built to scale10² → 10⁷ treesa scaling thesis

From one tree
to one crore.
Designed for both.

The hard part of a model like ours isn't the first hundred trees — it's making sure the ten-millionth tree gets the same honest treatment. Here's what stays the same forever, what we have to redesign at every order of magnitude, and what we expect to break along the way.

The whole thesis, in 30 seconds
10²
Pilot· we are here
10 — 1,000 trees
10³
Region
1,000 — 10,000 trees
10⁴
Multi-state
10,000 — 1 lakh trees
10⁵
National
1 lakh — 10 lakh trees
10⁶
Subcontinent
10 lakh — 1 crore trees
10⁷⁺
Beyond
1 crore +

What never changes

Five
immutable
principles.

The whole scaling argument rests on these. If any of them break under pressure, the thing we built isn't worth scaling. Each is written so a district lead in 2032 can hold us to it.

01
100% direct payment
UPI rails — and their analogues elsewhere — already handle ₹100 lakh crore a year. The direct-pay model scales infinitely; only ops scales linearly with us.
02
One photo, one tree, one number
Each tree carries a permanent ID and a public page from planting day to year 20. The proof unit never changes — just how we collect and verify it.
03
Native species only
No eucalyptus, no chir pine in new plantings, anywhere, ever. The species list is set by local foresters per district — but the rule is global.
04
Public failures
Dead trees are marked dead. Refunds are listed. Bad-actor farmers are named when removed. Transparency compounds — opacity rots.
05
The farmer is not our employee
Independent relationship. We make the introduction, we hold the standard, we drop them if they fail. We never own the trees or the land.

The scaling ladder · six phases
1
Phase 1 · 10² trees
Pilotwe are here
range
10 — 1,000 trees
where
1 district
farmers
5 – 30
ops shape
1 person · the founder
What works at this scale

Eyeball verification. Every farmer met in person. Every plot walked. Every photo screened by the founder in the platform's own message threads.


What breaks at this scale

Founder bottleneck. Burnout. Single-language. Can't sleep on Tuesdays.

What we build to get through
01
A directory.
Not a platform. Each farmer's profile is a single page maintained by hand.
02
In-house donor↔farmer threads.
Donors message their farmer directly on PlantTree — no phone numbers shared. The thread is the audit log.
03
A photo-proof standard.
Sapling + wooden tag + tree number, visible. Failure to post by day 14 = refund.
2
Phase 2 · 10³ trees
Region
range
1,000 — 10,000 trees
where
1 state · 6–10 districts
farmers
30 – 200
ops shape
1 founder + 3–5 district leads
What works at this scale

Federated verification. Each district lead is a farmer-respected local who vets peers, owns a 200-tree cluster, and gets a small monthly stipend paid out of an op-fund (NOT donor money).


What breaks at this scale

Spreadsheet chaos. Photo backlog. Farmer onboarding inconsistency. The founder's message inbox queues up.

What we build to get through
01
Farmer mobile app (PWA).
Native language. Camera with auto-watermark. Photo uploads tagged with GPS + tree number, posted straight into the donor thread.
02
In-house messaging at scale.
Templated planting briefs, status nudges, automatic 'photo received' confirmations — all on our own rails, with SMS/email fallback.
03
District-lead dashboard.
Each lead sees their cluster: due photos, missing tags, replant flags. Owns the data.
04
Section 8 / Trust registration.
Filed once we cross 1,000 verified trees and have ops to justify the paperwork.
3
Phase 3 · 10⁴ trees
Multi-state
range
10,000 — 1 lakh trees
where
3–5 states
farmers
200 – 2,000
ops shape
12-person core team + ~50 district leads
What works at this scale

State chapters with autonomy. Forest department MOUs in each state. FPO partnerships bring 100s of farmers at a time. 80G registration unlocks CSR rupees for ops.


What breaks at this scale

Tone of voice drifts state-to-state. Quality variance in plots. Verification visits not scalable on foot.

What we build to get through
01
CV-assisted photo review.
Auto-flag photos missing the tag, blurry, or geo-mismatched. Human still approves — model just queues priority.
02
Satellite NDVI cross-check.
Yearly canopy delta per plot, sourced from open Sentinel-2 data. Catches abandoned plots without a site visit.
03
State playbook.
Locked rules: native-only species lists, sapling cost bands, payment cadence, refund triggers. State leads can't override.
04
Open species + soil API.
Public dataset matching elevation × rainfall × soil → recommended native species. Anyone can query it.
4
Phase 4 · 10⁵ trees
National
range
1 lakh — 10 lakh trees
where
All India
farmers
2,000 – 25,000
ops shape
40-person core + state chapters
What works at this scale

Each state runs its own ops on shared standards. Public tree registry as open data — every tree's number, plot, farmer, species, lat/long, planting date, latest photo. Donors come in via UPI International (NRIs) without us holding FX.


