Founder case study · 2024—present

HERD

Plan less.
Go together.

A beautiful, shareable trip page that replaces group-chat chaos with one place to decide, coordinate, and go.

Founder & Product LeadiOS + Web

Scroll to unpack the work ↓
My roleFounder & Product Lead
TeamSolo
PlatformiOS + Web
StatusBuilding now

Group travel rarely fails because people lack ideas. It fails because coordination has no home.

01 · The problem

One trip.
Five apps.

Plans get scattered across chat threads, booking links, shared docs, saved maps, and payment requests.

The organizer becomes an unpaid travel agent. Decisions stall. Updates disappear. Nobody knows which plan is current.

Group chat“Which Airbnb did we pick?”
MapsLisbon saves
NotesItinerary_final_v4
Restaurants?
Ask Maya
Booking + emailFwd: Your stay is confirmed · 4 tabs open
Payments3 people still owe $186
HERD
Shared tripLisbon SummerJun 12–18 · 6 travelers
Maya, Jordan, Alex, and three more travelers
The design inputs

Build for the
real group.

These are product-planning inputs—not claimed outcomes. They define the coordination load Herd is being designed to absorb.

Design input5+

places where plans, links, maps, bookings, and payments typically scatter.

Target user2–4×

group trips planned per year by the initial customer segment.

Group model4–8

travelers: enough opinions to create coordination friction.

Working model6

members in the average trip used for the acquisition model.

ConversationGroup chat
PlacesSaved maps
LogisticsDocs + notes
BookingsEmail + tabs
MoneyPayment apps
DecisionsMemory
One shared source of truthLisbon Summer
02 · Product strategy

The link is the product.

Herd follows a web-first distribution model: one person creates a trip, shares a link, and the group can RSVP and coordinate without downloading an app.

01Zero-friction entry

See the trip before being asked to join it.

02Group-first by default

Every feature works for decisions shared by 4–8 people.

03Sharing is acquisition

Every trip page introduces Herd to the next organizer.

Early structure · Wireframe

One page.
Three jobs.

The first information architecture separated context, active planning, and group status without making travelers hunt through screens.

01
Trip context

Dates, destination, RSVP, and people remain visible.

02
Shared workspace

The active decision or itinerary stays central.

03
Group awareness

Activity and planning progress keep everyone oriented.

03 · Current production UI

One shared surface.

Reconstructed directly from Herd’s current component system: trip overview, itinerary, polls, expenses, and AI activity—not a speculative redesign.

DR
Your Trips

Lisbon Summer

Lisbon, Portugal · Jun 12–18

MJA
← My TripsLisbon SummerDR
OverviewItineraryPollsExpensesChat
Fri 13Sat 14Sun 15
Dear Breakfast9:30 AM · Saved by Maya +2
Alfama walking tour11:15 AM · 12 min walk
Prado1:45 PM · Table for 5
← My TripsLisbon SummerDR
OverviewItineraryPollsExpensesChat
Decide together

Where should we stay?

Alfama House 4
Bairro Alto Loft 2
Príncipe Real 1

Friday dinner?

MJA
3 of 5 have voted
← My TripsLisbon SummerDR
OverviewItineraryPollsExpensesChat
Trip spend
Total expenses€1,284.00
€1,284 of €3,000 spent · 43%
⌂
Alfama apartmentPaid by Darren · split 5 ways
€740
♨
Prado reservationPaid by Maya
€180
← My TripsLisbon SummerDR
What’s happening
H
Herd
New poll: “Where should we stay?” — vote now!
H
Herd
New activity added: Prado
✦
Herd AI suggestion
Found a more walkable Friday route.
Review suggestion →
Feature deep dive 01

Let the trip
earn the signup.

An invitation should open the plan—not an installation prompt.

