
Walter Zoo
A family day with red pandas, animal encounters and room to explore.
01–31.10.2026 · Outward + return by public transport01 / Create the reason to travel
Catalogue snapshot · 2 October 2026 · conditions apply
One destination. A national opportunity to connect experiences and public transport.
Gossau’s planning report dated 31 October 2022, p.25: 145 zoo-owned parking spaces; external parking needed on about 205 days/year. Historical planning evidence, not current occupancy. Thurbo, checked 17 October 2026: 50% admission discount for travel 1–31 October 2026 with outward and return public transport, including eligible tickets and subscriptions. The discount is on admission, not a blanket rail discount. Map OD data are modelled average weekday trips in 2025 to the destination tile, all purposes, not zoo visitors. Routing is a saved 14 April 2025 comparison. PACE is the established time/boarding classifier. Transit geometry in the source connects recorded stops; track and road bends are unavailable. Walking geometry is saved network geometry. The gap between the tile endpoint and the attraction is not silently joined with an invented route.
Gossau · 2022 · p.25 ↗ · Thurbo · 17.10.2026 ↗02 / Build on proven value
Catalogue snapshot · 2 October 2026 · conditions apply
Prototype catalogue presence—not visitor share or RailAway’s complete inventory. Unmatched areas are not automatically commercial opportunities. RailAway 2025 ↗
The three measures have different definitions: public transport revenue impact, partner leisure revenue, and tourism customers. Customers are not website visitors or experimentally proven additional journeys. The reported increase of 0.7 million means 700,000 travellers, not CHF. Coverage is a boundary-inclusive point-in-polygon join of the 40 platform offer coordinates against the supplied 5,976 original hotspot polygons, counting each area once. Sixteen offers match sixteen distinct areas. The remaining offers are not necessarily inaccessible or irrelevant: addresses and hotspot boundaries are imperfect proxies. No visit volumes or demand weights are present in the supplied GeoJSON. The beta-0.95 file is a modelled selection of hotspot areas, not an exhaustive attraction register. The source and method are retained with the data.
Audit: the supplied beta-0.95 source has 5,976 polygons, not 5,836. Forty catalogue coordinates match sixteen distinct polygons. 40 / 5,976 is 0.67%, but it is not area coverage: locations and areas are different units. The 0.27% displayed here is 16 / 5,976. The dot matrix represents actual polygon identities, ordered with matched areas first. Geographic starts use polygon bounding-box centres for display only; the coverage calculation always uses the original polygon geometry.
03 / The acquisition opportunity
Equal-length bars compare channel shares. Worldwide visits; Swiss audience shares differ.
Digital reach and travel demand are separate opportunities. No conversion rate is inferred between them.
Semrush Traffic & Market · September 2026 · worldwide · all devices · external modelled estimates. Direct includes unattributed traffic; it does not establish travel intent.
A scenario to test, not a sales forecast.
365 assumes representative average daily demand.
1,050,000 × shift × days × retained SBB fare
additional rail trips / day
Revenue before displacement, acquisition, platform and incremental operating costs. No commission income assumed.
Loading the published map…
The handover CSVs and workbook agree: 156,520 estimated visits, 124,261 Direct, 11,218 Organic Search. All channel shares use the same September 2026 worldwide/all-device denominator. Swiss visit shares: Freizeit 86.91%, MySwitzerland 23.54%, SwissActivities 24.16%, Ausflugsziele 67.43%; no CH-only channel splits. Direct is not synonymous with SBB referrals or train-first intent. The app benchmark is 3.5m monthly active users, published by design partner Unic in January 2025. The 1.05m planning estimate is a rounded 1.5m Good/Great weekday car trips × an assumed 70% non-commuting share. Full-precision source 1,523,423.3525 × 70% = 1,066,396.3468. The 30% commuting deduction is an assumption, not an observed leisure filter. A 30-day extension is a scenario, not a measured month. The expandable revenue calculator is hypothetical and does not mix trips with website visits.
Unic · 01.2025 ↗04 / Earlier in the decision
Swiss residents aged 15–79 · YouGov, January 2026
freizeit.sbb.ch · Semrush · 16.10.2026 · Worldwide
450 citations · 285 cited pages
September 2026 traffic model. A modelled zero is not proof of zero referrals—or zero mentions.
Visibility and referral visits measure different things. The opportunity: make the actual leisure offer discoverable.
YouGov: 1,239 Swiss residents aged 15–79, fieldwork 14–20 January 2026, published 3 February. Activity suggestions (66%) concern AI travel-planning users, not the whole population. Semrush screenshot: Freizeit, worldwide, 16 October 2026, proprietary score 36, 68 mentions, 450 citations, 285 pages. Platform mentions are 5 ChatGPT, 18 AI Overview, 19 AI Mode and 26 Gemini, summing to 68. They are not counts of discounted-offer citations. The separate railaway.ch score of 17 is Switzerland-only and is not directly comparable. The September Traffic & Market model assigns zero AI-channel visits; small/unattributed values may be missed. Original screenshots, CSVs and workbook are retained in research/semrush-2026-10-17.
YouGov · 03.02.2026 ↗05 / An exploratory search experiment
realistic German-language searches, recorded once on 16 October 2026. We checked whether the returned sources included SBB or RailAway.
Exploratory sample · not Google rankings or an AI citation benchmarkOfficial sources appeared in 6 of 12 searches—including 1 of 6 inspiration searches. Offer relevance and unsolicited transport advice were not consistently captured.
