# Airport-to-stay transport study — protocol v1

Prepared: 7 September 2026. Status: research design only; no observations collected in this release.
Publisher: TravelShifu. Contact: hello@travelshifu.com.
Reuse: this original protocol and blank template may be reused with attribution to TravelShifu and a link to https://travelshifu.com/evidence/#research-protocol. No third-party data or images are licensed by this permission.

## Research question

How do complete, comparable door-to-door journeys differ in total paid cost and elapsed time once access, waiting and last-mile travel are included?

Initial scope: a Jakarta-to-Bandung journey with a clearly specified origin and destination, and Bali airport-to-stay transfers. These are separate route strata, not one mixed league table. Global extensions require a new, published route register.

## Register the sample before collecting

For each exact origin/destination pair, register at least three distinct travel dates including a weekday and a weekend, two departure windows, and the modes actually available. Record one-person and two-person quote scenarios where sold. These are design targets, not a claim of statistical representativeness. Publish missing cells and why they could not be collected. Do not replace inconvenient observations silently.

Fix baggage, eligibility, refundability, party size, booking lead time and pickup/drop-off points within a comparison. Record requested travel date/time and quote-retrieval timestamp with IANA time zone. Compare like-for-like quotes retrieved in the same collection window. No automated scraping that breaches supplier access restrictions.

## Record the complete journey

Separate base fare, mandatory booking/tax fees, baggage, access transport, last-mile transport and other unavoidable charges. Record refundable deposits separately. Preserve original currency; publish any conversion source and timestamp, or report currencies separately.

Separate access, waiting, main transport and last-mile elapsed minutes. Tag each timing as observed, operator-stated, routing estimate or unknown. Quoted journey duration is not a measured journey. Record major disruptions and weather only where directly documented.

Keep the exact source URL and a private dated evidence reference. Never publish another person's ID, contact details, bank data or unredacted booking reference. Obtain consent before using identifiable original photographs. Store evidence with access controls and a documented retention decision.

## Calculate and publish

Cost = base fare + mandatory fees + baggage + access + last mile + other unavoidable costs. Refundable hold is a separate exposure measure. Unknown values are null, never zero. Do not combine per-person figures and party totals without explicitly converting them.

For each homogeneous route/mode stratum publish sample count, collection dates, median and observed range only once the sample is adequate for the stated claim. With fewer than five comparable observations, publish individual observations without a performance ranking. This threshold is an editorial reporting rule, not a statistical guarantee.

Do not infer safety, service quality, cancellation rights, future fares or national airport performance from this small convenience sample. Describe bias, unavailable inventory, dynamic pricing, access blocks and exclusions.

## Publication gate

An editor checks arithmetic and source references independently, confirms image/data rights and commercial disclosures, publishes the denominator and limitations, and records corrections in a version history. No affiliate commission affects inclusion or rank. Supplier terms may change after collection.

Release each populated CSV with its version, observation period, method, source references and licence scope. Until that gate is met, label all files protocol or blank template, not research findings.