In the simulated scenario, the model projected a change in circulation in the beverages department, subject to the model's assumptions."
What never appears
A sentence like "moving department X increases sales by 12%" fits no seal. Without an experiment, there is no cause. What fits is "worth testing": an operational or layout change decided with data, measured before and after, at the same time of day and with the same population. That is how a pilot becomes evidence.