Troopod · India Direction · Phase 1 · Combined Retention & Value Analysis

What to scale in India, the evidence.

A validated, brand-level analysis of 43 brands (30 recurring retainers, 13 one-time projects) across churn and LTV. The conclusion on who to focus on, and who to walk away from, is stated up front, with the reasoning. The full churn analysis and LTV analysis that justify it follow. Churn classifications validated by the business team; this is a retention map (no acquisition/CAC data yet), and the biggest single input to direction with founder market judgment layered on top.

Part I
Churn Analysis, why brands leave
Section 1

The overall churn picture.

Of the 24 eligible recurring brands (tenure ≥3 months), ~62% have churned. But "churn" is three very different things, and only some of it is ours to fix.

29%
Delivery fault
(our execution)
8%
Sales fault
(mis-sold / wrong-fit)
25%
Not our fault
(client / payment / shutdown)
Recurring churn breakdown (eligible, n=24), bars sum to ~62% total churn

The headline: ~38% of eligible brands churned for fixable reasons (delivery + sales) and ~25% for reasons outside our control. So the majority of churn is self-inflicted, which is good news, because self-inflicted means fixable. The rest of this section breaks down exactly how.

Section 2

Each type of churn, root causes & how to solve.

Delivery fault, 7 brands

Brands that left because our execution fell short. The single biggest controllable driver.
Timeline delays & slipped deadlines · 3 brands
Fflirtygo, Wiselife, Rrumi, work ran late; Wiselife's work ballooned to 5× the offered hours.
Fix: Lock scoped timelines per sprint with buffer; flag overruns early; cap scope-creep with written change-orders. A simple "are we on track" weekly status to the client kills the surprise factor.
Technical bugs shipped live · 3 brands
Kreo (Cloud SQL billing bug went live), Rrumi (recurring 404s, AWS data loss), Conscious Chemist (preview-vs-live GIT discrepancy).
Fix: A pre-deploy QA checklist + staging-vs-prod parity check. No feature goes live without a second set of eyes. The data-loss and billing-bug class of error is the most damaging and the most preventable.
Quality / "not what we expected" · 2 brands
Protouch (sub-standard experience, sticky cart), Bulbul (dissatisfaction → took team in-house).
Fix: Design/quality sign-off gates at milestones, with the client approving before build proceeds. Catch "this isn't it" at mockup stage, not after build.

Note: several of these brands also "went in-house" afterward, but that was the consequence of the delivery frustration, not the cause.

Sales fault, 3 brands

Brands that left because the sale itself was wrong, not because delivery failed.
Sold without budget to deliver value · 2 brands
Zenue (invoice/consumption-hours dispute), Manetain (mis-scoped at sales).
Fix: A minimum-engagement gate, don't sign brands whose budget can't fund enough hours to move their metrics. Set expectations on what the fee does and doesn't buy, in writing, at sale.
Expectation vs reality mismatch · 1 brand
Snuggly Monkey, what was promised at sale didn't match what delivery could do.
Fix: Sales-to-delivery handoff doc, what was promised, what's in scope, what success looks like, so delivery isn't inheriting a promise it can't keep.

Not our fault, 8 brands

We ended it, or the client left for their own reasons. Reported for honesty; mostly not preventable.
Client ran out of money / shut down · 2 brands
Creslia (shut down), Dr Water (paused for launch).
Partly preventable: better financial-health qualification at sale would have flagged the fragile ones early.
We ended it (non-payment) · 2 brands
Rabitat (untimely payments, ₹26L LTV, but our call), Formial (wouldn't agree to payment terms).
Partly preventable: stricter payment terms / advance billing up front avoids carrying non-payers for months.
Ghosted / went silent · 3 brands
Get Benefic, Oud Story, Not Rocket Science, RNR, stopped responding.
Hard to prevent but ghosting often masks dissatisfaction or budget stress, an early check-in cadence surfaces it before they vanish.
Section 3

The deep patterns, interaction effects, not single variables.

The real question isn't "do Seed brands churn?", it's "when does a profile churn, and what flips the outcome?" So here we test interactions (does variable B change what variable A does?) and compare churned brands against survivors with the same profile. At n=18 churns these are strong hypotheses, not proofs, cells get thin fast, and each is flagged.

