alirezarezvani/claude-skills / finance/skills/stock-analysis
stock-analysis
UnverifiedAnalyze, research or value a listed stock or company (US, global, NSE or BSE) with verified data, ratios, scores and valuation. Use whenever asked if a stock is a buy, what it is worth, or how a company looks.
references/sectors/retail-ecommerce.md
View on GitHub# Retail chains, e-commerce, marketplaces and quick commerce — sector playbook
Use this when: the company sells goods to end consumers through a store estate, a website/app, or a third-party platform — supermarkets and hypermarkets, value and specialty apparel chains, department stores, jewellery and electronics retail, horizontal and vertical marketplaces, 1P inventory-led e-commerce, food delivery, quick commerce and dark-store networks, and omni-channel hybrids of these.
This sector breaks the generic checklist more comprehensively than almost any other, because two accounting standards (IFRS 15 / Ind AS 115 on principal-vs-agent, and IFRS 16 / Ind AS 116 on leases) and one structural feature (negative working capital) between them corrupt revenue, margin, leverage, returns and free cash flow simultaneously. On top of that, the generic list contains **no operating KPI at all** — nothing that separates growth from opening stores versus growth from existing stores, or growth from more users versus more spend per user. Your job is to normalise everything to a comparable unit (GMV, or gross profit), rebuild unit economics from the notes, and judge the business on like-for-like and cohort behaviour rather than on the consolidated P&L.
## Contents
1. [Why the generic ratio set fails here](#why-the-generic-ratio-set-fails-here)
2. [The metrics that actually matter](#the-metrics-that-actually-matter)
3. [How to value companies in this sector](#how-to-value-companies-in-this-sector)
4. [Peer set construction](#peer-set-construction)
5. [Sector-specific red flags](#sector-specific-red-flags)
6. [Cycle and structural context](#cycle-and-structural-context)
7. [India vs global notes](#india-vs-global-notes)
8. [Checklist](#checklist)
---
## Why the generic ratio set fails here
Work through all seven before quoting any standard ratio. In most cases the correct action is to restate the metric onto a common basis, or suppress it entirely — not to caveat it.
**Revenue is not a comparable unit, so every revenue-derived ratio is arithmetically broken.** Under IFRS 15 / Ind AS 115 a **principal** (1P, inventory-led: a supermarket chain, an own-stock beauty e-tailer) books the full customer ticket as revenue. An **agent** (3P marketplace, food delivery, most horizontal platforms) books only its commission. The same ₹100 of consumer spend becomes ₹100 of revenue for one and ₹18 for the other. Consequently a 4% operating margin at a grocer (earned on gross sales) is a far better business than a 4% margin at a marketplace (earned on net commission), and OPM, EV/Sales and revenue-growth comparisons across the two models are meaningless unless normalised to GMV or to gross profit. Worse, companies **flip** between principal and agent treatment for parts of their business, generating optical revenue growth or collapse with zero economic change. Never compare margins across 1P and 3P without stating the basis.
**IFRS 16 / Ind AS 116 destroys EBITDA, D/E and ROCE comparability.** Retail is a leasehold industry. Since FY20 (India) / 2019 (IFRS; US GAAP ASC 842 for lessee balance sheets), rent is split into right-of-use depreciation plus lease interest. Three consequences: EBITDA margins for leased retailers jumped roughly 300–800bps overnight with no trading improvement; lease liabilities inflate reported "debt", so leased apparel and department-store chains routinely print D/E of 1–3x that is almost entirely rent; and ROU assets inflate capital employed, so ROCE fell. Pre-FY20 and post-FY20 EBITDA margin, D/E and ROCE series **are not the same metric**. A "D/E < 1" screen mechanically rejects every leased-store retailer and mechanically passes owned-property retailers regardless of economics — the exact opposite of the intended signal.
**FCF and working-capital signals are inverted.** Grocery, quick commerce and marketplaces run structurally negative working capital: the customer pays instantly, the supplier is paid in 45–90 days. Growth therefore *generates* cash — "float" — so FCF looks superb precisely while the business is consuming economic capital, and FCF **collapses when growth stops** even though profitability is unchanged. Standard FCF also ignores lease payments, a real contractual rent-like outflow that IFRS 16 parks in financing activities, making leased retailers look far more cash-generative than owned ones. Always compute FCF **after** lease payments, and model the working-capital unwind at lower growth.
