Dwell — Renter-Trust-Score Marketplace Case Study | Abhay Kumar
Concept · Pre-Launch Discovery

Dwell

Bank-verified trust infrastructure for shared living — giving 1.2M Canadian newcomers a portable rental reputation Equifax and TransUnion can't see.

My role
Discovery → strategy
Status
2 of 9 interviews done
Stack
React Native + Supabase + Flinks
Market
1.2M newcomers/yr
The problem

Equifax and TransUnion are blind to 1.2M people a year

1.2 million newcomers enter Canada every year with no Canadian credit history — which means no rental history, which means landlords reject them on sight regardless of how reliable they actually are. The result: 6-month housing searches and 25–30 applications to rent a single room.

  • References are unverified; WhatsApp screenshots aren't evidence. No portable record of "what kind of tenant is this" exists anywhere.
  • Genuinely reliable renters are turned away by a broken proxy signal — Canadian credit history — that has nothing to do with how they actually pay rent.
Discovery in action

Two conversations that changed the product

Interview #1 · Ashmita, newcomer renter

She'd viewed 20 rooms over 6 months — "mostly dingy homes." That single detail exposed a gap the original 3-pillar design didn't cover: a verified renter has no way to know if a listing is worth their time either.

→ Added the Property Trust Score as a 4th pillar — landlords and listings get scored too.

Channel interview · Real estate agent

A realtor renting via Facebook Marketplace pointed out the model's weak spot directly: "marketing is free… I don't know how you'd charge the individual owner." The small landlords the plan billed at $15/unlock are exactly the ones using free channels — and the real B2B gatekeepers are property management companies, not individual owners.

→ Monetization model flagged as unresolved before build; PMCs added as a target interview segment.

Personas

Four people, four different frictions

Maya · International student

No Canadian credit history. Rejected from 15 rooms before finding one — needs any signal that separates her from 40 other applicants.

Dev · Tech worker newcomer

Has income, no Canadian credit trail. Lost his dream apartment twice to Canadian-born applicants.

Priya · Young professional

Domestic renter, two roommates, WhatsApp chaos over bills. Wants the Ledger and Agreement pillars first.

Marcus · Small landlord

Owns 2–3 units, self-manages. Spends 4–6 hours per vacancy screening — can't tell good tenants from bad until too late.

Backlog

User stories that shaped the roadmap

US-01 As a newcomer renter, I want a portable, bank-verified trust score so landlords have a real signal instead of rejecting me on sight. Must
US-02 As a renter, I want day-to-day utility — ledger, tasks — before my bank is ever connected, so I get value before facing any friction. Must
US-03 As a renter, I want full data deletion on demand so I trust connecting my bank account at all. Must
US-04 As a landlord, I want to unlock a pre-verified renter profile for $15 so I skip hours of manual reference-checking. Should
US-05 As a renter who's toured bad listings, I want a Property Trust Score so I stop wasting time on poorly maintained rooms. Should
US-06 As a property management company, I want bulk directory access so I can screen tenants across every unit I manage. Could
Requirements

The 90-day MVP, spec'd as acceptance criteria

RequirementPriority
House creation + member invite — the core onboarding without which nothing else worksMust
Digital agreement with e-signature and Day-1 PDF exportMust
Expense splitting + ledger — the daily-retention feature (Splitwise replacement)Must
Basic manual Trust Score ("Thin File"), not shareable until Flinks-connectedMust
Flinks connection + score upgrade, gated behind an explicit directory opt-in toggleShould
Tenant directory + landlord unlock flowShould
Full Property Trust Score algorithm (post-viewing survey is Phase 1; full scoring is Phase 2)Could
My role

Running discovery and letting it actually change the plan

I ran the discovery interviews, and treated every one as a chance to break the existing plan rather than confirm it. Ashmita's answer added an entire pillar. The realtor's answer put the pricing model on hold rather than let it ship on an unvalidated assumption. I then designed the trust-score algorithm, anti-gaming mechanics, and database schema so the next 5 interviews have something concrete to react to.

What I designed

Four pillars, one trust graph

The Agreement

Digital house agreement with e-signature — enforceable, timestamped, shared with every housemate.

The Ledger

Shared expense splitting and task tracking with dispute resolution — the daily-use hook.

Renter Trust Score

A portable, bank-verified 300–850 behavioural reputation score, built from real rent and bill payment data.

Property Trust Score

The mirror score, born from Interview #1 — landlords and listings scored on response rate, condition, and dispute history.

Under the hood

What actually moves the score

Rent on-time rate
30%
Utility / bill payment
15%
Expense settlement
15%
Task completion
15%
Tenure
10%
Dispute rate (neg.)
10%
Endorsements
5%

A velocity flag lowers confidence on a suspiciously perfect record — 100% completion with zero disputes reads as gaming, not virtue.

Competitive moat

The renter's advocate, not another blacklist

Openroom and FrontLobby already own the landlord-protection, negative-data side of this market. Their model requires renters to distrust them — no one hands their bank data to the platform built to report them. Dwell's Renter Trust Score only works if renters willingly connect their bank, which only happens if the product is renter-owned and never becomes a blacklist. That's a moat built from trust, not a feature list — and it's the one thing an incumbent can't copy by shipping a new module.

Outcome

A build decision that waits on evidence, on purpose

Dwell is 2 of 9 discovery interviews in, with a working score algorithm, database schema, and a 90-day MVP plan ready to execute — and, just as importantly, a monetization assumption correctly flagged as unresolved rather than shipped blind. The build/no-build decision waits for the remaining 7 interviews, on purpose.

Roadmap

Discovery → MVP → B2B → scale

Now

7 remaining discovery interviews (4 renter, 3 landlord) plus a property-management-company interview, then the build/no-build call.

Days 1–90

7-feature MVP: house creation, agreement, ledger, task tracking, thin-file score, Flinks upgrade, directory.

Days 91+

B2B landlord portal + Property Trust Score, then US expansion after 500 verified Canadian scores.

More case studies

See the rest of the work, or get in touch.