Deep Pattern 1 · the strongest interaction in the data

Among your valuable brands, CRO is what separates retention from churn

Don't look at CRO across all brands, look at it only among big brands (₹50L–1Cr+), where it matters most:

Big + CRO: 6 brands → only 1 delivery-churned (Mia, Masakali, Happi Planet, One Science live; Rabitat not-our-fault). 5 of 6 retained.
Big + no-CRO: 5 brands → 3 delivery-churned (Kreo, Protouch, Wiselife; only Neesh & Visage survive).
So: a big brand without CRO is ~3× more likely to die on your delivery than a big brand with CRO.

Why: CRO gives a big, complex brand a shared success metric (conversion lift). Without it, the relationship is judged purely on flawless execution, and at that complexity, execution slips. Fix: never take a big brand on pure-tech; CRO isn't an upsell here, it's the retention mechanism. [n=11, directional]

Deep Pattern 2 · this one overturns an earlier assumption

High CS effort does not protect against churn, it signals strain

An earlier (shallower) cut suggested high-touch brands were stickiest. The interaction test says the opposite:

Delivery-fault churn by CS effort: High = 40% (2 of 5) · Medium = 38% (3 of 8) · Low = 18% (2 of 11).
High-effort brands churn on delivery at more than double the rate of low-effort ones.

Why: high CS effort isn't loyalty, it's a brand that's demanding and hard to deliver for. The effort is the symptom of complexity that then strains delivery into failure. Reframe: treat high CS effort as a risk flag, not a comfort signal, these brands need your best delivery resources, not just your most patient CSM. [n=24 eligible, directional but consistent]

Deep Pattern 3 · churned-vs-survivor matching

The Seed+big survivor proves the rule: CRO + the right delivery

Of 5 "Seed + sizable" brands, 4 churned on delivery and 1 survived (Happi Planet). Matching the survivor against the churned reveals the differentiator:

Survivor, Happi Planet: CRO + high CS, still live at 6.8mo.
Churned, Kreo, Protouch, Wiselife: all no-CRO. Conscious Chemist churned despite CRO+high-CS , the one exception, on a specific technical failure (animation/GIT).
Pattern: Seed+big is survivable, but only with CRO attached. Strip CRO and 3 of 3 died on delivery.

Fix: Seed + sizable is a high-value, high-risk profile, take it only with CRO and over-resourced delivery. It's your most expensive churn when it fails. [n=5, a hint, but a clean one]

Deep Pattern 4 · the brutal one

Nothing saves small brands, not CRO, not CS effort, not anything

We tested every combination on the 9 small/pre-revenue brands. The result is stark: 8 of 9 churned, across every permutation of CRO / no-CRO and high / low CS effort.

With CRO: NRS (gone), Zenue (gone), Bulbul (gone), Creslia (gone), Formial (gone), Get Benefic (gone), Tvishi (live, the lone survivor). Without CRO: Ethik (gone), Snuggly (gone).
CRO didn't save them. High CS didn't save them (Formial was high-CS, still gone). The only variable that predicts survival is not being small in the first place.

Fix: This isn't a delivery or offering problem, it's a selection problem. A size/budget floor at qualification is the only lever. Stop signing brands below viability; no amount of service rescues them. [n=9, 8 churned, strong]

The synthesis · what all four interactions converge on

A revenue-banded barbell, two segments, two failure modes

Segmenting the recurring book by the brand's own monthly revenue, controllable churn splits cleanly by band:

SegmentThresholdnFixable churnAvg LTV% of total LTVWhat retains them
Top≥₹50L/mo
(A/B band)
1117% with CRO
vs 60% without
₹7.8L81%CRO attach + delivery resourcing
Middle₹10–50L/mo
(C band)
7~17% w/ CRO
(pooled est.)
₹1.1L7%Behaves like top band: 3/3 CRO brands live
Bottom<₹10L/mo
(D/Pre-rev)
989% churned₹1.1L9%Nothing, qualification floor only
Top band (≥₹50L/mo · 11 brands · 81% of all LTV): churns on delivery. With CRO attached, delivery-fault churn is 17% (1 of 6); without CRO it jumps to 60% (3 of 5). High CS effort here is a risk flag (40% delivery churn), not a comfort.

