QuickClaimBack — deduction catcher for quick-commerce brands

1. One-liner

Reads a brand’s Blinkit, Zepto and Instamart settlement files and files every wrong deduction before the window shuts.

2. Trend signal — why now?

Quick commerce is now the primary channel for a huge slice of Indian D2C. ZOFF Foods says qcommerce is 65–70% of their business. Zepto raised $400M (CalPERS-led) and Swiggy pulled ~₹10,000 crore via QIP in the last cycle — the GMV is real and growing, and platforms plan 2,000–2,500 new dark stores across metros in 2026.

But the money going out is a black box. Every settlement statement carries deductions for returns, damage claims, pick errors, promo co-funding, platform and visibility fees — and, critically, “deductions are applied to the settlement for the period in which the incident is recorded, not necessarily the period in which the original order was placed,” which creates timing mismatches that make manual reconciliation break. On Instamart specifically, “deduction transparency on the seller dashboard is the weakest of the three platforms, and reconciling settlement statements manually against order data is a common operational burden reported by sellers” (confetti.design, 2026). Brands are doing this in spreadsheets, or not doing it at all.

The pattern is proven in the West: on Amazon, the average Indian seller recovers 1–3% of turnover just by catching weight/handling errors and unreturned items — and a whole tool category (Unicommerce UniReco, eVanik, Gonukkad, FireAI, SellerRocket) exists to do it. That category is Amazon/Flipkart-first. Quick commerce — with its own settlement format, its own deduction types, its own dispute windows — is bolted on as an afterthought, if at all. That gap is the opening.

Provenance:

• Signal 1 (demand): “Reconciling settlement statements manually against order data is a common operational burden reported by Instamart sellers; deduction transparency is the weakest of the three platforms.” — https://confetti.design/blog/blinkit-vs-zepto-vs-swiggy-instamart-for-selling — 2026
• Signal 2 (feasibility/economic): Indian marketplace sellers recover 1–3% of turnover by catching fee/deduction errors; a funded recon-tool category exists (UniReco, eVanik, FireAI) but is Amazon/Flipkart-first. — https://unicommerce.com/blog/marketplace-payment-reconciliation-the-invisible-margin-leak-killing-your-profitability/ — 2026
• Signal 3 (economic): Qcommerce is 65–70% of GMV for brands like ZOFF; Zepto raised $400M, Swiggy ~₹10,000cr; 2,000–2,500 new dark stores planned 2026 — the channel where deductions happen is scaling fast. — https://globalwebsters.com/blog/quick-commerce-unit-economics-blinkit-zepto-instamart/ — 2026 Category: Geographic arbitrage

3. The opportunity

The West solved “the marketplace deducted money it shouldn’t have” — SAFE-T claims, chargeback recovery tools, a whole cottage industry. India solved it for Amazon and Flipkart. Nobody has solved it for quick commerce, which is now where the volume — and therefore the leakage — actually lives.

The incumbents (UniReco, eVanik, BUSY, Gonukkad) are e-commerce reconciliation engines. Their model is COD/prepaid settlement matching for Amazon/Flipkart shipment-level data. Quick commerce breaks their assumptions: settlements are weekly (Blinkit) or bi-weekly (Zepto/Instamart), deductions land in the wrong period, damage/pick-error passbacks come from dark-store operators not the brand’s own returns, promo co-funding is negotiated per-campaign, and each platform’s statement format is different and often a semi-structured PDF/CSV dump. A generic recon tool shows you a mismatch number. It does not tell you which deduction is disputable, why, and it does not draft and file the claim inside the platform’s dispute window.

QuickClaimBack does exactly the narrow thing: ingest the QC settlement files, classify every deduction line, surface the ones that are wrong or unsupported, and produce the claim — ready to file, before the window closes. Not a dashboard. A recovery engine that turns leaked margin back into cash.