What breaks at this scale

Funding for ops gets serious — district leads need real salaries. Linguistic + cultural variance compounds. Bad actors try to game the model.

What we build to get through
01
Operator funding model.
Optional 'add ₹50 for the ops fund' checkbox at checkout. Separate ledger, audited quarterly. Never mixed with farmer payment.
02
10-language farmer interface.
Hindi, Marathi, Tamil, Telugu, Kannada, Bangla, Gujarati, Punjabi, Malayalam, Odia. Voice-first for low-literacy users.
03
Public tree registry.
Open, downloadable, queryable. Like a property record — every tree has a permanent ID and a public page.
04
Insurance pool for replants.
1% of every payment (with explicit donor opt-in) goes into a state-level fund for re-plantings beyond year 1.
5
Phase 5 · 10⁶ trees
Subcontinent
range
10 lakh — 1 crore trees
where
India + Nepal + Bhutan + Bangladesh + Sri Lanka
farmers
25,000 – 1.5 lakh
ops shape
Federated non-profit utility
What works at this scale

We stop being a website and start being public infrastructure — a non-profit utility, like UPI itself. Other NGOs, school programs, government schemes, and CSR teams all plant on our rails because they're the most boring, audited, undeniable rails available.


What breaks at this scale

Greenwashing risk: someone tries to mint dubious carbon credits on top of the registry. Political capture risk in some states. The brand has to stay stubbornly small even as the impact gets large.

What we build to get through
01
Open planting API.
Any organization can register a tree, pay a farmer, hand off verification to our standard. Free. No commercial license. Source-available.
02
Carbon-credit firewall.
Registry stays free + public. Credits are issued only with explicit farmer consent + farmer share. Most plots opt out — and that's fine.
03
Cross-border UPI rails.
Same direct-payment model in NPR / BDT / BTN / LKR currency corridors as they open up to UPI.
04
Independent audit body.
Annual third-party audit of every state chapter's books, plots, and refund record. Published in full.
6
Phase 6 · 10⁷⁺ trees
Beyond
range
1 crore +
where
Asia · Africa · Latin America
farmers
1.5 lakh +
ops shape
Mostly local, federated
What works at this scale

The thing we built becomes uninteresting infrastructure — like postal codes. India proved a direct-payment registry model could plant 1 crore trees with one-zero-zero percent passthrough. Other countries fork the protocol, run their own chapters, share the registry standard.


What breaks at this scale

Honestly, by here the protocol has to be more robust than the organisation. If PlantTree.life-the-NGO dies, the registry, the standards, and the farmer relationships must keep working.

What we build to get through
01
The protocol, not the platform.
Open spec for direct-pay, photo-proof tree registries. Independent national chapters. We're one implementation of many.
02
A boring foundation.
Long-tenured stewardship. No founder cult. No exits. Like the IETF or Wikimedia, but for trees.

What breaks at each jump · honest

The hard part isn't growing.
It's not breaking.

Every order-of-magnitude jump kills something that worked before. Naming the failure ahead of time is the only way to design for it. None of these are surprises — they're scheduled.

from
to
what breaks
the redesign
100
1,000
In-person verification breaks
Founder can't visit 1,000 plots. Solution: district leads as verifying peers, paid out of separate ops fund.
1,000
10,000
Single-inbox ops breaks
One founder inbox can't carry 10k tree events. Solution: in-house messaging at scale + farmer PWA + district-lead dashboards.
10,000
1 lakh
Single-language platform breaks
We can't run Tamil + Bangla + Marathi farmers off a Hindi-English site. Solution: 10-language farmer interface + voice-first UX.
1 lakh
10 lakh
Ad-hoc ops funding breaks
District leads need real salaries; founder's pocket is empty. Solution: 80G + CSR + opt-in donor tips, fully separated from farmer payment.
10 lakh
1 crore
Single-organisation governance breaks
No one NGO should hold a registry of 1 crore trees. Solution: federated state chapters + an open protocol others can implement.

Four independent levers

Scaling isn't one dial. It's four — each on its own clock, each with its own failure mode. We move People from founder-verified to federated-protocol. We move Tech from spreadsheet to satellite. We move Money from pocket-funded to audited utility. We move Trust from individual pages to an open registry. None of them can lag too far behind the others.

Federation, not hiring
People

We never employ farmers. We slowly devolve verification authority from one person → small clusters → state-level peers → independent chapters → an open standard anyone can run.

The progression
Founder verifiestoday
District leads
State chapters
Federated chapters
Open protocol
Automate verification, never trust
Tech

Tech does the boring stuff — flag suspicious photos, cross-check canopy growth from space, generate planting briefs. Humans approve every decision that affects a farmer's payment.