Bottom band (<₹10L/mo · 9 brands · 9% of LTV): churns on everything8 of 9 left (89%), including 6 of 7 that had CRO and the 1 with high CS effort. No operational lever changed the outcome.

Middle band (₹10–50L/mo · 7 brands · 7% of LTV): behaves like the top band, so we estimate it by pooling: across all ₹10L+ brands, CRO cuts delivery churn to ~17% (1 of 6) vs ~50% without. In the mid-band specifically, all 3 CRO brands are live and both churns were no-CRO (Fflirtygo, Manetain). So a mid-band CRO brand should retain like a top-band one (~83% before QA), just at lower LTV (median ₹1.3L vs ₹7.8L). A legitimate secondary focus, not a question mark.

The number that frames the strategy: the top band holds 81% of all LTV in 11 brands, and attaching CRO cuts their delivery churn from 60% to 17%, a ~3.5× swing on 81% of your value. The bottom band is 9% of LTV and ~unsavable. So: attach CRO + resource delivery on the ≥₹50L/mo band; apply a ₹20L/mo revenue floor at qualification. That targets essentially all controllable churn where the value actually is.

Method & honesty note: These are interaction effects on 30 recurring brands / 18 churns. Cells get thin fast (Seed+big is n=5, the CS-effort tiers are n=5–11). Treat Pattern 1 (big×CRO) and Pattern 4 (small) as solid directional findings; Patterns 2 and 3 as strong hypotheses to confirm as the book grows. The value here is the direction each interaction points, and all four point consistently at the same two levers.

Part II
LTV Analysis, where the value is
The most important LTV fact

Your LTV is brutally concentrated.

2 brands = 64%
of all recurring LTV (Visage ₹41L + Rabitat ₹26L of ₹105L total)
Cumulative share of total recurring LTV, brands ranked high to low

Top 3 brands = 69% of all LTV. Top 5 = 74%. The other 25 recurring brands together make up barely a quarter. This single fact governs how to read everything below: any "average LTV" is dominated by one or two giants, so we lead with median (the typical brand) and treat average as a separate, outlier-sensitive number.

Why this matters: the median recurring brand pays Troopod just ~₹1.3L over its life. The business looks like ₹3.5L/brand on average, but that average is a mirage created by Visage and Rabitat. Strategy built on the average would be built on two brands you can't replicate at will (and one of which, Rabitat, you already fired).

LTV by dimension

Brand size is the only clean LTV driver.

Unlike churn, which had rich interaction effects, LTV is mostly a simple story: bigger brands pay more. But even here, median vs average matters.

By revenue band

LTV by brand revenue band, median (solid) vs average (faded). ₹L
median average

Band A (₹1Cr+) is the only band whose average (₹11.1L) towers over the rest, but its median is just ₹2.2L. Translation: most big brands pay normally; one (Visage) pays enormously. Below Band A, median LTV is flat at ₹0.9–1.7L regardless of size. Size buys you a shot at a giant, not a reliably higher LTV.

What actually drives LTV, tenure, not deal size

Median LTV by fee × tenure quadrant (₹L) · n in labels

If you correlate the two inputs against LTV, tenure wins clearly: 0.85 vs 0.56 for fee. How long a brand stays drives its value far more than how much it pays per month. The quadrants make it concrete: a high fee with short tenure (₹0.9L) earns less than a modest fee with long tenure (₹1.3L). A big deal that churns early is worth less than a small deal that sticks.

This collapses LTV and churn into one problem. Since tenure (= retention = not churning) is the dominant LTV driver, LTV is mostly a retention story wearing a revenue mask. The lever for higher LTV isn't charging more or chasing bigger brands, it's keeping brands longer. Which means every churn fix from the churn analysis (CRO for big brands, delivery quality, the qualification gate) is also your highest-leverage LTV move.