4. Target market

• Primary customer: Finance/ops lead (or the founder wearing that hat) at an Indian D2C/FMCG brand doing ₹2–50 crore annual GMV on quick commerce — packaged food, personal care, beverages, health snacks. The brands for whom qcommerce is 40%+ of revenue and margins are already thin.
• Why they buy: In their words — a founder spent ₹6 lakh getting listed across all three platforms and three months later had ₹1.2 lakh in combined revenue; another says qcommerce is 65–70% of business but they spend 10–15% of GMV on the channel “with margins continuing to compress.” When platform take is 35–50% and gross margin has to clear 70% just to survive, a 1–3% leakage in wrong deductions is the difference between a positive and negative contribution margin. They feel this every settlement cycle.
• Rough TAM reasoning: India has thousands of D2C brands actively selling on Blinkit/Zepto/Instamart, with the count rising as dark stores expand. Even a serviceable slice of a few thousand mid-size brands paying ₹1,000–3,000/mo is a comfortable sub-$5M ARR business.
• Why now for them: Qcommerce just crossed from “experiment” to “primary channel” for these brands in the last 12–18 months. The deduction pain scaled with it, but no dedicated tool caught up. And the platforms enforce short dispute windows — money not claimed on time is gone — so a manual spreadsheet process is actively losing them cash every week.

5. Product sketch (MVP)

• Upload or auto-pull weekly/bi-weekly settlement files from Blinkit, Zepto and Instamart (start with file upload; API/portal pull where available).
• Line-item deduction classification: commission, logistics, damage, pick error, return, promo co-funding, penalty, platform/visibility fee — normalized across the three platforms’ different formats.
• Discrepancy flags: deductions with no matching order, duplicate deductions, deductions billed in the wrong period, promo co-funding above the agreed rate, damage/return claims exceeding order value.
• A “disputable pile” ranked by rupee value and by days left in the dispute window — so the biggest, most-urgent claims surface first.
• Auto-drafted dispute packet per claim: the reference IDs, the reason, the expected reversal amount, formatted for that platform’s seller-support process.
• Recovery tracker: what was filed, what was reversed, what’s pending, rupees recovered to date.
• Weekly “leakage report” e-mail: “₹X deducted this cycle, ₹Y disputable, Z days to file.”

6. AI angle — what’s load-bearing

Remove the AI and this is a spreadsheet no one maintains. The load-bearing work is (a) parsing three different, messy, semi-structured settlement formats — including PDF dumps and inconsistent CSV columns that change without notice — into a normalized deduction ledger, and (b) classifying each deduction line and reasoning about whether it’s supportable given the matched order data and the platform’s stated fee rules. That’s exactly what cheap LLM parsing + a classification layer got good at in 2025–26; before that you’d have hand-built a brittle parser per platform per format change. The AI also drafts the dispute language per claim. Without it, the product is manual reconciliation — which is precisely the burden brands already can’t keep up with.

7. Localization angle

This is the localization play — it’s the whole idea. The Western marketplace-deduction-recovery pattern (Amazon SAFE-T tools) applied to a channel structure that only exists at this scale in India: quick commerce with dark-store-level deductions, weekly settlement cycles, per-platform statement formats, and India-specific mechanics like GST on deductions and promo co-funding negotiated with a Category Manager. A generic global recon tool cannot read a Blinkit settlement PDF or know that Instamart’s dashboard hides the deduction detail. Pricing is India-native: ₹999–2,999/mo tiers work where a $49 tool wouldn’t, and a % -of-recovered success fee aligns with brands that don’t want another fixed SaaS bill.

