Road speed and taxi ranks: checking the traffic where cabs wait and pick up
A taxi rank works when cabs can reach it and leave it quickly. How to read live road speed and the jam factor around candidate and existing ranks at the hours that matter.
Part of the guide Location intelligence automation
A taxi rank needs two things: people who want a cab nearby, and roads around it that let cabs arrive and pull away without sitting in a queue. Live road speed tells you the second part. Measure the current speed, the free-flow speed and the jam factor on the roads next to each rank at the hours when it is used, repeat for a few weeks, and you can see which ranks are choked by traffic and which candidate spots would flow.
This is for city transport departments, airport and station operators, taxi associations, and ride-hailing teams choosing pickup points. It explains what the traffic data can and cannot tell you about a rank, how to read it, a made-up example in Dubai, the usual mistakes, and an AI employee that measures every rank for you.
What the traffic data measures, and what it does not
Traffic Flow measures how traffic is moving on the roads around each place at the moment the flow runs. For every rank you give it, it returns the current speed in km/h, the speed when the roads are clear (free flow), the jam factor from 0 for free-flowing to 10 for standstill, the congestion as a percentage, a level in words, how many road stretches were measured, how many are closed, and the busiest road nearby. The figures are averages over the roads within the distance you choose: the roads next to the place (about 150 m), or every road within 500 m or 1 km.
It does not count cars, taxis or passengers, and it has no peak hours of its own. A run at 8 am tells you how 8 am looks today. To know the evening peak, run it in the evening; to know a pattern, run it at the same hours for several weeks and keep the rows.
How to read speed around a rank
- Congestion, not raw speed. A rank on a 30 km/h side street moving at 25 is fine. A rank on an arterial road whose free flow is 70 moving at 20 is jammed. Read congestion_pct, which compares the two.
- Jam factor at the exit. Cabs leaving a rank join the busiest road. If busiest_road shows a high jam factor at the rank's peak hours, every fare starts with a wait.
- Closures. A road stretch marked closed next to a rank, often roadworks, can cut it off for weeks. roads_closed counts them.
- The same hour, many days. One morning can be odd. Ten mornings show a habit.
A made-up example in Dubai
Invented figures, for illustration only. A taxi operator in Dubai runs ranks at twelve hotels and malls and is told that drivers avoid two of them in the evening. It sets an AI employee to measure the roads within 500 m of each rank every weekday at 08:00, 13:00 and 19:00 and save the rows to a sheet.
After three weeks, the averages are clear. Ten ranks show light traffic at all three hours. Two show a jam factor above 7 at 19:00, with congestion around two thirds; at one of them the busiest road is the slip road every cab uses to leave. The operator moves the evening rank at that mall to the other entrance, on a quieter road, and asks the mall to change the signs. The other jammed rank is near a closure that ends in a month, so it is left alone and watched.
To pick new rank sites, the same operator first lists clusters of hotels, malls and hospitals that are close to each other, using the taxi ranks template, then measures the roads at each cluster at its busy hours and keeps the ones that flow.
Mistakes to avoid
- Reading speed as demand. Fast roads do not mean people want cabs there. Speed tells you whether a rank can work, not whether it will be used.
- Measuring at the wrong hour. A rank by an arena matters at the end of an event. Schedule runs for the hours the rank is busy, not the office day.
- Too wide a radius. At 1 km the average mixes the motorway with the lane the cabs use. For a rank, the roads next to it or within 500 m say more.
- One week of data. School holidays, rain or a festival change traffic. Collect several weeks before moving a rank.
- Forgetting the walk. A rank on a free-flowing road three minutes' walk from the hotel door may lose passengers to cabs hailed at the door.
An AI employee that measures every rank
- Schedule: Several times a day, for example 08:00, 13:00 and 19:00, on the days ranks are busy.
- My data: your ranks and candidate sites, each with a position.
- Traffic Flow: Roads within 500 m (or adjacent roads for a rank on a side street).
- A Google Sheets step that adds one row per rank and run: time, speed_kmh, free_flow_kmh, jam_factor, congestion_pct, traffic_level, roads_closed and busiest_road.
- Filter on Jam factor 7 or more, then Email to the operations desk.
The sheet fills with a row per rank for every run, ready for a weekly average by hour. The operations desk only gets an email when a rank is jammed at the moment of the run: the rank, its jam factor and congestion, the busiest road, any closed stretches and the traffic level in words. Over a month the sheet shows which ranks are habitually blocked and at which hour.
Questions
Does Traffic Flow count the taxis at a rank?
No. It measures speed, free-flow speed, the jam factor and closed roads around each place. It never counts vehicles.
What is the jam factor?
A number from 0 to 10 for how jammed the roads are: 0 is free-flowing, 10 is standstill. Closed roads count as 10.
How do I see the evening peak?
Run the flow at the evening hours you care about. Traffic Flow measures the moment it runs, so the timetable decides which hours you see.
Which radius should I use for a rank?
The adjacent roads (about 150 m) or 500 m. A 1 km circle mixes in roads the cabs never use.
Can I compare candidate rank sites?
Yes. Put candidates and existing ranks in the same table and compare their averages at the same hours over the same weeks.