**Current ratio and liquidity rules are inverted for the same reason.** A well-run grocer or quick-commerce operator should have a current ratio below 1.0 — that is supplier float, not distress. A "current ratio > 1.5" rule rewards the retailer that finances its own inventory and penalises the one with channel power. Judge liquidity on the cash conversion cycle trend, undrawn facilities and lease-adjusted fixed-charge cover instead.
**ROCE is gameable in both directions.** Franchise-led and asset-light formats (brand licensors, franchise apparel) show 40%+ ROCE on almost no capital, while post-IPO platforms sitting on very large idle cash balances show depressed ROCE that says nothing about the operating business. Owned-property retailers look permanently capital-inefficient versus leased peers doing the identical trade. Compute ROCE on **operating capital excluding surplus cash**, on a lease-consistent basis, and pair it with **return on incremental capital** (new-store payback) — average ROCE tells you about yesterday's estate, not about the rollout that the multiple is paying for.
**P/E and P/B are frequently undefined or irrelevant.** Most scaled platforms were loss-making for a decade, so P/E is negative or a meaningless four-digit number through the crossover year and mean-reverts violently for two years afterward. P/B fails because the assets that matter — brand, customer cohorts, seller network, delivery density, private-label IP, catchment locations — are unbooked, while accumulated losses can push book equity toward zero or negative. Conversely, Indian retail P/Es of 60–120x are not "expensive" in isolation: they price a decade-long store-rollout runway that a static P/E cannot express. Judge the multiple against the implied store count and density, not against the market.
**EV/EBITDA is usable but is the single most common site of a basis error.** The unforgivable and extremely frequent mistake is mixing bases: adding lease liabilities into enterprise value while using a pre-IFRS 16 EBITDA (double-counting rent), or comparing a post-IFRS 16 multiple against a pre-FY20 historical average. Pick one basis, state it, and apply it to every peer.
**The generic list has no operating KPIs, and seasonality makes the ones it has misleading.** A retailer can post 25% revenue growth with negative like-for-like sales, falling sales density and lengthening store payback — the checklist sees only the 25%. Add Q3 festive/holiday concentration (30–40% of annual profit for Indian apparel, gifting and jewellery retail; similar for Western Q4) and annualising a single quarter is actively misleading. TTM and LFL are mandatory, never optional. Finally, Indian new-age platforms headline an "Adjusted EBITDA" that excludes ESOP charges running 3–8% of revenue — a real, recurring, dilutive shareholder cost that the generic ratios silently absorb.
---
## The metrics that actually matter
Ranges are **indicative only**. They vary by market, format, cycle and reporting period, and post-IFRS 16 versus pre-IFRS 16 basis changes several of them materially. A company's own multi-year history and its direct sub-format peers always override any absolute band below.
| Metric | Definition / how to compute | Indicative healthy range | Why it matters |
|---|---|---|---|
| Like-for-like / same-store sales growth (LFL, SSSG), decomposed | Sales growth from stores open the full comparable prior period, excluding openings, closures, currency and acquisitions. Decompose into footfall/transactions × conversion × average basket value × items per basket. Check whether online and click-and-collect sit inside the LFL base — most Indian retailers include them, which flatters it. | Developed markets 2–5% grocery, 3–6% specialty. India 6–12% for a healthy format; best-in-class value formats have run 15–25% in expansion phases. Below food/CPI inflation = real volume decline. Negative for 2+ consecutive quarters is structural. | The single most important retail metric and it is absent from the generic list. Total revenue growth can be manufactured indefinitely by opening stores at declining returns; LFL isolates whether the existing asset base is getting healthier. Volume-led LFL is high quality; price-led LFL during inflation reverses. Falling footfall masked by a rising ticket is the classic pre-decline signature. |
| Sales density — revenue per sq ft, per store, per dark store | Annualised revenue ÷ average carpet/retail area. Track the level, the trend, and the **gap between mature-store and new-store density**. Analogues: GMV per dark store per day (quick commerce), sales per checkout (grocery), GMV per active seller (marketplace). | India: best-in-class value grocery ~₹32,000–36,000/sqft/yr; value apparel ~₹10,000–14,000; department stores ~₹9,000–12,000. US: mass-market big-box ~$500–700/sqft, warehouse clubs $1,000–1,500+. Quick commerce: a dark store generally needs ~₹8–10 lakh/day GMV to be contribution-positive. | Density determines whether rent, staff and inventory can be absorbed at all. A retailer whose density falls as it expands is buying growth by moving into weaker catchments or cannibalising itself. It is the cleanest early evidence that the rollout runway — which is what a 70x P/E is paying for — is exhausted. |