8. Business model — path to $1M–$5M ARR

• Pricing: Two-part. Base SaaS ₹1,499/mo per brand (up to 3 platforms), plus a 15–20% success fee on rupees actually recovered. Success fee makes the pitch risk-free — “we only win when you get money back.”
• ACV: ~₹40,000–70,000/year blended (₹18K base + recovery share). For a brand doing ₹5cr QC GMV, 1.5% leakage caught = ₹7.5L found, 15% share = ~₹1.1L — the tool pays for itself many times over, so ACV skews up for larger brands.
• Rough math to $1M ARR: ~₹8.3cr ARR ≈ 1,200 brands at ₹70K blended ACV. Achievable within the mid-size QC-brand pool.
• Rough math to $5M ARR: ~₹42cr — needs ~3,000–4,000 brands or a meaningful move up-market to larger FMCG players plus a second channel (adding Flipkart Minutes, BigBasket Now, and general e-comm recon as an upsell).
• Expansion path: Start with deduction recovery → add settlement forecasting (cash you’re owed and when) → add fee-benchmarking (“your promo co-funding is 4pts above category”) → become the finance layer for a brand’s entire qcommerce P&L.

9. Go-to-market wedge — first 100 customers

• Free audit as the hook. Offer any brand a one-time free deduction audit: they send one month of settlement files, we return “here’s ₹X you can still claim.” This is falsifiable, concrete, and self-qualifying — the number sells the subscription. This is the whole wedge.
• Scrape the seller lists. QC brands are discoverable — they list on Blinkit/Zepto, they appear in D2C directories, they post GMV milestones on LinkedIn. Build a list of 1,500 mid-size brands, send a personalized “we found ₹___ leaking in your category” cold e-mail/DM tied to the free audit.
• QC-ops agencies and consultants (the confetti.design / Base / decodegrowth crowd advising brands on Blinkit/Zepto) already sit between us and the buyer. Rev-share referral: they bring the brand, we split the recovery fee. 3–5 such partners = pipeline.
• Category-manager and D2C founder communities on LinkedIn and in operator WhatsApp/Slack groups — post the anonymized “we recovered ₹X across N brands last month” number monthly. Recovery numbers are shareable proof; leakage is a story every QC founder recognizes.
• Land on the pain moment: target brands right after a bad settlement cycle — content around “why your Blinkit payout was short this week” pulls the exact person searching at the exact moment.

10. Build complexity — justification

Medium. Off-the-shelf: LLM parsing/classification, standard web stack, file ingestion, a claim-tracking workflow. The custom work is the per-platform settlement normalization and the order-matching logic across timing mismatches — non-trivial and it needs babysitting when platforms change their statement formats. Realistically 3–4 months for a small team to a credible v1 covering all three platforms; a single-platform (Blinkit-only) MVP could ship in ~8 weeks to start proving recovery.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	Reading a brand’s own settlement data and filing legitimate disputes on their behalf.
Ethical — no harm / dark patterns	✅	Recovers money genuinely owed; disputes must be truthful or they get rejected by the platform anyway.
Market exists (evidence above)	✅	Manual recon is a documented burden; a funded recon-tool category exists for adjacent channels.
1–5 person team can build this	✅	Medium complexity, off-the-shelf AI, phased per-platform rollout.
Launchable with <$50K / ₹40L	✅	Solo/pair build, no capital-heavy dependencies; inference cost is modest at low volume.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	16/20	Felt every settlement cycle; directly hits contribution margin on a channel that’s 40–70% of revenue. Not quite hair-on-fire because many brands have half-given-up rather than actively hunting a fix.
Demand evidence	15	12/15	Documented manual-recon burden + proven adjacent recovery economics (1–3% of turnover) + funded recon category. Missing: a direct verbatim of a QC brand asking for this specific tool.
Build feasibility	15	11/15	Doable in 3–4 months, but per-platform format normalization is fiddly and breaks when platforms change formats — ongoing maintenance tax.
Distribution clarity	15	12/15	Free-audit hook is concrete and self-qualifying; brand lists are scrapeable; agency referral path is real. Conversion from free audit to paid still unproven.
Revenue mechanics	15	12/15	Success-fee model aligns incentives and de-risks the buy; base SaaS gives floor. Depends on recoverable amounts being large enough — 1–3% benchmark supports it.
Time to first revenue	10	7/10	Free audit → paid can close in weeks once recovery is shown; success fee earns as soon as first claim reverses. Platform dispute cycles add lag before the first reversal lands.
Defensibility	10	6/10	Moat is accumulated per-platform parsing rules + a recovery-outcome dataset (which deductions actually get reversed) + workflow lock-in on the finance team. Copyable, but a head start compounds.
Total	100	76/100