By offering, and a trap

LTV by offering · median vs average ₹L
median average
The trap: by average, No-CRO LTV (₹4.4L) beats Has-CRO (₹2.7L) , which would wrongly suggest "skip CRO for value." But that's entirely Visage, a ₹41L no-CRO outlier. By median, the two are basically equal (₹1.7L vs ₹1.3L). CRO doesn't cost you LTV, and remember from the churn analysis, it's what retains your big brands. Don't read this average as "CRO is worth less."

By category

LTV by category · median vs average ₹L · grey = n≤3
median average

"Skincare ₹11.4L" and "Baby/Kids ₹9.2L" look like premium categories by average, but their medians are ₹1.9L and ₹0.9L. Both averages are single-brand artifacts (a big skincare brand; Rabitat in Baby/Kids). By median, no category is meaningfully richer than another. Category is not an LTV lever, it matters for retention (Health/Wellness retains best), not value.

By funding & CS effort

Both are outlier-driven, not real signals. "Series D ₹41L" is literally just Visage (n=1). "Series A ₹26L" is just Rabitat (n=1). "High CS effort ₹6.4L" is Visage + Rabitat again. Strip the two giants and funding-stage and CS-effort have no clean LTV pattern at this data size. We flag these rather than chart them as findings, they'd just be re-plotting the same two brands.

The synthesis

What LTV actually tells you.

1. The typical brand is worth ~₹1.3L (median), not ₹3.5L (average). Plan capacity, pricing, and CAC tolerance around the median, not the mirage average.
2. LTV concentration is a fragility risk. 64% of value in 2 brands means losing one is catastrophic , and you already lost one (Rabitat). The strategic priority isn't just "get more LTV," it's "de-risk the concentration" by building a fatter middle of ₹2–5L brands.
3. Size is the only thing that reliably moves LTV, but it moves the ceiling (your shot at a giant), not the median. So pursue bigger brands for upside, but don't expect size to lift the typical deal.
4. The LTV and churn analyses point the same way. The brands worth most (big) are also the ones that churn on delivery (from the churn analysis). So your highest-LTV brands are your highest-risk, which means delivery quality isn't just a churn fix, it's LTV protection. Keeping one Visage-scale brand alive is worth more than winning ten median ones.

Method note: LTV = fee × tenure (+ upsell) for retainers, project fee for one-time. It is revenue LTV, not profit, the CS-cost overlay (Phase 3) will adjust it, and given high-effort brands carry the most LTV, profit may compress the top. Median is the honest central measure here given the extreme concentration; averages are shown only to expose where they mislead.

One-time book

One-time projects, a different business.

13 one-time projects, analysed on their own terms. The economics are fundamentally different from retainers: value is a single fee (no tenure compounding), and "success" means the project lands clean and ideally leads to a repeat or referral, not month-over-month retention. So churn and LTV mean different things here.

₹9.9L
Total booked
(vs ₹105L recurring)
₹0.6L
Median project fee
(avg ₹0.8L)
6 of 11
active projects
overrunning duration
One-time · churn

"Churn" barely applies, but scope does.

Only 2 of 13 projects "churned" in the delivery sense (Threesixty, unhappy with designs; Lumov, poor execution, both delivery-fault). The other 11 are live or completed. So churn isn't the one-time risk. The real risk is the project never ending.

Expected vs actual duration (months) · one-time projects
expected actual
The core one-time problem. 6 of 11 active projects run well past their promised duration, Vetycos (1mo → 9.3mo, 9×), Dawn (1mo → 9mo, 9×), Indoin (4×). A one-time fee assumes bounded work; when a 1-month project runs 9 months, the effective rate collapses and "finished" deals silently consume delivery capacity. This is the one-time equivalent of churn, value leaking through unbounded time.

How to solve: hard scope + time boundaries written into one-time contracts; a defined "done" with change-orders for anything beyond; and a deliberate decision at overrun, either bill it as a retainer or close it. Don't let project scope drift indefinitely.

One-time · LTV

Low value per deal, and it doesn't compound.

One-time LTV is just the fee, no tenure multiplier, so it's structurally small: median ₹0.6L, max ₹2.4L (Puracy). The whole book (₹9.9L) is less than a tenth of the recurring book, and 6 of 13 clients are pre-revenue brands getting a site built from scratch.