| Occupancy cost ratio (rent + CAM + property tax as % of sales) | Post-Ind AS 116/IFRS 16 you must **reconstruct** this from the lease note: ROU depreciation + lease interest, or the undiscounted lease commitment schedule. Also record the fixed vs revenue-share split and the escalation clause (India: typically ~5% p.a., or ~15% every 3 years). | Grocery/value retail 2–5% of sales; apparel/specialty 8–12%; mall-based department stores 12–16%. Above ~15–18% the format is structurally fragile. Revenue-share leases (Indian malls, commonly 12–18% of sales) are far safer in a downturn than fixed rent. | Rent is the retailer's operating leverage and its main route to insolvency: a 200bps rise in occupancy ratio can erase the entire EBIT of a specialty retailer. Ind AS 116 removed it from the P&L, which is precisely why the generic checklist misses it. High-fixed-rent, low-density formats break first when LFL turns. |
| GMV / GOV / NMV growth and take rate | GMV = gross value transacted. Take rate = platform net revenue (commission + fulfilment + ads + payments) ÷ GMV. Decompose take rate into commission / logistics / advertising / other — each has a different margin and a different ceiling. Pin the definition: gross or net of cancellations, returns, discounts, delivery fee and taxes. | Blended take rates: large global auction/horizontal marketplaces ~13–14%; craft/curated marketplaces ~20%+; large 3P ecosystems ~45–50% of 3P GMV all-in but ~15–20% commission-only; Indian horizontals 5–12%; deliberately low-take social commerce 4–6%; Indian food delivery ~19–22% of GOV; quick commerce ~15–18%. | GMV is the true scale variable and reported revenue is not comparable across models — take rate is the bridge between them. Take-rate expansion is the primary margin lever and a direct measure of pricing power over sellers. But a rising take rate alongside decelerating GMV or seller churn is fee extraction, not monetisation. **GMV is not revenue** and must never be substituted into a revenue ratio. |
| Contribution margin per order (₹/$ per order and % of GMV) | Net revenue per order less **all** variable costs: last-mile/rider payouts, packaging, payment gateway and COD handling, customer discounts and cashbacks, returns and RTO cost, customer support, variable warehouse cost — before corporate overhead, technology and brand marketing. Show mature versus new cities/cohorts separately. | Indian food delivery ~7–8% of GOV (₹35–45/order) at maturity. Quick commerce: mature dark stores ~4–6% of GMV, blended often 1–3% during expansion. 1P e-commerce 8–15% of GMV. Must be positive **and rising with density**, not merely with discount withdrawal. | This is the metric that says whether the business model works at all, and no generic ratio captures it. A platform can post improving consolidated EBITDA purely by throttling growth spend while unit economics stagnate. Widening contribution margin is the only evidence that scale creates genuine cost advantage. Rebuild it yourself — companies quietly exclude rider incentives, platform fees or discount spend to manufacture positivity. |
| Cohort retention, repeat share, transacting users and frequency | % of an acquisition cohort still transacting in year 2/3/4, and that cohort's GMV versus its year-1 GMV (net revenue retention). Alongside: monthly/annual transacting users (MTU/ATU), orders per user, and repeat customers' share of orders. Decompose GMV growth as users × frequency × AOV. | Repeat customers should drive >70–80% of orders at a mature platform. Year-3 cohort GMV of 60–100%+ of year-1 is strong (best-in-class cohorts spend *more* over time). Frequency: Indian food delivery ~3–4 orders/month for a core user; quick commerce ~3–5/month; fashion e-commerce ~3–5/year. Watch for the retention curve **flattening**, not just its level. | Retention is the asset. A platform with 40% year-2 retention must replace half its base annually with paid acquisition, permanently capping margins regardless of scale. Cohort curves also reveal bought growth: rising new-user counts with steepening cohort decay means progressively worse customers. This is where terminal value actually comes from, and it is invisible in P/E and ROCE. |
