Route overlap: cutting miles nobody needed to drive

Route overlap is the quiet line item on your budget. Two morning buses trace the same collector streets ten minutes apart. A deadhead from the yard mirrors another bus’s outbound leg. A tier was copied from last year’s spreadsheet, so you are still running shadow coverage in a neighborhood that could be one vehicle if the windows lined up.

None of that shows up as a bold expense. It shows up as fuel, hours, and wear spread across the whole year — plus the opportunity cost of buses you think you need because the map still looks “full.”

If you want the companion topic — moving seats instead of just miles — read empty seat miles. For how software looks at the whole network instead of one route at a time, see manual vs algorithmic route planning.

What counts as overlap

Think of overlap as duplicate coverage: same road segment, same kind of work, not meaningfully more passengers served. Common shapes:

• Parallel pickups in one time band — two buses nibbling the same cluster.
• Legacy templates copied forward so routes “rhyme” geographically without anyone asking if one bus could do both.
• Deadhead that stacks — multiple vehicles repositioning through the same corridor empty.

Overlap is not the same as two buses on a main road at different times; context matters. The damage case is redundant capacity in the same window, especially when ridership on those legs is thin.

How to see it without guessing

You need something better than eyeballing PDF maps. A practical minimum is recent GPS (a few weeks to a couple of months) plus which run each trace belonged to. Manifests or boarding counts help you decide whether overlap is carrying riders or just burning miles.

A simple workflow: group traces by route and day, snap to segments, build a matrix of how often routes share the same links in the same part of the morning. Flag the worst corridors, then cross-check ridership. High overlap and weak loads is usually your first cut list — not the route everyone depends on for coverage.

If your data is messy, you are not alone. Most programs start with exports plus a short GPS sample; cleanup is part of the job, not a reason to postpone measurement.

Why overlap survives so long

Bell times and tiers that were never harmonized create parallel waves through the same streets. Stop policy — a stop per driveway era — leaves you with more legs than you need. Manual spreadsheets encourage copying a leg from last year instead of re-solving the network. Driver habit can reintroduce parallel travel after a optimize-and-print exercise if nobody reconciles actual paths with plan.

Driver scheduling matters too: if shifts are built before routes are consolidated, you can “save miles” on paper and still pay for two drivers to do redundant work. Driver shift scheduling ties into that.

Fixing it without wrecking service

Consolidate stops where walking distance and safety policy allow — and communicate early. We have a full walkthrough of rules and politics in stop consolidation.

Nudge pickup windows slightly so one vehicle can thread adjacent clusters without inventing a twenty-minute delay. Pickup window optimization is the long version.

Rethink tiers so neighborhoods are not served twice because of schedule shape. Tiered school bus routing covers the tradeoffs.

Shrink deadhead by yard assignment and staging — overlap often lives in the first and last miles.

Mixed school + shuttle operations should optimize together when vehicles and rules allow; otherwise you optimize two silos and wonder why miles do not fall. See mixed fleet routing.

Every consolidation needs a ride-time and equity check. Saving miles by pushing walks onto the same families every time is a different problem than efficiency — call it what it is and design around it. When you shrink fleet coverage, keep contingency plans honest so a breakdown does not erase the savings in one bad morning.

A ninety-day shape (not a religion)

Weeks one and two: pull GPS and schedules, pick three worst corridors, write down baseline miles and complaints. Weeks three to six: pilot one zone — consolidate or reassign, communicate, watch ride time and calls. Weeks seven to ten: roll what worked to similar geography; update shifts and driver briefings. Weeks eleven and twelve: remeasure miles, document what changed, leave spare capacity for weather and breakdowns.

Rough ROI is just arithmetic once you have baseline miles and a cost-per-mile you believe: if a slice of your annual miles is redundant and you cut that slice in half, the fuel and maintenance line moves — labor may follow, but contracts and hiring plans deserve their own conversation, not a blog footnote.

Parents usually accept small walk increases when the message is specific — which stop, which dates, how pickup time moves — and when you leave a channel for edge cases (safety, disability, genuine hardship). “We optimized” is a bad email; “here is your new corner and why” is a better one.

Bottom line

Overlap is a measurement problem first. Once you can name the corridors, you can merge work — stops, windows, tiers, deadhead — instead of hoping a sharper highlighter fixes next year’s budget.

If you want to see network-level optimization on real data, the live demo is the quickest path. For setup and imports, our docs and help center collect the practical stuff in one place.

Related reading: Dwell time optimization · Fleet route optimization — 60-day playbook · Route contingency planning

— Emrah G.