UX Prototyper · Reinvention set · Auction Details · 31 Aug 2026
Make First Class worth the bid, before the bid.
The opportunity: a customer looking at an auction cannot tell what First Class actually gets them on this train, so they bid low or not at all. The axis: what the customer needs in order to bid higher — each variant assumes a different reason people underbid. Journey, prices, auction state and the sticky footer are identical across all three; only the assumption is under test. This is deliberate reinvention, not a fidelity error — today's Auction Details screen is the thing being replaced. This set supersedes the 19 Aug set (v1–v3) and is built to the raised standard approved 31 Aug. The screens below are the live prototype files, embedded, never redrawn — each links to its clickable prototype.
Unverified, and shown anyway — read this first
The cabin photography does not exist (A shows labelled placeholders, never stock). The £94 walk-up fare is illustrative and tagged "to verify" wherever it appears — we may not hold it at bid time (open with Ben). The proof-chip counts are sample data; the survey isn't live. The cancelled-train promise is unconfirmed policy (Ops). The comparison cells and struck-through lounge reason are written to be defensible, not from operator data. The journey date was moved to Wed 09 Sep 2026 because the brief's date is in the past. Tokens are provisional and the typeface is a fallback.
1 axisWhat the customer needs in order to bid higher. Everything else is held identical across the set.
3, distinctA photo-led listing, a pricing ledger, a plain-words page. The one-sentence test below is the check.
1 cheap controlVariant C needs no photos, no fare feed, no model. It is the baseline A and B get judged against.
0 fake assetsEvery placeholder and illustrative number is labelled on the screen itself, not only in the handoff.
The same moment, three ways
Kings Cross → York · LNER · current bid £19 · buy now £42
One solution — the same Auction Details page, the screen a customer lands on when they open this auction, built three structurally different ways. Every variant keeps the same spine: journey identity at the top, the live auction state, and the identical bid/buy footer with the fee line. Only the lead changes. The phones are live: scroll them, tap the photos, open the sheets. For the full flow, open each variant's clickable prototype.
Variant A · See it
A product page for the seat itself
A photo-led listing that sells the seat before it shows the auction.
M
Mechanism. Airbnb's listing page. You see the space before you see the price, so the price arrives as a verdict on something you have already valued. Photography answers the objection prose cannot — P3: “doesn't scream to me that it's worth spending this amount on.”
H
Hypothesis. If people can see the actual coach before they reach the bid control, they will name a higher bid, because today they are pricing something they have never seen. Behavioural, not a primary metric — the A/B Test Template owns the metric.
A
Assumptions. Real cabin photography from LNER, which we do not have. Every image is a labelled placeholder tagged “PHOTO · OPERATOR PACK”, never a stock photo. Fallback: benefit list only — variant C shows what that looks like. Proof-chip counts are sample data, tagged so on screen.
A ledger that prices First Class against what you already have.
M
Mechanism. The reference-price anchor, done honestly. Airbnb shows the full price before you commit; eBay shows the price you're beating. A number next to a real reference reads as a verdict; alone it reads as a guess.
H
Hypothesis. If the bid sits next to the walk-up fare and the rows that actually change, people will name a higher bid, because they cannot judge £19 in a vacuum. Behavioural, not a primary metric — the A/B Test Template owns the metric.
A
Assumptions. The £94 walk-up First Class fare per service — we may not hold it at bid time (open with Ben), and it is tagged “to verify” on the screen itself. No fare, no anchor: the design falls back to the struck £56 and loses its lead. Comparison cells are written to be defensible, not from operator data.
A near-empty page that states the deal in plain words and keeps everything else one tap away.
M
Mechanism. Monzo's plain language — people trust what they can restate in their own words — organised around Airbnb's “show all amenities” disclosure. If people already know what First Class is (P5 described it unprompted), more information is the wrong medicine.
H
Hypothesis. If the page says less, more people will reach the bid control with a number already in mind, because the blocker is doubt, not missing information. Behavioural, not a primary metric — the A/B Test Template owns the metric.
A
Assumptions. Almost nothing, which is the point — no photos, no fare feed, no model. This is the cheap control, buildable today with copy alone. Two lines still need confirming before a live test: the cancelled-train promise (Ops) and the £94 anchor, tagged “to verify” on screen.
A photo-led listing that sells the seat before it shows the auction.
B in one sentence
A ledger that prices First Class against what you already have.
C in one sentence
A near-empty page that states the deal in plain words and keeps everything else one tap away.
The distinctness test, run before publishing: describe each variant in one sentence. No two of these fit the same sentence, so none is a skin of another.
Read before treating any of this as a decision
These are drafts to test, not final design — production design is Liz's team's call. Every screen names First Class explicitly, and class disambiguation rests on that naming alone (Aaron's call, 2026-08-31). The strike-through marks only stated absences and must never touch a seat reservation. No scarcity copy, no Trustpilot, no "X people upgraded on this route", no invented saving percentage — all deliberately absent. And Auction Details is easily bypassed from search results (Rob); what proportion of customers open it at all is an unanswered gate in Scope.md, so the winning layout may need to migrate to Results and Confirm & Pay to matter. Full detail: handoff-auction-2026-08-31.md.