| CAC, CAC payback, LTV:CAC and discount intensity | CAC = performance marketing + first-order discounts/cashbacks/referral spend ÷ new customers. Payback = CAC ÷ monthly contribution per customer. Discount intensity = total customer discounts and promotions as % of GMV, **whether booked as a revenue reduction or as marketing expense**. | LTV:CAC > 3x; payback < 12 months (< 6 months excellent for high-frequency models). Sustainable platforms discount 2–5% of GMV; 10%+ signals demand that does not exist at true prices. Scaled Indian platforms should trend below 5–6% of GMV on ad + promo combined. | With minimal switching costs, acquisition economics *are* the business model. The accounting trap: platform-funded discounts can be booked as a revenue reduction (correct when paid to the customer) or as marketing expense — the latter simultaneously inflates revenue, gross margin and apparent take rate. Reconstruct discount spend from the notes and re-net it against GMV. |
| Gross margin (on a restated COGS definition) and private-label mix | State explicitly what sits in COGS — logistics, fulfilment, warehousing, shrinkage and last-mile are inside COGS for some filers and in opex for others. Alongside: own-brand sales as % of total and the margin spread over national brands. | Grocery 14–18% (deep-value formats deliberately ~14–15% to drive density); value apparel 30–40%; department stores/branded fashion 45–55%; marketplace net-revenue gross margin 60–80%. Private label: 15–25% is good for a grocer (European hard discounters >70%); own brand typically carries 500–1,500bps more margin. | Private label is the main structural margin lever in retail — it converts scale into margin without raising prices and defends against a marketplace commoditising your assortment. But cross-check rising GM against LFL: margin expansion with decelerating volume is harvesting, not compounding. Restate all peers onto one COGS definition before comparing. |
| Inventory turns, shrinkage, markdowns and ageing profile | COGS ÷ average inventory; days inventory outstanding; shrinkage (theft, damage, expiry) as % of sales; markdown as % of gross sales; and the ageing note (% over 90/180/365 days) plus the provisioning policy. | Grocery/FMCG retail 10–14x turns (~26–36 days). Value apparel 3–4x (90–120 days). Branded/department fashion 2.5–3.5x (100–150 days). Shrinkage <0.5% is best-in-class grocery, 1–2% acceptable, >2% is process failure. Markdowns should be stable as % of sales. | Inventory is where retail fraud and retail failure both live. Rising days-inventory with flat sales is the most reliable single predictor of a future margin shock, because unsold stock must eventually be marked down or written off. The ageing bucket and the year-on-year provisioning **rate** are more informative than the turns ratio; quiet policy changes here are a governance signal. |
| Cash conversion cycle, negative working capital and supplier float | Days inventory + days receivable − days payable. Decompose: is negative WC driven by genuine inventory efficiency, or purely by stretching payables? Cross-check against supply-chain finance / reverse factoring / channel financing, which converts trade payables into economic bank debt. | Grocery/quick commerce −10 to −40 days; large 1P e-commerce around −30. Some deep-value grocers run slightly *positive* by design, paying suppliers fast for cash discounts — that is a strategy, not weakness. Apparel retail +30 to +90 days. Marketplaces strongly negative. | Negative working capital is the retail float: it makes growth self-funding and is a genuine competitive asset — but it inverts the FCF signal, looking strongest during expansion and cratering when growth stalls. Model the unwind. Reverse factoring is the most common way retailers keep real debt off the debt line; check the cash-flow classification and the auditor's note. A 15–20 day jump in payable days without a stated procurement change is not efficiency. |
| Return rate, RTO and COD mix | Returns as % of gross orders/GMV (post-delivery), plus RTO — orders refused or undeliverable and shipped back, a distinctly Indian problem driven by cash-on-delivery. Track COD share of orders and the cost of the reverse leg (typically full forward logistics plus refurbishment or write-off). | Fashion e-commerce 25–40% returns in developed markets, 20–30% in India. Electronics/grocery 2–8%. RTO healthy <5% of orders; >10% is economically ruinous. Indian COD share has fallen from ~60% to ~20–30% with UPI; a **rising** COD share is a negative mix signal. | A returned fashion order can cost 1.5–2x the contribution it would have earned, so a 5pp rise in return rate flips a category from profitable to loss-making with no change in reported revenue. Returns provisioning is soft: revenue is recognised net of expected returns, so an understated return assumption directly inflates current revenue and profit. GMV reported gross of returns overstates true scale by a quarter or more in fashion. |