The progression
Eyeball + spreadsheettoday
Farmer PWA
CV photo triage
Satellite NDVI
Open registry API
Op funding, never farmer skim
Money

The donor's ₹500 always goes 100% to the farmer. Ops are funded from a separate ledger — grants, opt-in tips, CSR — that grows on a different axis from the planting volume.

The progression
Founder's pockettoday
Small grants
Opt-in ops fund
CSR + 80G
Audited utility
Show the work, always
Trust

The platform is opaque to no one. Every tree, every payment, every refund is on a public page that doesn't require login to view. The audit is the brand.

The progression
Per-tree pagestoday
Public refund log
Open data exports
Tree registry API
External audits

What scales by itself vs what we have to redesign
Scales naturally · already good for 10⁷

The infrastructure we sit on.

  • UPI rails
    Handles ~₹100 lakh crore / year. Cost per transaction: ₹0. No re-design needed for any volume we'll ever hit.
  • SMS + email (transactional)
    Notification fallback for in-house messages at ~₹0.12/SMS. Survives billions.
  • Sentinel-2 + Landsat
    Free, public satellite imagery on a 5-day refresh. NDVI is well-understood.
  • Aadhaar / DigiLocker
    Farmer KYC at population scale. We tap, we don't build.
  • The directory model itself
    A static per-tree page is just HTML — cacheable, indexable, archivable, costs ~₹0/year/tree.
Has to be redesigned · the work

The parts we own and rebuild.

  • Verification authority
    Today: founder eyeballs every photo. Tomorrow: federated peer-verification with CV + satellite triage.
  • Operator funding
    Today: founder's pocket. Tomorrow: audited ops ledger, opt-in tips + CSR — never mixed with farmer payment.
  • Onboarding new farmers
    Today: a coffee at the chai shop. Tomorrow: FPO bulk onboarding + voice-first PWA in 10 languages.
  • Plot ecology screening
    Today: a forester walks the plot. Tomorrow: open species API (elevation × rainfall × soil → recommended species) reviewed by humans.
  • Governance
    Today: founder + 3 friends. Tomorrow: independent state chapters + external audit + an open protocol that outlives any one org.

The architecture, at 1 crore trees
Layer 1 · the rails
Public infrastructure
UPI · Aadhaar · Sentinel-2 NDVI · SMS/email · GST · Bharat Nidhi. Plumbing we never have to build or maintain.
Layer 2 · the protocol
The PlantTree spec
Open, source-available standard for: tree IDs · photo-proof format · refund rules · species lookup · farmer registry. Anyone can implement.
Layer 3 · the chapters
State implementations
Uttarakhand, Tamil Nadu, Karnataka, Maharashtra, Odisha … each runs autonomously on the protocol. Independent boards. Shared standards.
↓ implements ↓
Farmer PWA
10 languages, voice-first, offline-first
Donor app
Browse · pay direct · track 20 yrs
District-lead dashboard
Cluster ops, replant flags
Public tree registry
Every tree, queryable, downloadable
Audit & refund ledger
Live, public, third-party reviewed
↓ serves ↓
1.5 lakh
farmers planting
~50 lakh
donors over the decade
1 crore
trees with a permanent page
The point of the architecture

The protocol is more important than the platform. If PlantTree.life-the-org dies, the registry, the standards, the farmer relationships, and the public pages all keep working — because no one part holds the whole. That's the only way an org promising twenty years of growth updates can credibly promise twenty years.


What compounds — the long game
moat 01
Trust capital
Every public photo, every public refund, every named-and-removed farmer adds to a stack that competitors can't buy. By 1 lakh trees, the brand is the audit.
moat 02
Species & plot data
By 10 lakh trees we have the most granular planting registry in India — elevation × rainfall × soil × species × year-1 survival. Open data, useful to everyone, owned by no one.
moat 03
Farmer network
FPOs and panchayats start coming to us. The directory effect: more verified farmers → easier donor choice → more payments → more farmers worth verifying.
moat 04
Forest dept relationships
Each state MOU lowers the cost of the next state's MOU. By national phase, planting on degraded community land is procedurally easy across India.
moat 05
Diaspora rupees
UPI International + 80G means an NRI in Toronto can send ₹500 to a farmer in Almora with the same friction as a domestic donor. That base scales to lakhs of donors.
moat 06
Time itself
A tree paid for in 2026 is more valuable in 2046. Our incentive structure rewards survival, not planting volume. The further out we go, the more our work has compounded.

At 96 trees, this looks like
one person's project.
At 1 crore, like infrastructure.

Both are intentional. The first hundred prove the unit economics. The next ten million prove the protocol. Neither is the goal on its own — both are the same idea, designed at very different scales.

The whole thesis
  1. Five invariants that never change.
  2. Six phases, each redesigning ops.
  3. Four levers, on their own clocks.
  4. One open protocol — bigger than us.