What Is a Flow Map?

A flow map shows movement, exchange or connection between places. Lines, arrows or bands connect origins and destinations, while width, colour, direction or animation can represent attributes such as volume, mode or time.

Examples include migration between regions, freight between ports, commuter journeys, airline connections and trade between countries.

The data behind a flow

At minimum, a flow record needs an origin and a destination, and it may also contain a magnitude, direction, category or timestamp.

An origin-destination table does not necessarily contain the actual path travelled. Drawing a curved or straight line between two cities may represent a relationship, not a recorded route.

That distinction should be clear whenever the geometry could be mistaken for physical movement along the displayed line.

Encoding magnitude and direction

Line width is commonly used for flow magnitude because thicker lines can represent larger volumes, while arrows or tapered bands can show direction and colour can distinguish modes, categories or opposing directions.

Avoid assigning too many variables to one line. A flow map can quickly become difficult to decode if width, colour, opacity, animation and arrow style all carry independent meanings.

Reducing clutter

Flow maps are especially vulnerable to overplotting because hundreds of lines can cross in the same central areas and obscure one another.

Possible strategies include filtering small flows, aggregating nearby origins and destinations, bundling similar paths, separating modes into multiple views, or allowing interactive selection.

The goal is not to preserve every record visually at all costs. It is to keep the relationships necessary for the map's question.

Route geometry vs connection geometry

A schematic line says “these places are connected”, whereas a network-derived path says “movement is represented along this route”. Those are different claims.

If the map shows actual road, river or flight trajectories, the source and routing method matter. If it shows only OD relationships, a geometric connection may be sufficient as long as the audience does not read it as a literal path.

Common problems

Large flows can hide small ones, opposing directions can overlap, long lines can dominate visually simply because they cross more of the map, and dense hubs can become unreadable.

A flow map can also imply causality or exchange where the data only record co-occurrence or inferred connection. The semantic meaning of a “flow” should therefore be stated clearly.

When should you use one?

Use a flow map when the relationship between places is central to the question—not merely the value at each place.

If only totals by location matter, proportional symbols may be clearer. If the actual network is the subject, a network map may be more appropriate. If the main goal is to compare origins and destinations numerically, a matrix or table can sometimes communicate the pattern more honestly than a dense web of lines.

Direction and net flow are different concepts

A map can show two directional flows between the same pair of places—100 trips from A to B and 70 from B to A—or it can show the net result of 30 from A to B. Those representations tell different stories.

Net flow simplifies the map and can reveal dominant direction, but it hides the volume moving in the opposite direction. Gross two-way flows preserve more information, although they can create overlapping symbols. The choice should follow whether the reader needs exchange volume or directional balance.

Time can be part of the flow

Flows often change by hour, day, season or year. Animation can reveal temporal dynamics, but it also makes comparison difficult because previous states disappear, so small multiples or filters may provide a more stable way to compare periods.

If time is essential, decide whether the audience needs to observe movement, compare snapshots or inspect exact values. The most visually dynamic representation is not automatically the most informative.