One-time fee by project (₹L), ranked
Where one-time LTV is actually worth it: the top of the range, Puracy (₹2.4L, established US brand), Beenext (₹1.5L, VC fund), VGR (₹0.8L). These are big, established logos paying real fees for a bounded build. The bottom (Green Gainz, Threesixty, Momentum at ₹0.3–0.4L) is small work for small brands, low fee and high overrun risk.
One-time · who buys it

Mostly institutional one-offs and pre-revenue builds.

The one-time book splits into two clear types, and neither is a recurring-ICP brand:

Type 1 · big established logos (the good ones)

Bounded builds for credible names

Puracy (US, acquired), Beenext (VC fund), Campus-type names, VGR. They pay the biggest fees, the logo has credibility value, and they're not expecting a retainer. Worth pursuing opportunistically.

Type 2 · pre-revenue / small brands building a first site

Small new brands, high overrun risk, low fee

6 of 13 are pre-revenue (Dawn, Vetycos building from scratch). These are exactly the brands that overrun the most (Vetycos 9×, Dawn 9×), a new brand's site is never "done." Low fee, high time-drain. Take with caution and hard boundaries, or not at all.

Deep dive · the fixable churn

Delivery & sales, the core issues, and how to solve them.

Fixable churn is ~38% of the eligible book, split into delivery fault (29%, 7 brands) and sales fault (8%, 3 brands). Unlike the not-our-fault churn, every one of these traces to a specific, repeatable failure. Below: each root cause, the brands it cost you, and the concrete fix.

DELIVERY FAULT · 7 brands · 29%
SALES · 3 · 8%

The fixable churn, by owner. Delivery is the larger share and the dominant churn driver overall.

DDelivery fault, 3 root causes
All 7 delivery churns trace to one of three failures. "Went in-house" appears often but is a consequence of these, not a cause.
A
Timelines slipping
3 brands · ₹5.4L LTV lost
What happened
Fflirtygo (timelines, then hired in-house dev) · Wiselife (slipped + work ran 5× the offered hours) · Rrumi (timelines slipped alongside bugs). Deadlines moved, scope crept, and the client lost confidence.
How to solve
  • Scoped sprint timelines with built-in buffer, agreed in writing at kickoff
  • Weekly "are we on track" status to the client, kill the surprise factor
  • Written change-orders for anything beyond scope (stops the 5×-hours creep)
  • Flag any slip the moment it appears, not at the deadline
B
Bugs shipped to production
3 brands · the most damaging & most preventable
What happened
Kreo (a Cloud SQL billing bug went live) · Rrumi (recurring 404s + AWS data loss) · Conscious Chemist (preview-looked-fine but live broke, a staging-vs-prod GIT discrepancy). Things that worked in preview failed in production.
How to solve
  • Pre-deploy QA checklist, nothing goes live without a second set of eyes
  • Staging-vs-production parity check (the Conscious Chemist failure mode)
  • Automated error/uptime monitoring to catch 404s before the client does
  • Backup-before-deploy discipline (the Rrumi data-loss class is unforgivable to a client)
C
Quality below expectation
2 brands
What happened
Protouch (sub-standard experience, persistent sticky-cart issue) · Bulbul (general dissatisfaction → took the team in-house). The work functioned but didn't meet the bar.
How to solve
  • Design/quality sign-off gates at each milestone, client approves before build proceeds
  • Catch "this isn't it" at mockup stage, not after the build is done
  • A defined quality bar (performance, UX checks) every deliverable must clear
SSales fault, 2 root causes
These brands didn't leave because delivery failed, they left because the sale was wrong. The fix lives at qualification, not execution.
D
Sold without budget to deliver value
2 brands
What happened
Zenue (disputes over invoice amount & consumption hours, the budget couldn't fund the work needed) · Manetain (mis-scoped, too small to move the needle). Signed brands whose budget couldn't fund enough work to see results.
How to solve
  • The ₹20L/mo revenue floor, the master gate from the verdict tab
  • Minimum-engagement size: don't sign if the fee can't fund enough hours to move metrics
  • Set in writing at sale what the fee does and doesn't buy
E
Scope / expectation mismatch at sale
2 brands (overlaps D)
What happened
Snuggly Monkey & Manetain, what was promised at the sales stage didn't match what delivery could realistically do, so the gap surfaced fast (both churned inside 3.2 months).
How to solve
  • A sales→delivery handoff doc: what was promised, what's in scope, what success looks like
  • Delivery reviews the scope before the deal closes, no inherited promises it can't keep
  • Set realistic expectations on timeline and outcomes up front
What to fix first
Pre-deploy QA checklist + staging-prod parity (cause B)
Highest damage, lowest effort. Bugs and data-loss in production are the most unforgivable churn and the most preventable. Start here.
DELIVERY · this week
Written timelines + weekly status + change-orders (cause A)
Kills both the deadline-slip churn and the 5×-hours scope creep. Process, not headcount.
DELIVERY · this month
Milestone quality sign-offs (cause C)
Client approves at each stage; catch dissatisfaction at mockup, not post-build.
DELIVERY · this month
₹20L floor + sales→delivery handoff doc (causes D, E)
One gate plus one document removes most sales-fault churn. Owned by sales, enforced at qualification.
SALES · next deal
The deep-dive in one line
Of the 10 fixable churns, 6 are pure delivery process (timelines, QA, sign-offs, no extra headcount) and 4 are a qualification gate. None require hiring or new tools, just discipline at deploy and discipline at sign. That's what converts 38% retention toward the ~92% ceiling.
Phase 1 · The verdict

