How to measure congestion around a store, branch by branch
Compare the speed on the roads beside each branch with their free-flow speed, at the hours customers arrive, and turn the readings into a fair table of every store.
Part of the guide Location intelligence automation
To measure congestion around a store, take the current speed on the roads beside it and divide it by the speed those same roads reach when they are empty. The further the ratio falls below one, the more congested the store's approach is. Do this at the hours your customers actually arrive, repeat it over a few weeks, and you have a figure you can compare across every branch.
This guide is for retail and restaurant chains, pharmacy and bank networks, car dealers and anyone with branches that customers drive to. It walks through choosing which roads count, the arithmetic, how to compare branches fairly, a made-up example and the mistakes, and then sets it up as an AI employee.
Which roads count as around the store
Congestion only matters on the roads a customer uses to reach the door or the car park. For most high-street shops that is the street in front and its nearest junctions. For a retail park or a big-box store it can be every road within half a kilometre.
The Traffic Flow block offers three distances: the roads adjacent to the place (about 150 metres), every road within 500 m, or within 1 km. Pick one per type of branch and keep it. Changing the distance halfway through changes the numbers, and the comparison breaks.
The arithmetic, in plain terms
Traffic Flow returns the current speed and the free-flow speed for the roads it measured, already averaged and weighted by the length of each stretch. From those you can work out the ratio yourself, or use the congestion percentage it gives: how much slower than free flow, never below zero.
- Current 24 km/h, free flow 40 km/h: ratio 0.6, congestion 40%.
- Current 38 km/h, free flow 40 km/h: ratio 0.95, congestion 5%.
- Current 12 km/h, free flow 50 km/h: ratio 0.24, congestion 76%.
The jam factor, from 0 to 10, says the same thing on a fixed scale, and the traffic level puts it in words from Free flowing to Standstill. For a table that store managers will read, the percentage and the words are the easiest to follow.
Comparing branches fairly
One reading tells you little. A branch might look jammed because a bus broke down that minute. Three habits make the comparison fair.
- Fix the hours. Choose two or three hours when customers arrive, such as late morning, lunchtime and early evening, and read every branch at the same hours.
- Collect enough days. Three or four weeks of weekday readings smooth out one-off events. Keep weekends separate.
- Use the median, not the worst day. Sort each branch's readings for an hour and take the middle one. It resists the odd bad afternoon.
- Check confidence. Each reading carries a confidence from 0 to 1. Drop readings with very low confidence before averaging.
- Note closed roads. A reading with a closed road nearby belongs in a separate column; it says something different from everyday congestion.
A made-up example: a pharmacy network in Lyon
The numbers below are invented for illustration. A pharmacy group runs nine branches in and around Lyon and wants to know why prescription collections at two branches drop on weekday evenings.
| Branch | Median congestion at 18:00 | Usual level | Busiest road |
|---|---|---|---|
| Presqu'ile | 22% | Light | a central avenue |
| Part-Dieu | 58% | Heavy | a ring road junction |
| Villeurbanne | 18% | Light | a local street |
| Gerland | 51% | Heavy | a riverside road |
Invented figures for four of the nine branches, four weeks of weekday readings at 18:00, every road within 500 m.
The two branches with falling evening collections are the two with Heavy congestion at 18:00. The group cannot change the roads, so it extends the online order-and-collect window at those branches into the late morning, when the same roads read Light, and promotes it to customers who usually come in the evening.
Mistakes to avoid
- Comparing speeds instead of ratios. A branch on a fast road will always show higher speeds. Compare against each branch's own free flow.
- Reading once and deciding. A single snapshot can be an accident or a delivery truck. Collect weeks, not minutes.
- Treating congestion as footfall. Slow roads do not mean more shoppers, and fast roads do not mean fewer. For presence around a store, Mobility Footfall measures people-hours, which is a different question.
- Expecting car counts. Speed data does not count cars. Only UAE roads have hourly counts, through the Road traffic history dataset.
- Mixing distances. Keep the same road distance for every branch in a group.
Setting it up as an AI employee
An AI employee can take the readings, keep them and send the weekly table, without anyone opening a spreadsheet by hand. It is a good example of what location intelligence automation looks like in a branch network.
- Schedule: weekdays at your chosen hours.
- My data: your branches, from a table with their addresses.
- Traffic Flow: within 500 m (or adjacent roads for town-centre shops).
- Sort and limit: highest congestion first.
- A Google Sheets app block adds one row per branch per reading, building the history.
A second flow, scheduled every Monday, can read the sheet's last four weeks and have xMap LLM write a short summary of which branches got better or worse, sent by Email to the area managers. The templates Road speeds outside my depots and Score every neighbourhood for a new branch are good starting points to adapt.
Questions
What is a good congestion figure for a store?
There is no universal line. Compare each branch with the others in the same group at the same hour; a branch far above the rest at the hours customers arrive is the one to look at.
Does congestion mean more customers walk past?
Not necessarily. Congestion describes how traffic moves. For how busy the area is with people, use footfall, measured in people-hours of presence.
How many branches can I measure at once?
One run measures up to 50 places, and places within 300 m of each other share a reading. Larger networks can be split across several flows.
Can I see which road is causing the slowdown?
Each reading names the busiest road, the one with the highest jam factor, and counts how many road stretches are closed.