Design brief · Q4 2026
Raise the bar of the upgrade funnel
The upgrades funnel is primarily a handful of screens. Our customer interviews highlighted the main problems across that journey, and this brief gives direction on how we solve them without introducing too much scope. Every requirement is traced to the problem it came from, so you can always see why it is there.
For UX design
From Aaron
Date 1 September 2026
02
Auction Details
5 opportunities
03
Bid entry sheet
Designed, never released
04
Confirm bid / BUN
3 opportunities
The ask
Raise the quality and the bar of the upgrade funnel. The scope is the three screens where these problems land: Results → Auction Details → Confirm bid / Buy now.
Why now. Q4 needs the monthly transacting base to go from 2.6% to 4.5%. Getting there means taking bold steps to significantly improve the UX, and staying focused on solving these problems in a sequenced way that delivers the business target.
The UX itself is causing a lot of the friction today, and we see it in the drop-off rates. That gives us guidance we would otherwise be guessing at: we already know which pages need the most work. Auction Details needs it most.
Design boldly. We will sequence it together. Do not pre-cut it to what you think Q4 can afford — that is my job with engineering.
Context, and who we are designing for. 1.7 million customers have signed up to Seatfrog, and 0.4% of them have ever transacted. That is the audience. Every requirement here is framed for someone who has potentially never placed a bid and never bought anything from us, because that is who almost everyone is.
The five opportunities
All five come from the customer interviews, themed. All five ladder to the same place: more of the people who start an upgrade finish one.
01
Help customers judge whether an upgrade is a fair deal
"Is this a good price?"
Trying to moveMore of the people who search end up with an upgrade
02
Make the value of first class obvious before they bid
"What am I actually getting?"
Trying to moveMore people bidding at all. Not the size of the bid
03
Make the fee feel fair, not a penalty
"Am I being stung by the fee?"
Trying to moveTrust. Nobody meets a number they were not expecting. The price they first see is the price they pay
04
Help customers judge how likely they are to win if they place a bid
"If I bid, will I win?"
Trying to movePeople who open an auction going on to place a bid
05
Make bidding clear enough that customers don't feel they must bid late to win
"How does this work, and when should I bid?"
Trying to moveBids landing earlier, without lowering them
The solutions themselves have not been scoped out. Where there are two options that both look good, pick the one we think moves a number in this column. Put the variant away, and we can use a test to narrow down to the best path.
Where each opportunity lands
The strongest requirement each screen carries for each opportunity. Read the columns, not the rows: it shows what each screen is being asked to hold.
| Opportunity | Results | Auction Details | Bid entry | Confirm |
|---|
| Help customers judge whether an upgrade is a fair deal | Must | Must | Must | Must not |
|---|
| Make the value of first class obvious before they bid | Should | Must | — | Must |
|---|
| Make the fee feel fair, not a penalty | Must | Must | Must | Must |
|---|
| Help customers judge how likely they are to win if they place a bid | Out | Should | Must | Should |
|---|
| Make bidding clear enough that customers don't feel they must bid late to win | — | Should | Must | — |
|---|
Design with assumptions in mind
Where something is unconfirmed or does not exist yet, design as if it does and write the assumption beside it.
The example is imagery. Design these screens with great first class images and video in them. If we do not have those today, the design has just surfaced a crucial dependency, and closing it is Product's job, not yours. That is a better outcome than a screen designed around what we happen to have.
Don't overload the screen
Today there is too much on these screens. That is the problem. Strip back to what is essential.
- Only ever display what is exactly needed for the job that screen does. That is the principle we need in place, and it applies before anything gets added.
- If something takes key real estate and is too risky to remove, highlight it. Don't quietly keep it, and don't quietly cut it.
- If you think something critical is missing, that is a conversation, not an addition.
We are designing for someone in a pre-bid or pre-buy state who has never transacted with us. They need exactly enough to decide, and nothing that arrives after the decision belongs on these screens.
And this design does not have to serve everyone. The product vision is an experience that changes by segment, so it does not have to carry everything for everyone at once.
What is out of scope
The homepage, and redesigning it
It is about getting more people into the funnel. If the funnel is broken, more people entering it does not convert efficiently. We fix it in this order deliberately
Any post-transaction experience
Not this problem
Any post-bid experience
Not this problem. We can assume those flows inherit whatever design is introduced here
Must fixed
Should wanted
Could if the screen can carry it
Yours to decide a real design decision
Must not / Out deliberately excluded
01
Results
Job
Which of these trains is worth a tap, in one scan
Moment
Post-search, scanning, mobile
Watch for
Overcrowding. This screen is for scanning, and it stops working as a list before it starts looking crowded, so it cannot be judged by eye. Check that someone can still find their train
01 Help customers judge whether an upgrade is a fair deal
Must
The customer can say what the price is being measured against
Should
The reference is the station walk-up fare for the same move
Yours to decide
How we do this. Boundary: they can say what it is measured against either way, and scanning must not slow
02 Make the value of first class obvious before they bid
Should
One line of what differs, e.g. "More space, a table, a definite seat". The concept line, for someone who does not yet know what first class is, and the only thing on this page answering "what am I getting"
Could
The same line made dynamic to time of day, e.g. morning leads with breakfast, evening with a drink. The segmented version of the row above, not an alternative to it — it lands once we know the customer, where the concept line is what someone new needs first
Could
Class named on the result itself. Only material on one service today (LNER)
Could
Thumbnail of the actual carriage on this operator. It is the element that forces the card decision below, so treat it as part of that decision rather than an addition
03 Make the fee feel fair, not a penalty
Must
Every price fee-inclusive
04 Help customers judge how likely they are to win if they place a bid
Out
Anything about likelihood of winning. This screen is not for odds thinking, and it would add overload
All Applies across every opportunity
The decision here, and it is yours with Product
How much this page displays, as a principle — not this card in isolation, but the rule we hold ourselves to afterwards.