The ICP, and how high it can go.

Retention is the master lever, it drives churn directly and drives LTV (tenure correlates 0.85 with value vs 0.56 for deal size). The ICP depends on one thing: whether delivery is fixed. Two states below, then the retention ladder, the book verdict, and next steps.

1ICP today, before solving any churn
Nothing fixed yet, delivery, sales, and client-side churn all still in play. So the gate must be tight, screen out the brands prone to any of it up front.

FOCUS

  • ≥₹20L/mo revenue + not Seed-funded (the ICP)
  • + ≥₹50L & CRO for extra LTV upside
  • Health/Wellness & Skincare
  • not-Seed is the strongest single filter, lifts both axes
75% retention (≥₹20L+not-Seed, n=8)₹7.8L avg LTV

GATE OUT

  • <₹20L/mo revenue (the master gate)
  • Seed-funded, even at ≥₹50L (4 of 5 churn)
  • No-CRO on big brands (60% delivery churn)
  • Unbounded one-time projects
38% retention, no gate8/9 small churned

Tight gate, smaller book, but the only way to protect retention while churn is still unsolved. ≥₹20L + not-Seed is the ICP, 75% retention and ₹7.8L avg LTV across the cohort, the best balance of both axes. not-Seed beats CRO as the second filter (CRO alone gives 67%). A ₹20L floor (vs ₹10L) carries ~₹1L more LTV per brand and gates out one more sales-fault churn, for ~3 points less retention, a worthwhile trade. Even this is capped by unfixed delivery churn, which State 2 removes.

2ICP after solving delivery & sales churn
Sales churn is solved by the ₹20L floor (all three sales-fault brands were sub-₹20L), delivery churn by QA. Two levers, but the floor handles sales for free, so you can relax everything else.

FOCUS (widened)

  • ≥₹20L/mo revenue (floor only, drop not-Seed)
  • CRO preferred, no longer mandatory
  • Seed now viable, even high-value (their churn was 100% delivery)
  • More volume at the same retention
92% retention (₹20L floor + QA, n=13)₹6.0L avg LTV

GATE OUT (only)

  • <₹20L/mo revenue, still the one hard rule
  • Below it, brands churn for non-delivery reasons QA can't fix
  • That's the whole gate now
  • Unbounded one-time still capped
1 rule just the ₹20L floor

With the ₹20L floor in place, delivery and sales churn are both handled, the floor structurally removes sales-fault (all three were sub-₹20L), and QA removes delivery-fault. Retention reaches 92% with essentially only not-our-fault churn left, while the book stays wide (n=13). One floor + delivery QA does it.