| Fulfilment and last-mile cost per order (and % of GMV) | Warehousing, packing, line-haul, last-mile and rider/partner payouts, per order and as % of GMV. Track against order density (orders per sq km per day) and average delivery distance/time. | Large global 1P: fulfilment + shipping ~15–17% of revenue. Indian food delivery: cost per delivery ₹55–70, of which the platform bears ₹20–30 after customer delivery fees. Quick commerce ₹35–50/order at maturity. Should fall 10–20% a year in a scaling network. | This is where the theoretical scale economies of e-commerce either materialise or do not. Delivery cost is a function of **route density, not company size** — which is why national scale does not guarantee profitability while local density does. Cost per order falling as GMV grows is empirical proof of a moat; flat cost per order means a linear-cost logistics operator wearing a technology multiple. |
| Advertising and ancillary monetisation (% of GMV, % of gross profit) | Seller/brand advertising and sponsored listings, subscription and loyalty fees, fulfilment services sold to sellers, payments and seller lending — separated from core commission. Track ad take rate and ads as % of total gross profit. | Ad take rate: largest global 3P ecosystem ~5–6% of GMV; craft marketplaces ~2%; Indian marketplaces 1.5–3.5% and rising; Indian food delivery ~1–2% of GOV. Ads carry 70–95% incremental margin. Subscription members should show 2–3x the frequency and AOV of non-members. | For every scaled marketplace, advertising rather than commission is the profit engine — at the largest platforms the ad business alone exceeds the operating profit of the entire retail segment. The path from breakeven to a 15–25% EBITDA margin is almost entirely an ad-monetisation story, so this drives the terminal margin in any DCF. Counterweight: rising ad load degrades organic relevance, so monitor ad take rate against conversion and repeat rate together. |
| New-store / dark-store payback, capex per store, mature-store EBITDA, maturity mix | Capital to open a store (fit-out + initial inventory + pre-opening) ÷ annual store-level cash EBITDA. Alongside: store-level EBITDA margin by vintage (year 1, year 2, mature), % of the estate still immature, net additions and the closure rate. | India large-format grocery: ₹15–30 crore per **owned** store (payback ~5–6 years) versus ₹3–6 crore leased; value apparel ₹1–2 crore with 2–3 year payback. Quick-commerce dark store ₹40–60 lakh, target payback 12–24 months. Mature store-level EBITDA 8–12% grocery, 18–25% apparel. Closure rate >3–4% of the estate annually signals format failure. | Return on **incremental** capital, not average ROCE, determines whether a rollout creates value: a company earning 25% ROCE overall while opening stores at 8% incremental returns is destroying value while screening beautifully. The maturity mix also explains reported margins mechanically — fast openers always show depressed consolidated margins, so mature-cohort economics are the only honest steady-state read. Refusal to disclose store-vintage economics is itself informative. |
| Lease-adjusted leverage and fixed-charge coverage (EBITDAR), plus pre-Ind AS 116 EBITDA | (Net debt + lease liabilities) / post-IFRS 16 EBITDA, **or** the older convention (net debt + 8× annual rent) / EBITDAR. Fixed-charge cover = EBITDAR / (interest + rent). Also restate EBITDA pre-Ind AS 116 (rent back in opex) and compute FCF after lease payments. | Lease-adjusted net debt/EBITDAR < 3.0x comfortable, 3.0–4.0x manageable, >4.5x fragile. Fixed-charge cover > 2.0x healthy, < 1.5x distress. Rating agencies, Indian ones included, use the 8×-rent capitalisation convention. | Retail insolvencies are almost always rent insolvencies, and generic D/E either misses rent entirely (pre-2019 accounting, or US GAAP operating-lease presentation) or counts it as debt without adding rent back to EBITDA. Pre-Ind AS 116 EBITDA is also the only way to compare an Indian retailer against its own pre-FY20 history. |
| ESOP charge as % of revenue and diluted-share bridge (India new-age; global tech-retail) | Share-based payment expense from the P&L note as % of revenue, plus outstanding options/RSUs as % of share capital and the 3-year dilution trend. | Indian new-age platforms commonly 3–8% of revenue at listing, should fall toward 1–2% within 3–4 years. Outstanding pool often 5–10% of share capital. | Companies headline "Adjusted EBITDA" excluding it. It is a recurring, cash-equivalent, shareholder-borne cost and a direct transfer of value. Always add it back into the earnings base before applying any multiple, and model the dilution separately in per-share terms. |
Sourcing notes: in India, LFL, sales density, store count by format, GMV/GOV, take rate, contribution margin and dark-store cohorts appear in the **quarterly investor presentation and concall transcript**, not in the Ind AS financials — read the transcript and the shareholder letter. Lease and inventory-ageing detail lives in the notes to accounts and the CARO annexure. In the US, segment revenue, fulfilment cost lines and lease maturity schedules are in the 10-K; store counts, square footage and comparable-sales definitions are in the MD&A and the 8-K earnings release; the comparable-sales *definition* is footnoted and changes are disclosed there.