- Fair deal and fee own the Musts here. First class value rests on one Should and two Coulds, on the assumption Auction Details picks it up one page in.
- If we are creating overload, pick one deliberately: redesign the card so it can carry it, or remove components until the load comes down.
- What we must not do is keep adding and hope scanning holds.
02
Auction Details
Job
They are not ready to commit. They have shown curiosity and tapped through to find out more about this trip. This page gives them what could persuade them to bid
Moment
Still deciding whether this is for them. Looking for information and reassurance, not a close
Watch for
Four of the five opportunities want this screen
This screen is carrying more than any other, by a distance. Four of the five opportunities want it.
41% of users interact with the tabs. Of those, 22% go straight back to the details page, 35% check the first class benefits and 25% check how this works. We cannot get bounce rates or page duration today. Front-end analytics, so directional.
What that tells us is customers are not finding what they came for. So there is room to make bold decisions here, including cutting.
01 Help customers judge whether an upgrade is a fair deal
Must
The customer can understand whether this price is good, by seeing what it is compared against
02 Make the value of first class obvious before they bid
Must
Show the full benefits of first class
Should
Compare against standard class perks
Could
Catering on this service
Could
Lounge access, where the station has one
Could
Benefits shown dynamically per operator
Should
Images of the first class in action
Could
Video of the carriage
Yours to decide
Whether depth sits on the page or behind a slide-in sheet
03 Make the fee feel fair, not a penalty
Must
Every price fee-inclusive, fee disclosure intact
04 Help customers judge how likely they are to win if they place a bid
Should
The screen says something about their chances that the bid count does not. Today 14 bids reads as "you probably wouldn't win" and a low count reads as "a good chance" — the same signal, read two opposite ways, and it is the only signal on the screen
Should
An indication of how likely they are to win at the number they have in mind
Should
How many seats are winnable on this auction
05 Make bidding clear enough that customers don't feel they must bid late to win
Should
What happens to a bid, stated on the page, not behind a tab. It is a Must on the bid entry sheet, where the number is actually typed
Should
Say plainly whether bidding early costs anything
Must not
Let a countdown own the fold. Urgency chrome is what teaches the behaviour we are removing. A prominent timer while asking people not to bid late is arguing with our own design
All Applies across every opportunity
Must
Keep the sticky footer's two actions, bid and buy
Yours to decide
Which of the four things a bidder needs leads: is this my train · what am I getting · what it costs and is that good · how this works. All four must be present in every version; only the order is open
Do not treat any existing design as decided.
03
The bid entry sheet
Job
The number, and the timing, are both chosen here
Watch for
It is a sheet, not a page
01 Help customers judge whether an upgrade is a fair deal
Must
Say that the amount they enter is the maximum they will pay, and that they may win it for less, because we bid up on their behalf in £1 increments
Must not
Remove the buy-now ceiling from view while they choose. It is already there today
Must not
Carry the station price reference. By now the question has changed from is this good to what am I committing to
03 Make the fee feel fair, not a penalty
Must
Fee disclosure intact, fee treatment unchanged
04 Help customers judge how likely they are to win if they place a bid
Must
They can judge how likely their number is to win, before they commit it. Not a feeling that it is probably fine, a judgement they could defend
Must
If the screen cannot state a likelihood, it says what would make one possible, or gives a rule of thumb. Silence is the one thing it may not do
Must not
Frame a likelihood as a prompt to bid higher. The aim is that more people bid, not that the same people bid more
05 Make bidding clear enough that customers don't feel they must bid late to win
Must
Say what placing this bid commits them to, including whether anything is charged if they do not win
Must not
Imply the amount is final if it is not, or provisional if it is
Yours to decide
Told, explained or shown: one plain line at the point of action, a compact "how this works" card, or the market demonstrated by example
All Applies across every opportunity
Must
If the unreleased sheet already answers a requirement in this table, say which, and start from it rather than redrawing
Not design's
The default is pre-filled at the minimum bid, which anchors at the floor. Raising it moves yield, so it needs a pricing owner. Flag it, do not solve it
04
Confirm bid / BUN page
Job
Confirm what they are buying, and remove the reason to doubt we will deliver it
Watch for
This page already cannot take what is on it
01 Help customers judge whether an upgrade is a fair deal
Must not
Carry a price reference. By now they have chosen this auction
02 Make the value of first class obvious before they bid
Must
The first class perks they are bidding for. Existing content, relocated, above the bid box
Must
"You're bidding for First Class", named in the bid box itself
Should
"If this train is cancelled, you're on the next one. No charge."
Yours to decide
Reading order. Whether what you get is read before what it costs
04 Help customers judge how likely they are to win if they place a bid
Should
Give them direction on what their chances are. Today this is the first and only place a bid can be amended — tapping Place a bid comes straight here — so this is the first chance they get to act on it
03 Make the fee feel fair, not a penalty
Must
The fee as one named item in what the price buys: upgrade + booking fee
Must
Editing a bid floors at starting price plus booking fee
What I need back
01
A coverage line per screen: where each of the five opportunities is answered, and what got sacrificed to fit it. An opportunity answered nowhere is a legitimate answer if you say so.
02
What did not fit. The most useful page you can send me.
03
Which Musts fight each other. They were written by five separate documents and nobody has checked them against one another on one screen.
04
Not polished production design. We slice this into Q4 sequencing first, and polish follows the slice, not the other way round.