Early wireframe
Open link→View trip→Choose RSVP→Join group
Production UI reconstruction

Lisbon Summer

San Francisco → Lisbon, Portugal
Jun 12–18 · 7 days$1,500–$2,000/person · Public
Who’s going5 travelers
MJAS+1
Join

Lisbon Summer

San Francisco → Lisbon, Portugal
Jun 12–18 · 7 days4/8 going · 1 maybe

Your RSVP

Choose the status the group should plan around.

✓ Going? Maybe× Can’t

Selecting a status saves immediately to your trip membership.

Alternative

Native-only

Strong installed-app continuity, but distribution stops at the download prompt and trust must form before travelers can see the plan.

Alternative

Account-first web

Web links travel easily, but an immediate account gate spends trust before the invitation has shown why continuing is worthwhile.

Chosen direction

View-first link

The link distributes anywhere, earns trust with visible trip context, and turns authentication into a continuation of the RSVP action.

ConstraintInvited travelers may have no Herd account or installed app.
DecisionOpen the shared trip context before asking for authentication.
Why it mattersThe destination, dates, and group make the join request legible and worth continuing.
Product target<2s

Trip-page load

Product target<45s

Invitation to joined traveler

Feature deep dive 02

Turn opinions into visible decisions.

Group chat collects reactions. The product has to show what the group has decided, what remains open, and who still needs to respond.

Coordination sequence Early wireframe
01Ask one questionWhere should we stay?
02Vote + RSVPEach traveler owns a response.
03Show shared state4 of 6 voted · 2 RSVPs open
Production UI reconstruction
Lisbon SummerGroup planning
6 travelers
Where should we stay?4 of 6 voted
Alfama House2 votes
Bairro Alto Loft1 vote
Príncipe Real1 vote
Who’s goingRSVP status
MMayaGoing
JJordanMaybe
AAlexWaiting
Planning Progress3/5
✓Build your itinerary
✓Invite your group
Pin bookings or links
✓Create a poll
Collect RSVPs
ActivityLatest updates
HHerd
New poll: “Where should we stay?” — vote now!
HHerd
Maya changed their RSVP to going
Alternative

Chat-only discussion

Conversation stays familiar, but decisions disappear into message history and the organizer must repeatedly reconstruct consensus.

Alternative

Organizer-controlled planning

One person can move quickly, but the group loses agency and every preference, status change, and follow-up flows back through the organizer.

Chosen direction

Visible group state

Polls capture preferences, RSVP states show attendance, planning progress exposes open work, and activity history keeps changes legible to everyone.

ProblemRepeated questions and ambiguous consensus turn the organizer into the group’s memory.
DecisionStructure only the moments that need a shared answer or accountable next step.
Visible signalPolls, RSVP states, planning progress, and activity history resolve into one shared trip state.
Measurement proposal

Invitation-to-RSVP conversion

Instrument invitation opens and the first saved RSVP to understand whether shared context leads to participation.

Measurement proposal

Poll participation

Track eligible members who submit a response before a poll closes, without treating a target as an achieved result.

Measurement proposal

Time-to-decision

Measure elapsed time from poll creation to a recorded group choice.

Measurement proposal

Organizer interventions

Count reminders and manual follow-ups needed before the group reaches a visible decision.

Feature deep dive 03

AI that joins
the plan.

The product direction is to bring a useful first draft into the shared trip. Today, Herd returns suggestions for travelers to review and explicitly accept.

Working model · current behavior Future workflow concept
Lisbon SummerBuild an itinerary
Shared with 6 travelers
DestinationLisbon, Portugal
DatesJun 12–18 · 7 days
Budget$1,500–$2,000/person
Trip vibeFood · local culture · easy pace
Build the day plan

Herd uses the trip details and the group’s saved preferences as generation context.

Lisbon SummerItinerary generation
In progress

Building your trip...

Named stages make progress—and recovery—visible.

✓Reading trip contextComplete
✓Matching group preferencesComplete
✦Building the day planIn progress

If generation fails, the group can retry without losing its trip inputs.

