# The Stranded Report — FULL Methodology (v1.1.1) *This is the long-form reference version (internal + downloadable "full methodology" link). The short on-page version journalists actually read is `METHODOLOGY-onpage-short.md`.* **Campaign:** The Stranded Report — Britain's best and worst stations to be stuck at **Client:** Stasher · **Status:** v1.1.1, matches the published dataset (collected 6–13 July 2026, Aberdeen correction 14 July). Final version published on-page with the campaign. --- ## What this study measures When a train is delayed, some stations make the wait bearable — somewhere to sit, a cheap meal, a safe and functioning environment — and some don't. The Stranded Report ranks Britain's ~50 busiest transport hubs by a single **Station Survival Score (0–100)**: how survivable a delay there actually is. A higher score means a more bearable wait. Everything in the ranking comes from named public data sources, listed below with the periods they cover. No estimates, no scraped review sentiment, no paid data. ## Which stations are included The 50 hubs are selected by a stated rule, applied in order until the list reaches 50: **every London terminal in the official top 50 busiest stations; the airport stations (Heathrow, Gatwick, and Stansted, which qualifies on its own passenger numbers); every other station in the official top 50; then the busiest station in each of Britain's largest remaining cities and urban areas**, using the Office of Rail and Road (ORR) *Estimates of Station Usage* (April 2024–March 2025, published December 2025). A city can appear more than once where two of its stations are in the national top 50 (Glasgow and Liverpool both do). London commuter interchanges outside the terminals (Farringdon, Stratford and similar) are excluded by the rule, not by judgment. ## What goes into the score The score has two halves, published as their own league tables alongside the overall ranking. ### The "getting stuck" half — 35%. How likely your journey is to go wrong there. **Punctuality & cancellations — 20%** Measured **at each station**, not blended from operators: the share of trains arriving within 3 minutes and the share cancelled, from the ORR's station-level performance data (Table 3130, April 2025–March 2026, weighted by recorded stops). The ORR first published this station-level series in March 2025 and badges it "management information" rather than accredited statistics, with the latest periods provisional; we use the regulator's figures as-is and repeat those labels. **Passenger complaints — 15%** ORR complaints per 100,000 journeys (April 2025–March 2026). This data only exists per train company, so each station's figure is a blend of the companies serving it, weighted by each company's share of that station's services during the data period (companies under 3% of services, and any that began operating after the period, are excluded). This replaces review-mining approaches with the official record of passenger dissatisfaction. ### The "being stuck" half — 65%. How bearable the wait is when it does. **Facilities & comfort — 25%** From National Rail's official per-station data (retrieved 6 July 2026): toilets, accessible toilets, baby changing, Wi-Fi, waiting rooms, staffing hours, and step-free access category. Scored on a published point allocation (free toilets 3, waiting room 2, Wi-Fi 2, 24/7 staffing 2, step-free category A 1). **Food, drink & value — 25%** Coverage of four passenger needs, from the verified venue layer: quick coffee, a budget option, a sit-down venue and free rest space, each scored by whether it exists inside the station or within an 8-minute walk (70%), plus the share of the waking day (06:00-23:00) with at least one verified food venue open (30%). Outlet counts do not affect the score; they appear on station cards as attributed colour only ("N outlets on the official station map", or "at least N food businesses registered at this station" from the FHRS/FHIS register, matched by station postcode and address). **Crime rate — 15%** Recorded crimes at each station from **British Transport Police** open data (data.police.uk), covering **February 2024–January 2025** — the most recent 12 months published (BTP's data feed has an acknowledged pause after January 2025; see the data.police.uk changelog). Expressed as **crimes per million passenger entries and exits** (ORR usage data; CAA passenger statistics for Heathrow), so busy stations aren't punished for volume alone. Scottish stations are included: BTP still polices Scotland's railways. **What is and isn't counted:** figures are matched to the anonymised map point data.police.uk publishes for each station. This counts crime recorded at that location only — it excludes crime recorded at separate nearby points and crime recorded on trains, and will therefore read substantially lower than whole-station totals some BTP FOI responses report. The same definition was applied identically to all 50 stations, so relative rankings are directly comparable even though absolute counts are not comparable to FOI-sourced press figures. ## The Delay Survival Tool: band plans and waivers The tool's plans per delay band (30/60/120/180 min) fill named slots (quick coffee, budget food, sit-down, fresh air, main stop, itinerary) from the verified layers; usable time = delay − 15 min buffer − 2× walking (40 min buffer at airports). "Glance" items (monuments, viewpoints) never fill a primary slot. **Data contract: a slot that cannot be honestly filled is an absent key, never a null.