---
## How to value companies in this sector
No single multiple works across this sector. Pick the method from the sub-model, and state the basis explicitly every time.
### A) Mature physical retail (grocery, value apparel, department stores, jewellery, electronics)
**Primary: EV/EBITDA, with absolute basis consistency.** Either (i) post-IFRS 16 — EV includes lease liabilities in net debt and EBITDA is post-IFRS 16 (typically 8–14x in developed markets, 25–45x for premium Indian compounders); or (ii) lease-adjusted — EV = market cap + net debt + 8× annual rent, over EBITDAR. **Never mix the two.** Secondary: EV/Sales as a sanity check (0.3–1.0x developed grocery, 2–5x high-growth Indian retail) and P/E for genuinely mature, low-growth names. Indian sell-side convention is to present "EV/EBITDA pre-Ind AS 116" alongside reported specifically to preserve comparability with pre-FY20 history — replicate that.
**DCF is the honest workhorse for Indian retail**, because headline P/Es of 60–130x are pricing a decade-long rollout runway that no static multiple expresses. Build it bottom-up: store additions × sales density × mature store EBITDA margin × incremental capex per store, with an explicit **terminal store count**. Then invert: at today's price, what store count and density are implied, and are they physically plausible given catchment arithmetic and competitor estates?
**Property adjustment (opco/propco).** Where the retailer owns its real estate — some Indian large-format grocers own the large majority of their stores, as do several US big-box chains — split it: value the operating business on EV/EBITDAR with a **market rent charged**, then add the property at a cap rate. Without this, owned-store retailers look permanently and misleadingly capital-inefficient against leased peers.
### B) Marketplaces and platforms (3P e-commerce, food delivery, horizontals, social commerce)
Because reported revenue is a net number and near-term EBITDA is small or negative, climb the ladder: **EV/GMV (or EV/GOV) → EV/Net Revenue → EV/Gross Profit → EV/Contribution Profit → EV/EBITDA once real.** EV/GMV benchmarks: 0.3–0.8x for low-take-rate horizontals, 1.5–3.0x for high-take-rate, high-frequency platforms — but EV/GMV is only interpretable **paired with the take rate**, since EV/GMV = EV/Revenue × take rate. **EV/Gross Profit or EV/Contribution Profit is the most defensible near-term multiple**, because it normalises away the 1P-versus-3P revenue-recognition distortion entirely; this is the single most useful comparability fix in the sector.
The primary method remains a **steady-state-margin DCF** structured as: GMV growth → take rate → contribution margin → mature EBITDA margin → reinvestment. Every metric in the table above feeds one link of that chain. Terminal margins you must be able to defend: 15–25% of net revenue for a scaled marketplace, 4–8% of GMV for food delivery, 2–5% of GMV for quick commerce.
**Cohort/LTV valuation** is the rigorous alternative for high-frequency models: value = NPV of existing cohorts (observed retention × contribution per cohort) + option value on future cohorts. It is the only method that makes an unprofitable, fast-growing platform valuable for a defensible reason rather than a narrative one.
### C) Indian new-age platforms — SOTP is the market convention
Listed Indian internet/retail platforms are almost universally valued **sum-of-the-parts**, because segments have wildly different economics and maturity: a food-delivery segment on DCF or EV/EBITDA, a quick-commerce segment on EV/GOV, a B2B supplies arm on EV/Sales, a going-out/ticketing arm on EV/Revenue — less a holding-company discount, **plus net cash added separately**. Post-IPO cash piles are large; never let them dilute the operating multiple, and never let treasury "other income" flatter consolidated PAT. Beauty/fashion platforms split naturally into a high-margin near-profitable vertical (EV/EBITDA) and a loss-making one (EV/GMV). Do the segment split yourself even where the company reports consolidated.