Lisbon SummerYour first draft
Editable by the group
OverviewItineraryPollsExpensesChat
Fri 13Sat 14Sun 15
Dear Breakfast9:30 AM · Breakfast
Edit
Castelo de São Jorge11:15 AM · Culture
Edit
Miradouro de Santa Luzia12:30 PM · Viewpoint
Edit
Prado1:45 PM · Lunch
Edit
✦
Herd AI suggestion

Saturday pairs Museu Nacional do Azulejo with Time Out Market Lisboa to reduce cross-city travel.

Review suggestion →
Alternative

Standalone chatbot

Flexible conversation can generate ideas, but trip context, group preferences, and every accepted choice must be copied back into the shared plan.

Alternative

One-click result

A fast reveal feels simple, but opaque generation hides progress, makes failure ambiguous, and presents a polished answer before the group can shape it.

Future direction

AI inside the trip

The concept would keep shared context, show progress, land as an editable itinerary, and remain visible to everyone planning together.

Product direction
Keep AI inside the shared trip workflow.

The future concept makes inputs, generation state, and editing visible; current production stops at reviewable suggestions until a traveler accepts one.

Measurement proposal

Generation completion

Instrument suggestion requests, returned results, failures, and retry outcomes; measure any future full-plan flow separately.

Measurement proposal

Time-to-first-plan

Measure elapsed time from the first suggestion request to the first explicitly accepted itinerary activity.

Measurement proposal

Edit rate

Track accepted AI-sourced activities that travelers later change or remove.

Measurement proposal

Accepted suggestions

Count reviewed AI proposals that a traveler explicitly adds to the shared itinerary.

Working method

From principle
to proof.

I balanced product, design, and implementation by turning each decision into the next artifact—then using the working surface to test the reasoning.

  1. 01

    Frame

    Distill the coordination problem into one rule that can guide product choices.

    OutputProduct principle
  2. 02

    Prioritize

    Choose the smallest phase that can prove value before expanding the system.

    OutputScoped phase
  3. 03

    Prototype

    Make the critical path tangible early enough to expose gaps and tradeoffs.

    OutputWireframe
  4. 04

    Build

    Translate the chosen path into accessible behavior across desktop and mobile.

    OutputWorking responsive surface
  5. 05

    Verify

    Check interaction states, breakpoints, focus, reduced motion, and source accuracy.

    OutputQA evidence
04 · Product decisions

Reduce the work
around the trip.

01

Show value before signup

Invited travelers see the trip first. Joining becomes a continuation of the experience, not a gate before it.

Lisbon Summer · 5 travelers
02

Turn opinions into decisions

Polls replace open-ended chat spirals with a lightweight, visible path to consensus.

03

Put AI inside the group

Suggestions appear in the activity flow with context, attribution, and a clear accept-or-ignore choice.

✦ Herd AI
Found a more walkable route →
05 · Experience system

A loop designed
to travel.

The organizer’s workflow naturally introduces Herd to the rest of the group—and every participant can become the organizer of the next trip.

CreateStart with destination, dates, budget, and vibe.
ShareSend one beautiful trip link.
JoinRSVP with minimal friction.
CoordinateVote, plan, chat, and split expenses.
RepeatA traveler creates the next trip.
Growth model · target

Six travelers.
Two future hosts.

If two people from an average six-person trip organize a future trip, the modeled viral coefficient can move above 1.0. This is a product target, not a measured result.

MJASKD
6 exposed travelers2 target organizersK > 1.0 goal
06 · Where it goes

Build the trip page first.

The roadmap deliberately starts with the shareable unit before layering coordination depth and native distribution.

Now
The trip pageCreation, sharing, RSVP, attendees, and core overview.
Next
CoordinationPolls, itinerary, reminders, and logistics.
Then
Money + intelligenceExpenses, AI planning, chat, and shared media.
Later
Native growthPush, deep links, profiles, and mutual travelers.
500
Beta-user launch targetPlanning at least one trip with 3+ members within 60 days. A success criterion for launch—not a current outcome.
Back to the beginningHerd case study↑