What to remember
- Define mass, distance and the friction exponent.
- Use Reilly for a two-centre breaking point.
- Use Huff for multi-centre attraction probabilities.
- Validate with visits, mobility, competition and local constraints.
Calculate Reilly’s breaking point
For centres A and B separated by D, the distance from A is D ÷ (1 + (mass B ÷ mass A)^(1/n)). With n = 2, A = 400,000, B = 100,000 and D = 60 km, the point is 40 km from A and 20 km from B.
The 50/50 boundary is theoretical and only reflects the declared inputs.
| Input | A | B | Result |
|---|---|---|---|
| Mass | 400,000 | 100,000 | Ratio 0.25 |
| Total distance | 60 km | - | 60 km |
| Exponent | 2 | 2 | Square root |
| Breaking point | 40 km from A | 20 km from B | Theoretical boundary |
Move to the probabilistic Huff model
For origin i and centre j, utility is mass_j ÷ distance_ij^n. Probability is that utility divided by the sum across accessible centres.
With A of mass 10,000 at 5 km and B of mass 20,000 at 10 km, n = 2 gives utilities 400 and 200, or about 66.7% for A and 33.3% for B.
Read a modelled area rather than ground truth
Huff probability depends on attractiveness, distance, included competitors and exponent. It is neither observed visits nor a market-share forecast.
Travel time, road networks, barriers, income, assortments and habits require separate data and calibration.
Bound the connection to Plot
Plot locally contains a Retail Gravitation tool with Reilly, Huff, 41 corridor positions and a configurable exponent. It is locally tested and not deployed as a public runtime.
It is not a GIS: no polygons, isochrones or automatic site optimisation. Map handles operational flows and risks and does not own this method.
Frequently asked questions
Do Reilly and Huff produce the same trade area?
No. Reilly gives a two-centre breaking point; Huff distributes probability across multiple centres.
Can travel time be used?
It can in an adapted model, but the documented Plot calculation uses corridor distance and does not prove a routing engine.
Does the model choose the best location?
No. It explores assumptions; field evidence, site economics and human judgment remain necessary.
Full sources
- REAT manual, Reilly and Huff formulas and examples, complete PDFComplete source available online · Read the complete source
- Dynamic Huff model and store choice, complete scientific article PDFComplete source available online · Read the complete source
- Huff model and trade area using mobility data, PDFComplete source available online · Read the complete source
- Insee, small trade areas and living areas, PDFComplete source available online · Read the complete source