13. Qualitative modifiers

Founder-fit tags

domain-expertise-required · technical-heavy — you need someone who understands qcommerce settlement mechanics cold (or a category-manager advisor) plus the engineering to parse and match messy financial data reliably.

Key assumptions to validate (3–5)

1. Assumption: Wrong/disputable deductions really are ~1–3% of QC turnover for mid-size brands. How to test: Run free audits on 10–15 brands’ one-month settlement files; measure actual disputable rupees found.
2. Assumption: Platforms will reverse a meaningful share of well-formed disputes within their windows. How to test: File 30–50 real claims across the three platforms; measure reversal rate and time-to-reversal.
3. Assumption: Brands will pay a 15–20% success fee (not just want the free audit). How to test: Convert the free-audit cohort — put the paid contract in front of them once they see the recovered number.
4. Assumption: Settlement formats are stable enough that parsing maintenance doesn’t eat all the margin. How to test: Track format-change frequency across three platforms over 8–12 weeks.

Risk flags

1. Platform dependency: The whole product reads platform-controlled settlement files and files through platform-controlled dispute channels. A format change or a hostile policy change can break ingestion or throttle disputes overnight.
2. Incumbent expansion: UniReco/eVanik/FireAI could add real QC support and use their existing e-comm seller base to distribute — the window to build the QC-specific moat is finite.
3. Recoverable-amount risk: If real disputable leakage turns out closer to 0.5% than 1–3% for most brands, the success-fee economics thin out and the pitch weakens.
4. Dispute-cycle lag: Reversals depend on platform seller-support responsiveness; slow reversals delay the success-fee revenue and dampen the “risk-free” pitch.

14. Structured verdict

Score:                  76/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       Technical founder with a qcommerce category-ops advisor (or an ex-category-manager co-founder)
Time to revenue:        6–10 weeks (free audit → paid; success fee lands on first reversal)
Capital to launch:      ₹4–8 lakh ($5–10K)
Top 3 assumptions to validate first:
  1. Disputable deductions are ~1–3% of QC turnover — free-audit 10–15 brands and measure.
  2. Platforms reverse well-formed disputes at a usable rate — file 30–50 real claims, measure reversal %.
  3. Brands convert from free audit to a 15–20% success fee — put the contract in front of the audit cohort.
Kill criteria:
  - Abandon if free audits on 15 brands surface <0.5% of turnover in disputable deductions.
  - Abandon if platform dispute reversal rate on well-formed claims is <20%.
  - Abandon if an incumbent recon tool ships credible three-platform QC support before your v1.

15. Next step — 1-week validation sprint

• Day 1–2: Recruit 8–10 QC brands (via D2C LinkedIn groups and QC-ops consultants) willing to hand over one month of Blinkit/Zepto/Instamart settlement files for a free audit. No product yet — do it by hand.
• Day 3–4: Manually parse and match the files against order data. Catalog every deduction, flag the disputable ones, and total the recoverable rupees per brand. File 5–10 real disputes to test the reversal channel.
• Day 5: Decide. Go if median disputable leakage across the cohort is ≥1% of QC turnover and at least 3 of the audited brands say yes to a 15–20% success-fee contract. No-go if the leakage is trivial or nobody will pay once they’ve seen the free number.

The falsifiable result: a real rupee figure of disputable deductions per brand, and a real count of brands who signed a paid contract after seeing it. Not “they liked it” — a recovered-rupees number and a signature.