How it works
Screening, not forecasting
Urban Pathways does not forecast demand, deliberately; it answers a narrower question. It does not predict patronage and does not produce a business case. What it does is screen an entire road network and rank where a new or upgraded surface-transit corridor is most likely to be worth a detailed look — a screening and prioritisation step that sits upstream of network design, scheduling and detailed modelling, and feeds them a shortlist rather than replacing them.
The five factors
Every candidate corridor is scored against Surface Corridor Prioritisation, our five-factor method. Each factor carries a fixed headline weight:
| Factor | Weight | What it captures |
|---|---|---|
| Demand Potential | 30% | Where travel demand is concentrated but underserved |
| Transport Gap | 25% | Congestion and car-dependence relative to available transit |
| Travel-Time Competitiveness | 20% | How well transit could compete with driving on that corridor |
| Infrastructure Feasibility | 15% | Running-way and lane availability, depot and charging capital feasibility, grade and junction geometry |
| Regulatory Ease | 10% | How straightforward the corridor is to approve and deliver |
These five factors and their headline weights are all we publish. The sub-factor construction and internal scoring logic behind each factor are retained as trade secret; we do not disclose thresholds, coefficients, an elasticity model, or fare or cost parameters, in this document or anywhere else on the public site. That boundary is deliberate, not an oversight, and it holds across every page of this site.
Where the data comes from
Urban Pathways is built to run on open, widely available data rather than a bespoke survey: road-network graphs from OpenStreetMap, timetable and route data from GTFS, census and commuting-flow data, congestion data, and development-pipeline (planning-approval) records, sourced per market. This open-data foundation is also what keeps the method viable in markets where formal transport data has historically been thin — more than 10,000 transit agencies across 100-plus countries now publish GTFS, and the Mobility Database alone catalogues more than 6,000 GTFS, GTFS-Realtime and GBFS feeds across 99-plus countries (MobilityData, Mobility Database catalogue), while OpenStreetMap coverage keeps improving in previously under-mapped geographies — Srinagar’s mapped road network, for instance, grew from 1,154km in 2014 to 2,259km in 2022 (Springer Nature, OpenStreetMap road-network growth study), and global road-surface attribute coverage rose from roughly 33% to 36% between assessments (HeiGIT).
Validation against observed patronage
Demand catchment, the input behind Demand Potential, correlates with observed bus boardings at Spearman 0.31 (p<0.001), across roughly 339 Sydney statistical areas, checked against TfNSW Opal data from 2016. That correlation is moderate, not strong, which is why results sit at Level 2 rather than a higher tier: it checks the demand input the method is built on, not the ranking itself against realised patronage on the corridors it recommends.
Mode-aware by design
Urban Pathways does not just rank corridors — it selects among bus classes for each one: diesel bus, electric bus, articulated BRT, trackless tram or minibus/paratransit, favouring dedicated-right-of-way modes on congested, high-demand corridors and lighter modes where demand is thinner. Demand responds to the service structure being tested, across all twelve service-model architectures the screen evaluates per corridor, rather than staying fixed while that structure changes around it. Where several of those architectures share a hub, Urban Pathways also costs consolidating them into one branching trunk, attributed separately from the fleet decision on the same page. For the mode it selects, Urban Pathways also sizes the fleet and prices the diesel-to-electric decision — that scenario work has its own page: Fleet transition.
Confidence tiers
Every screening result carries an explicit confidence tier stating how much verification stands behind it — from an indicative first pass through to results checked against domain-panel review. The Sydney pilot, for example, sits at Level 2: indicative screening, with domain-panel verification still pending. We surface the tier next to every result, not in a footnote, because a transit authority should be able to see at a glance how much weight a number can bear. A tier moves only when new evidence justifies it — a correlation checked and held, a data source brought in and run against real inputs — not when code ships or a sprint ends.
A worked example: Sydney
An illustrative pilot run of Urban Pathways against Sydney’s road network shows the method above in practice. It is a demonstration of how the method surfaces candidate corridors, not a finished business case, and not a recommendation that any specific corridor be built.
Sydney has far more candidate transit corridors than any single planning cycle can fund to detailed design, and deciding which to study first has traditionally rested on planner judgement and route inertia rather than a systematic scan of the network. Screening Sydney’s network against the five factors above identified 211 activity centres and 4,780 candidate corridors, of which 4,184 were currently unserved by existing bus routes, on a screening run as at 28 August 2026. The corridors ranked highest were returned with their confidence tier attached, and with the mode best suited to each — favouring dedicated-right-of-way options such as trackless tram or articulated BRT where demand and congestion were both high.
Illustrative signals surfaced by the screen
- Macquarie Park ↔ Lindfield/Roseville — an orbital gap between two activity centres with no direct cross-town service today; the screen’s mode auto-selection flagged this as a trackless-tram-scale corridor.
- CBD ↔ Mascot — a corridor carrying a large commuting flow between the centre and an inner-suburban activity centre; the screen’s mode auto-selection flagged this as articulated-BRT scale.
- Surry Hills ↔ Bondi — an inner-eastern corridor the screen flagged at BRT scale.
- Bella Vista ↔ Schofields — a north-west growth-corridor pairing, reflecting the rapid activity growth in that part of Sydney.
Across the higher-scoring corridors generally, the screen’s mode auto-selection logic tended to favour dedicated-right-of-way modes — trackless tram and articulated BRT over standard bus — on corridors combining higher congestion with higher demand potential. That pattern is itself a screening signal about where road-based congestion may be eroding the case for mixed-traffic bus, not a conclusion about what any specific corridor needs.
Why this stays Level 2. The Sydney pilot has not been calibrated against observed patronage, ticketing data, or a funded transport model, and it does not replace four-step or activity-based demand modelling. It is a prioritisation screen: a way to triage a very large candidate set down to a shortlist worth modelling properly. Nothing here should be read as a patronage estimate, a business-case result, or a commitment to construct any corridor. The restraint is deliberate — this example exists to show what the tool screens for, not to tell Sydney what to build.
Checking a figure on this site
Every external statistic on this site carries its source next to it — World Bank, C40 Cities, Springer Nature, HeiGIT or MobilityData. Every Sydney pilot figure is dated to the screening run it came from, 28 August 2026. That is what a reader can check today.
What it hands off — and what it doesn’t do
Urban Pathways discovers and ranks corridors. It does not design routes, build timetables, roster crews, or run traffic-flow modelling — that work belongs to network-design, scheduling and detailed-modelling tools, and to the engineering consultancies and internal teams that run them. Urban Pathways’ job finishes where theirs begins: handing over a short, confidence-tiered list of corridors worth their time.