Most route operations are running territories that were drawn once, by someone who has left, using a paper map. They then spend years optimising sequences inside those boundaries — polishing the inside of a badly shaped box.
Territory design is the highest-leverage planning decision in a route business, and the least frequently revisited.
What territory design decides
A territory (or "district", or "master route") is a stable allocation of customers to a route and usually to a driver. Redrawing them changes:
- Workload balance — whether one driver finishes at 15:00 and another at 19:30.
- Drive-to-serve ratio — the proportion of the day spent driving rather than delivering.
- Driver familiarity — knowing the loading bay, the manager and the awkward gate code.
- Account ownership — who the customer calls, and who gets blamed.
- Growth headroom — whether new customers can be absorbed without a redesign.
Design to about 85% of the available shift. A territory planned to 100% has no room for a new customer, a traffic incident or a driver having a bad day — and will be re-planned in anger within three months.
The four objectives, and why they conflict
| Objective | Pulls towards | Conflicts with |
|---|---|---|
| Balanced workload | Equal hours per route | Compactness |
| Compactness | Tight, contiguous shapes | Balance when demand is uneven |
| Service consistency | Stable boundaries | Efficiency as demand shifts |
| Growth capacity | Deliberate slack | Short-term efficiency |
There is no plan that maximises all four. Deciding the priority order before you start is the actual work; the software just executes your choice. Most operations should rank: balance, then compactness, then consistency, then capacity — but a business built on relationship selling may well invert the first and third.
Getting the workload measure right
Territories balanced on stop count are almost always unbalanced in reality, because stops are not equal. Balance on estimated workload:
`` Workload = service time + travel time within territory + fixed daily overhead ``
Estimating service time properly is covered in service time estimation. For territory work you need the average and the variance: a territory with five 45-minute stops behaves very differently from one with thirty 8-minute stops even if the totals match.
Where visit frequency varies (some customers weekly, some monthly), balance on workload per cycle, not per day, then let the day-level planner distribute within the cycle.
The redesign process
1. Freeze a clean data set. Twelve months of stop history: location, date, service duration, volume, revenue. Clean the geocodes first.
2. Compute demand per customer. Visits per cycle, average service time, average volume. Flag the top 5% by workload — they usually need individual handling.
3. Choose your seeds. Either existing depot-adjacent anchor points, or algorithmically generated centroids. Seed count equals target route count.
4. Assign with constraints. Let the tool allocate customers to seeds, respecting capacity, workload ceiling and any hard rules (this customer must stay with this driver; this postcode is a different legal entity).
5. Review with the planners. Every automated territory plan contains three decisions that are technically correct and operationally stupid — a river crossing that looks short on the map, an industrial estate with one entrance, a customer whose manager will not deal with a new driver. Planners find these in an afternoon.
6. Model the transition. Which customers change driver? Who tells them? Sequence the change so no driver loses more than a third of their accounts at once.
7. Set a review cadence. Quarterly drift check, annual redesign. Territory quality decays continuously as customers open, close and change volume.
Measuring whether it worked
Track before and after:
- Hours per route, and the spread between the longest and shortest
- Drive time as a percentage of total shift
- Stops per hour
- Overtime hours per week
- Missed time windows
- Route count required for the same volume
The spread metric is the one to watch. A redesign that cuts the gap between the longest and shortest route from three hours to forty minutes has succeeded even if total distance is unchanged, because you have just recovered the capacity that was hiding in the short routes.
When not to redesign
- During a peak season. Obvious, routinely ignored.
- When your service time data is fiction. Fix the measurement first, or you will balance to imaginary workloads.
- Immediately after an acquisition before customer records are reconciled — you will design around duplicates.
- When the real problem is capacity. If every territory is over-full, redesign redistributes pain rather than removing it. You need another route.
Frequently asked questions
How often should territories be redesigned?
A light drift review every quarter and a full redesign annually suits most distribution businesses. High-churn operations — courier, some field service — benefit from continuous rebalancing instead, moving a handful of accounts each month rather than everything at once.
Should territories be geographic or account-based?
Geographic wins on efficiency; account-based wins on relationships. Businesses where the driver sells or upsells at the stop usually accept a 5–10% efficiency penalty to keep account continuity. Decide deliberately rather than letting the software decide for you.
How do I handle a customer that dominates a territory?
Treat very large accounts as dedicated routes or as a fixed anchor around which the rest is balanced. Averaging them into a normal territory produces a route that is either half-empty or hopelessly overloaded depending on the week.
Can I redesign territories in a spreadsheet?
Up to roughly 15 routes and a few hundred customers, yes, painfully. Beyond that the balance-versus-compactness trade-off has too many interactions to hold in your head, and you need a tool that can evaluate hundreds of alternatives.
What do I tell drivers whose routes change?
The truth, early, with the workload numbers. Territory changes are read as punishment when they arrive unexplained. Showing a driver that their route was 22% over the average, and that this is why they were always finishing late, converts most objections.
