It designs your zones 6 ways, then shows you why it picked one.
RobustRoute plans balanced delivery zones and real-road routes for distribution and field teams. It evaluates a batch of designs, scores them against your constraints, and recommends one — and writes nothing until you apply it.
- 6
- designs evaluated
- 1
- parallel batch
- 0
- written without you
One request. A batch of candidates. A recommendation you can audit.
This is a real run against the Riyadh demo dataset — 50 stops, captured 2026-08-03. Same three views of it: what was considered, what the planner said, and how it scored.
6 designs · one parallel batch
· awaiting apply · nothing written
| # | Method | Balance score | Largest ÷ mean (≤ 2.00×) | Verdict |
|---|---|---|---|---|
| 01 | Polar sectors4 zones · 15 · 13 · 12 · 10 | 0.856 | 1.20× passes the fairness ceiling | ADOPTED |
| 02 | Constrained clustering6 zones · 11 · 9 · 9 · 7 · 7 · 7 | 0.821 | 1.32× passes the fairness ceiling | REJECTED |
| 03 | Constrained clustering4 zones · 15 · 14 · 13 · 8 | 0.785 | 1.20× passes the fairness ceiling | REJECTED |
| 04 | Polar sectors6 zones · 11 · 10 · 9 · 8 · 6 · 6 | 0.774 | 1.32× passes the fairness ceiling | REJECTED |
| 05 | Polar sectors8 zones · 9 · 8 · 7 · 6 · 6 · 5 · 5 · 4 | 0.750 | 1.44× passes the fairness ceiling | REJECTED |
| 06 | Constrained clustering8 zones · 9 · 7 · 7 · 7 · 6 · 6 · 5 · 3 | 0.738 | 1.44× passes the fairness ceiling | REJECTED |
Rank 01 · Polar sectors
Tightest spread in the batch — 15 down to 10, and every zone clears the ceiling.
Fairness ceiling: the largest zone may not exceed 2.00× the mean. Every candidate above clears it — ranks 02–06 are rejected on balance score alone. Real run, Riyadh, 50 stops, captured 2026-08-03. The record is kept with the run.
Different operations need different shapes.
Each method produces a characteristic geometry. The planner picks between them on your data — you are not committing to one at the start.
Constrained clustering
Natural groupings, balanced by workload.
Polar sectors
Wedges anchored on a depot.
Isochrone bands
Cut by drive time, not distance.
Balanced grid
Rotated rectangles aligned to the data.
Manual polygons
Draw the boundary yourself.
Fixed ERP format
Uniform cells where a downstream system demands them.
Every candidate is scored, checked against a ceiling, and kept on the record.
Scored, not guessed
Balance score is computed from the actual spread of stops per zone — for the adopted design, 15 · 13 · 12 · 10.
A ceiling, not a preference
The largest zone may not exceed 2.00× the mean. A design that breaches it is rejected regardless of how well it scores.
The record is kept
Every candidate, its score, its constraint check and the reason it lost stay attached to the run.
Plan the territory. Dispatch the day.
Co-equal products, one customer dataset. Use either on its own, or run both against the same data — no forced bundle, no extra integration step.
Strategic planningCommand Center
Build balanced delivery zones and save real-road route plans you dispatch from week after week.
- AI-assisted zoning
- Real-road routing
- Versioned plans
Daily operationsLast Mile Delivery
Drop today's orders, set your drivers, get optimized routes in seconds.
- Stops-based pricing
- Per-agent caps
- Full dispatch history
It proposes. You apply.
A run reads your data and produces a recommendation. Nothing is written to your zones, routes or agents until you press apply — so exploring a different shape costs you nothing but the time to look at it.
· awaiting apply · nothing written
Run it on your own data.
Upload a customer file and see the batch it produces. Start on the free tier — no card.