How we work out when a place is busy

Our busyness chart is modeled, not measured. Google does not sell live or historical crowd data through any public API, so nobody showing you a popular-times chart for free is reading a sensor — they are either scraping, paying a third party, or modeling it. We model it, and we say so on every chart.

Where does the curve come from?

Three inputs, none of them a measurement:

1. A shape for the kind of place. A bar, a coffee shop and a bowling alley fill up at different hours in a way that holds across cities. Each venue category carries its own hourly curve and its own weekly shape.
2. How well known the place is. Review volume sets the ceiling of the curve, so a place with four thousand reviews reads busier at peak than one with forty. Without this every venue saturates at the top of the chart and the shape stops carrying information.
3. Its actual opening hours, so the curve does not claim a crowd at an hour the doors are locked.

Can I get real numbers?

Yes, and the app is built to take them. Two commercial providers sell genuine forecasts and live readings, and the busyness layer is a swappable seam — set one key and every chart switches from modeled to measured, with the confidence label changing to match. We run the modeled version by default because it is free and honest about being a model, not because it is as good.

Why show a model at all?

Because the question it answers is mostly about shape, not precision. “Is nine o’clock better than seven here” is a question a category curve answers well. “How many people are inside right now” is one it cannot answer at all, and the chart does not pretend to.