SPINE — 52 mobile artboards

Crisis link lives in the utility bar, above the brand — it outranks navigation on every screen and never scrolls out of a person's mental model.

Only two things survive the collapse: the mark and Menu. The crisis link keeps its own row rather than being folded inside the menu.

Full-width rows at 48px+ so the menu is usable one-handed on a cheap phone; Sign in sits last because nothing upstream requires it.

The disclaimer is body copy under a 2px rule, not fine print — it is the one claim the product must never let a user miss.

A solid header block instead of a photo: the H1 and the no-login CTA are the only things competing for the first screenful, and the stat strip carries its own honesty line rather than a triumphant claim.

Type-to-filter over 36 tappable cards rather than a native select: the reduced grid shown mid-typing proves the layout still holds when only three options remain.

One question per screen with Back always present, so a wrong tap costs one step and never restarts the flow.

Bands, not a rupee field — asking for an exact number invites a wrong answer and implies a precision the schemes never check. "I'm not sure" is an allowed answer.

The helper line explains why the question is asked, so a fourth-attempt aspirant understands a later exclusion instead of feeling judged by it.

The final button names the reward ("See my matches") rather than "Finish", and the copy states this answer filters nothing out.

The deadline is the card's header — bigger than the scheme name — because a missed date makes eligibility irrelevant; "last verified" appears on every card so trust is dated, not assumed.

Three concrete exits and the free helplines sit on the empty state, so the worst outcome in the flow still hands the user somewhere to go.

Schemes without a deadline still get the same header slot, labelled "Rolling intake", so the eye never has to work out why one card is missing its top line.

The empty state names the exact filter pair that produced zero and offers to drop one of them — cheaper for the user than re-deciding the whole query.

The eligibility checklist admits an unknown with a "?" row rather than guessing a pass — and the outbound button is labelled with its destination so leaving SPINE is expected, not a surprise.

Free cards are large, individually tappable numbers; paid platforms are compact rows on a tinted band below, so visual weight matches what the user can actually afford.

Call-first: the number is the largest element on the page and the anonymity promise sits directly under it, where hesitation actually happens.

Price is stated before any description of the service, and the free route is boxed at equal weight immediately beneath — a user who cannot pay never has to scroll to learn there is an alternative.

Anonymity is a default-on checkbox in the booking flow, not a setting buried elsewhere — and unavailable slots stay visible as struck-through cards so the grid keeps its rhythm.

The disclaimer is the headline, not a footnote under a green tick — a fake confirmation that reads as real is the one failure this screen cannot afford.

Continue without an account sits above the providers because the eligibility checker must never be behind a login; DigiLocker leads the providers since it holds the documents the schemes ask for. No Meta option.

Phone-first with the field open on arrival: the cheapest identifier for this audience gets zero extra taps, and the copy states the minimum data collected.

Two groups of three at 44px+ each so the code is readable and typable one-handed; the demo code is disclosed on screen rather than faking a delivered SMS.

The error keeps the entered digits and points at the likely mistake instead of clearing the field; error weight comes from a 2px rule and an icon, never colour.

Expiry disables the inputs and promotes resend to the primary action, so the only sensible next tap is the biggest thing on screen.

Optional toggles ship off and each card explains the consequence in one sentence, so "Only what's required" is a real one-tap route rather than a dark-pattern decoy.

Onboarding reuses the matcher screens inside a three-step wrapper and carries over any answers already given, so signing up ends on real results rather than an empty dashboard.

One field, one button, and a reassurance that a forgotten password does not cost the user their saved schemes.

Errors sit under the field they belong to and the rule is stated plainly with no arbitrary complexity demands — password theatre punishes this audience hardest.

The dashboard opens on the single deadline that matters most, boxed at 2px; the full match list, counts and role switcher descend in order of urgency below it.

One CTA and a dashed preview of the three cards that will fill the page — the empty state teaches the payoff instead of apologising for having no data.

Remove is a low-contrast sibling of Open rather than an icon, so a destructive tap is deliberate and readable at a glance.

The empty shortlist names the exact gesture that fills it and offers the matches the user already has, so it is never a terminal screen.

Status tags pair a shape with the word so they survive greyscale and colour-blindness, and the standing note kills any impression that SPINE is filing on the user's behalf.

The empty tracker states plainly that marking is manual, then carries the nearest deadline in the footer so the screen still does useful work.

Each document says which scheme demands it, so a user can judge what to chase first; the list is derived from saved schemes rather than a generic checklist.

Rather than an empty box, it explains the mechanism and gives the common three documents so a user can start collecting before saving anything.

The consent toggles are mirrored here verbatim so the DPDP decision is revisable in the place users look; destruction is guarded by typed confirmation, not a colour.

The gate lists what is still open rather than only what is locked — a login wall on this product must never look like the end of the road.

File size and last-updated sit on every row because data cost and staleness both matter here; no download counts, since popularity is not a quality signal.

The empty state admits the coverage gap by name and offers the report route, so a missing language reads as a known limitation rather than user error.

Every card wears a CONCEPT label and there is no member count, no activity feed and no testimonial — nothing that would imply a community that does not exist.

The concept tag precedes the H1 and the intro denies partners before describing them, so the page cannot be screenshotted into an implied endorsement.

"What SPINE is not" gets equal structural weight to "what SPINE is" — the disclaimers are the product's core claim, so they sit in the body, not the footer.

A stated response time sets expectations, and the form redirects urgent distress to a phone line rather than a three-day queue.

A summary at the top plus per-field messages that say what to do; typed input is preserved and errors are drawn with rules and icons so greyscale loses nothing.

A demo success screen returns the user's words with a copy action, so effort spent typing is not silently thrown away by a fake confirmation.

Gaps are listed at the same weight as strengths, itemised specifically enough to be actionable rather than a compliance paragraph.

The policy opens with the architectural fact that makes the rest credible, then itemises storage in a table a person can check against their own expectations.

Six numbered clauses in ordinary sentences rather than a wall of legalese — a terms page nobody can read protects nobody.

The reassurance about saved work comes before the exits, and both exits lead to the two things a person actually came for rather than a generic homepage.

A server failure must not sever a person from crisis support, so the helpline is repeated on the error page with the note that it works independently of SPINE.