Source: stored query/result archive. Underlying search engine and geographic personalisation were not controlled. Re-run the protocol before making a market-wide claim.
The twelve observations are recorded search-result sets, not twelve ChatGPT answers or matched causal experiments. There are six inspiration queries, three transport-intent queries and three deals queries. Source presence is measured by official sbb.ch/railaway.ch hosts, including subdomains, excluding proxies. No destination-first paired test was recorded; it is a recommended next experiment. Result titles and URLs are retained, but full assistant answers are not: attraction/discount/unsolicited-PT dimensions remain not assessed unless directly supported. The 201-entry colleague workbook is a separate generated catalogue with unknown prompt history; it is not pooled into this benchmark. Existing offer visibility is an opportunity to measure, not proof that RailAway is universally absent.
06 / Introducing SBB Freizeit
Catalogue snapshot · 2 October 2026 · conditions apply
A working local prototype using the 40-offer dated repository catalogue. Search, category filters, map markers and offer detail work without live AI or reservations. Expired entries are excluded from recommendations but retained in the catalogue evidence. National view is the default. Nearby ordering uses geographic distance, not a claimed journey time. Supported saved routing comparisons are shown as historical examples only. The separate application at /app.html is preserved. The SBB Freizeit name and sbb.ch/freizeit address are proposed branding; no SBB production integration or rights approval is asserted. Red #EB0000, white/neutral hierarchy and tab patterns follow the public SBB design references, using system fonts rather than an unlicensed brand font.
SBB · colours ↗ · SBB · brand ↗07 / 01 / SBB-owned channels
Catalogue snapshot · 2 October 2026 · conditions apply
A local SBB Mobile concept, adapted from Park_n_Ride_Potential/web/src/feeder/mockup.ts: status bar, SBB header, journey fields, connection cards and persistent icon/label tabbar. Explore, home selection, category filters, save/unsave and notification handoff work deterministically. Saving family offers changes the local recommendation order; no personal mobility records, login or backend is connected. A production pilot would need user permission, preference controls, opt-outs, notification frequency limits and agreed SBB placements. The sample notification uses the verified October 2026 admission offer.
SBB · Mobile tabbar ↗08 / 02 / AI & organic discovery

A family day with red pandas, animal encounters and room to explore.
01–31.10.2026 · Outward + return by public transportClear, verifiable content
Indexable pages. Accurate conditions. Crawlable links. Regular updates.
For travel in October, RailAway offers 50% off admission when you travel there and back by public transport. Check dates, eligibility and your connection.
Intended behaviour after publication—not an observed recommendation.Structured data helps interpretation. It does not guarantee indexing, AI inclusion or additional bookings.
The product and extracted facts use the same structured offer object, so dates and eligibility cannot diverge across the demonstration. The inspect panel exposes JSON-LD using TouristAttraction, Service and Offer. No invented numeric ticket price, live availability, ratings or endorsements are added. Prerendered DE/EN sample pages, metadata, internal links, sitemap and robots examples are exported under exports/discovery-proof for review; they are not new deployed product surfaces or evidence of indexing. Future publication requires approved URLs, crawl policy, freshness checks and supplier fulfilment. Google says ordinary SEO fundamentals apply to AI features; special AI files or markup do not guarantee inclusion. Measure offer-URL citations separately from brand mentions and referral visits.
Google Search Central ↗ · Schema.org · Offer ↗09 / 03 / Targeted social
Big eyes. A little adventure.sbb_freizeit A day to remember together. Meet the red pandas at Walter Zoo.
01–31.10.2026 · with outward + return public transport
Campaign concepts, not published posts or achieved results. Regional delivery would use agreed geographic audiences—not individual mobility histories.
Marketagent survey commissioned by Dertour Suisse, reported by TRAVEL INSIDE on 8 October 2026. 66% refers to surveyed social-media users; 54% refers to respondents already inspired by social media, not the full population. Self-reported influence, not causal conversion or year-on-year growth. Instagram and vertical video are designed concept interfaces with a verified destination image; there is no live campaign, influencer endorsement or claimed engagement. Click-through opens the same offer object as the website, app and assistant demonstration. The motion is an editorial pan of the supplied still photograph, not purported zoo video footage. Geographic targeting is proposed; no individual mobility-to-social matching.
TRAVEL INSIDE · 08.10.2026 ↗10 / The 42hacks mobility advantage
Catalogue snapshot · 2 October 2026 · conditions apply
Agree the offers, channel access, outcome reporting and a capped pilot budget.
Data: 2025 upscaled average-weekday directed OD trips, matched by exact origin/destination tile IDs to saved 14 April 2025 ordinary car/PT routing. These are destination-area trips of all purposes, not attraction visits or a forecast for next weekend. MIV = car + motorbike; total uses the source total field; PT sums train, bus, tram, u-bahn and ferryboat only when all components are present. PACE uses unrounded seconds and boardings, zero parking allowance. Good/Great PACE and at least 50% car share define an explicit screening rule; origins are ranked by existing car demand, not conversion probability. The criterion is editable in source and requires pilot validation. No numerical ROI or incentive response is inferred. Broad vs targeted changes campaign geography, not total destination demand. A smaller list is not itself evidence of higher campaign efficiency. Regional randomized holdouts, impressions/spend, redemptions, refunds, displacement and additional PT trips are needed to measure cost per incremental trip. Aggregated OD data are never linked to individual social profiles. The 56% population-coverage claim has no verified definition in the available evidence and is omitted. The source snapshot and read-only exporter are retained.