3How retention climbs as you gate
Read two ways: the best gate today vs the best once delivery is fixed. The gate loosens as delivery improves.
GatenRetention today+ delivery fixedAvg LTV
No gate (all recurring)2438%67%₹3.5L
≥₹20L revenue floor1354%92%₹6.0L
≥₹20L + not-Seed (the ICP)875% ★ best balance88%₹7.8L
≥₹20L + CRO667%83%₹4.4L
Two reads of "best", depending on whether delivery is fixed. Today (no delivery fix): ≥₹20L + not-Seed is the ICP gate at 75% with the highest LTV (₹7.8L), not-Seed screens out the volatility you can't yet absorb. Once delivery QA is in: the plain ≥₹20L floor reaches 92% and keeps the most volume (n=13), because QA now saves the delivery churns the tighter gate was avoiding. So the gate loosens as delivery improves. ≥₹20L + not-Seed tops the LTV column too (₹7.8L), so it wins on both axes, the best retention-plus-value balance. (We tested finer bands: ₹10L+not-Seed edges retention slightly at 78% but ₹20L carries ~₹1L more LTV per brand and gates out one more sales-fault churn, so ₹20L is the pick.)
4Recurring vs one-time

RECURRING · the engine

₹105L
91% of revenue
  • Median LTV ₹1.3L → ₹3L+ achievable
  • Value compounds via retention
  • Scale this
vs

ONE-TIME · the tool

₹9.9L
9% of revenue
  • Median fee ₹0.6L, doesn't compound
  • 6 of 11 overrun duration
  • Cash & logos only
Final conclusion
Two moves take retention from 38% to ~92% and median LTV from ₹1.3L toward ₹3L+: gate harder at the front, execute harder on delivery.
  • 1Gate at ≥₹20L/mo + not-Seed. The ICP, 75% retention and ₹7.8L LTV today, the best balance of both axes.
  • 2Fix delivery. Then the gate can relax to the ₹20L floor alone and retention reaches ~92%, while the book stays wide.
  • 3Layer ≥₹50L + CRO for LTV. That's where 81% of value sits, the upside on top of the retention base, not the base itself.
  • 4One-time is a tool, not the engine. Use it for cash and logos with hard scope limits, never as the growth motion.
!Limitations
  • · Small cells. The ICP sub-segments are 4–9 brands. These are directional patterns, not statistical proof, the projections (67%, 92%) are modelled, not guaranteed.
  • · Retention map, not acquisition. No CAC, win-rate, sales-cycle, or market-size data, so this says what to keep and deepen, not how cheaply to win or whether the segment is big enough to scale into.
  • · Survivorship. Partially mitigated (under-3-month brands flagged out), not fully solved, true cohort decay curves need ~80–100 brands. Re-run when the book is ~3× larger.
  • · Theme tags are interpretive. Top-level fault (delivery/sales/not-us) is team-validated, the finer sub-tags (timeline/bug/quality) are Claude's reading of the reason text.
Next steps
Gate at ₹20L/mo revenue + not Seed-funded
The ICP gate, 75% retention and ₹7.8L LTV. The ₹20L floor removes all sales-fault churn and most not-our-fault, not-Seed removes the volatility that drives delivery churn.
SALES · now
Build delivery QA (see the Delivery & Sales Fixes tab)
Pre-deploy checklist, staging-prod parity, milestone sign-offs, weekly status. Takes the gated book to ~92% and unlocks the big-Seed segment.
DELIVERY · the ceiling-maker
Attach CRO to every ≥₹50L brand (the LTV layer)
≥₹50L brands hold 81% of LTV, CRO cuts their delivery churn 60%→17%. This is where the value concentrates, layered on top of the ₹20L + not-Seed retention base.
SALES + DELIVERY
Add a month 3–4 retention review
Most churn lands in the 3–9 month window. A mid-relationship results review catches drift before it becomes an exit.
CSM
Re-run this analysis at ~80–100 brands
Confirms the directional patterns with cohort-grade rigour, and adds acquisition data (CAC, win-rate) for a full ICP.
PHASE 2