When to reroute based on traffic: jam factor thresholds and the cost of a detour
A detour is only worth it when the jam costs more minutes than the extra distance. How to set jam factor thresholds and work out the trade in plain arithmetic.
Part of the guide Location intelligence automation
Deciding when to reroute based on traffic comes down to one comparison: the minutes the usual road will cost at its current speed against the minutes the detour adds by being longer. Switch only when the jam costs more than the detour by a clear margin, because a longer road is a sure cost while a jam can clear in ten minutes. A jam factor threshold is the shortcut that tells you when that comparison is worth doing at all.
This is written for dispatchers, fleet supervisors and anyone who sends vans, couriers, technicians or cash vehicles along the same main roads every day. It covers what the jam factor means, a set of thresholds to start from, the arithmetic of a detour with a made-up example in Bangkok, the readings you should not act on, and how to have a flow watch your key roads and tell drivers only when a switch pays.
The jam factor in one paragraph
Traffic Flow measures, at the moment a flow runs, the roads around a point: the current speed, the free-flow speed (how fast the same roads move when they are empty), a jam factor from 0 for free to 10 for standstill or closed, the congestion as a percentage, a level in words and how many road stretches are closed. It describes speed only. It does not count cars and does not forecast, so every reading is a snapshot of now.
Thresholds to start from
Thresholds work best as bands, each with its own action, rather than a single trigger. These are a starting point to adjust after a few weeks of your own readings, not rules from anywhere else.
| Jam factor | Level | Suggested action |
|---|---|---|
| 0 to 3 | Free flowing or light | Keep the planned road. Nothing to work out. |
| 3 to 6 | Moderate | Keep it, unless the detour is short. Rush-hour slowness of this kind is usually already in planned times. |
| 6 to 8 | Heavy | Work out the trade (below). Switch if the detour wins by your margin. |
| 8 to 10 | Standstill | Switch unless the detour is also heavy. |
| Any, with roads closed | Closed | Switch, and warn anyone waiting at the far end. |
A starting set of bands; tune them to your vehicles and roads.
What a detour really costs
Travel time on a stretch is its length divided by its speed. That is all the arithmetic you need: minutes = km / speed in km/h x 60. Compare the usual road at its current speed with the detour at its current speed, then add the minutes you lose turning off and joining again.
Here is a made-up example in Bangkok, with invented numbers. A delivery van normally takes a 7 km stretch of an expressway to its depot. At free flow it moves at 60 km/h: 7 minutes. This morning the stretch reads a jam factor of 7.4 and a current speed of 14 km/h.
- Usual road now: 7 km at 14 km/h is 30 minutes.
- Detour: 11 km on a parallel road reading 32 km/h is about 21 minutes, plus 3 minutes for the extra turns: 24 minutes.
- Saving: 6 minutes. If your margin is 5 minutes, switch.
Now suppose the expressway reads 22 km/h instead. It costs 19 minutes, the detour still 24: stay put, even though the level says heavy. The band told you to look; the arithmetic told you not to move. That is why a threshold on its own leads to needless switching.
Readings not to act on
- Low confidence. Each reading carries a confidence from 0 to 1. A low one at night, measured from few vehicles, is weak ground for sending a driver five extra kilometres.
- A single point. One jammed junction does not mean the whole road is jammed. Read several points along the stretch and use the slowest sustained part, not one spike.
- Both roads jammed. When the alternative is heavy too, the detour adds distance for nothing. Compare both before telling anyone to switch.
- Planned times. Route planner times are travel times for usual traffic. They are a good plan for an ordinary day, not a reading of this morning, so do not use them as the current cost of the usual road.
Points along a road, not a road
Traffic Flow measures the roads around a place, so a corridor is watched as a row of points: a junction every few kilometres on the usual road, and the same on the alternative. Put them in a table with a name, the stretch each belongs to and its length. Choose the roads next to each point, about 150 m, so the reading stays on the road you mean and not a side street.
If your fleet runs in the United Arab Emirates, the Regional data block also holds road traffic history there: cars counted and their speed, hour by hour since January 2021, on the main roads near each place. That shows which hours a stretch usually slows, so you can set different bands for the morning and the afternoon. Elsewhere, build the same knowledge by keeping your own readings over the weeks.
A flow that only speaks when switching pays
- Schedule every 15 minutes, only during your driving hours, on weekdays.
- My data: the table of points on the usual road and on the alternative.
- Traffic Flow with Roads set to the roads next to each point.
- Calculated field: minutes for each point's stretch, from its length and speed_kmh.
- Data analyst or xMap LLM: add up each road's minutes and say whether the detour wins by your margin, or whether a road is closed.
- Branch: only the runs where switching wins go on, to a message in Slack or WhatsApp for the drivers on that corridor.
Adding Traffic incidents in the same flow names the cause when a stretch goes red: a closure, roadworks or an accident. Drivers trust a switch more when they know why. Most runs end silently at the Branch, which is the point: drivers hear from the flow only when a change saves them time. This kind of watch on your own roads is part of the AI employee for field operations work, alongside road closure alerts and route plans.
Questions
What jam factor should trigger a reroute?
Use it to decide when to check, not when to switch. Many teams start working out the detour from about 6, switch at standstill unless the alternative is jammed too, and always switch when a road is closed.
Does Traffic Flow predict how long a jam will last?
No. It measures speeds at the moment the flow runs. Run it often in your driving hours to see whether a jam is building or clearing.
Can I see how many cars are on the road?
Traffic Flow gives speeds, not counts. Car counts by hour exist only for roads in the United Arab Emirates, in the Regional data block's road traffic history.
How often should the check run?
Every 15 minutes during driving hours is enough for most fleets. Running it outside those hours spends searches on roads nobody is using.