** Free rest space scores via the station's recorded waiting room (`metrics.waiting_room`, from National Rail data) or a verified free-entry venue. Two documented human waivers apply: Bournemouth and Milton Keynes Central each have one genuine close dwell option rather than two; their plans are stay-put-weighted and their cards must not imply outing variety. Airport stations (Heathrow, Gatwick, Stansted) are exempt from stop requirements by design and carry stay-put plans. ## How the score is built Each metric is normalised across the full cohort (best station = 100, worst = 0 on that metric) and combined using the weights above. The exact formula ships inside the published dataset (`scoring.formula`) and every station carries its six sub-scores, so the overall score is recomputable from the public file alone: survival = 0.20×(0.7×within-3 + 0.3×cancellations) + 0.15×crime + 0.25×facilities + 0.25×food-and-drink coverage + 0.15×complaints. Punctuality combines within-3-minutes performance (70%) and cancellations (30%) within its 20% share. The full scored dataset, including every sub-score, is published alongside the campaign. A station can rank very differently on the two halves, and that is information, not a contradiction — both half-scores and their ranks are in the published dataset. ## The Delay Survival Tool The interactive tool on the campaign page uses an additional data layer: named venues inside or near each station (with opening hours from the venue's own site or OpenStreetMap, never guessed) and nearby places worth visiting during a longer delay, with route-based walking times. **Airport stations are treated differently by design:** there is genuinely nothing within walking distance of Heathrow, Gatwick or Stansted's terminals, so rather than pretend otherwise, the tool gives them a "stay put" plan built from landside amenities. Where a venue's hours could not be verified, the tool labels them as estimates. ## How we handle prices Pilot research showed that branch-specific prices at stations usually cannot be verified from published sources — so prices never drive the ranking. Where prices appear (in the Delay Survival Tool and editorial highlights), each carries a tier label, source, and date: - **Fixed national price** — chains with one advertised nationwide price (e.g. the Sainsbury's £3.95 meal deal, confirmed on sainsburys.co.uk and in trade press, June 2025). - **Official venue menu** — prices from a venue's own published menu for the specific station site (e.g. J D Wetherspoon publishes a free PDF menu per named pub, including pubs on station concourses). - **Published studies** — dated third-party investigations of station food prices, cited as their authors' findings, never as ours. - **Verified at this station** — a price individually confirmed at a named outlet with a dated, reproducible source. Delivery-platform menus are **not** used: the platforms' terms prohibit automated collection, and their listings routinely mislabel non-station branches as station ones. ## Quality control - Every datapoint carries a source URL and retrieval date, published in the dataset. - All figures were collected in a stated window (6–13 July 2026) and re-checked at publication freeze. - Independent verification recomputed every published number from the raw sources before release (final dataset: 2,600+ automated cell checks plus 10 raw-source spot-checks, zero discrepancies). - The top three and bottom three stations receive an additional human review before publication. - Known data quirks are disclosed rather than smoothed over — including the BTP publication pause (window stated above) and a geocoding fault in the source data for Cardiff Central (records are correctly labelled by station name, which is what we count; the faulty map coordinates are ignored). ## Fairness & right of reply The ranking measures institutions, not passengers or staff. Airport hubs are scored on their rail station perimeter, with airport-level context noted. Named operators and station managers are offered right of reply ahead of publication. Prices and facilities change; figures are correct as of the stated collection dates.