**Critical Indian adjustment:** restate "Adjusted EBITDA" to **include ESOP cost** before applying any multiple, and model option-pool dilution separately in the per-share bridge.
### D) 1P inventory-led e-commerce
EV/Sales and EV/Gross Profit near term; EV/EBITDA at maturity. These are economically retailers with a distribution overlay — benchmark their gross margin, inventory turns and contribution margin against **physical retail** peers, not against asset-light marketplaces.
### What not to use
- **P/B** — brand, cohorts and network are unbooked and accumulated losses distort equity. Exception: property-heavy retailers, where book at least anchors to land.
- **Trailing P/E through the loss-to-profit crossover** — mathematically unstable and economically uninformative.
- **EV/Sales compared across 1P and 3P** — the denominators are different quantities.
- **Any multiple applied to GMV as if GMV were revenue.**
- **Post-IFRS 16 EV against pre-IFRS 16 EBITDA**, in either direction.
Always close by inverting the valuation: what mature EBITDA margin, terminal store count, or terminal GMV share does today's price require — and has any comparable company anywhere ever achieved it?
---
## Peer set construction
The defining rule: **peers must share a revenue-recognition model, a lease posture and a maturity stage.** Violating any one of the three makes the comparison arithmetic, not analysis.
Split on these axes, and never blend across them:
- **Principal (1P) vs agent (3P)** — the hard boundary. A 1P grocer and a 3P marketplace cannot be compared on margin, EV/Sales or revenue growth. If you must span them, convert both to EV/Gross Profit or EV/GMV.
- **Format and basket** — grocery/food (high frequency, low ticket, low GM), value apparel (low ticket, high GM, high turns), department stores and premium fashion (high ticket, high GM, low turns), jewellery (very high ticket, gold-price-linked, hallmarking and making-charge economics), electronics (low GM, financing-driven). Sales density, occupancy ratio and inventory turns differ by multiples across these.
- **Owned vs leased property** — until you have run the opco/propco adjustment and charged market rent, an owner and a lessee are not comparable on ROCE, EBITDA margin or D/E.
- **Franchise/FOFO vs company-owned (COCO)** — franchised chains book wholesale revenue at lower margin on almost no capital; company-owned chains book retail revenue at higher margin on real capital. ROCE and OPM are non-comparable; compare **system-wide sales** and store-level economics instead.
- **Delivery model** — food delivery, quick commerce (10–30 min, dark stores, owned inventory), scheduled e-grocery, and horizontal e-commerce have entirely different cost curves, frequencies and AOVs. Quick commerce is *inventory-led* even when the parent is a marketplace; treat it as 1P.
- **Maturity stage** — a chain with 40% of its estate under two years old will always show worse consolidated margins than an identical mature chain. Compare mature-cohort economics, or explicitly adjust for the maturity mix.
- **Geography and density** — delivery economics are set by population density and labour cost. Indian quick commerce, US same-day and European e-grocery are not the same industry. Also split by urban tier: metro-only formats and tier-2/3 formats have different rent, wage and basket structures.
Practical construction rules: use **TTM** figures, never annualised quarters, because of festive/holiday concentration. Restate every peer onto a single lease basis and a single COGS definition before tabling anything. Where an Indian conglomerate houses retail inside a larger listed entity, use the **segment disclosure** and compare segment-level EBIT and capital employed, never the consolidated entity. For diversified platforms, build peer sets segment-by-segment, matching each segment to its own comparables.
---
## Sector-specific red flags
Ordered roughly by how often they precede a de-rating.
1. **GMV/NMV/GOV definition changes or silent redefinition.** Switching between GMV, NMV and GOV, or including cancelled, returned and RTO orders in headline GMV. In fashion this can overstate true transacted value by 25–40%. Any mid-series definition change, or a quiet footnote change to inclusions, is a deliberate flattering of growth until proven otherwise.
2. **Principal-vs-agent (gross-vs-net) revenue-recognition flips.** Optical revenue growth or collapse with zero economic change. The giveaway is revenue growth diverging sharply from gross profit growth or from GMV growth.
3. **Customer discounts booked as marketing expense rather than as a revenue reduction.** This inflates revenue, gross margin and apparent take rate simultaneously. Reconstruct discount spend from the notes and re-net it against GMV; discounts above ~5–8% of GMV suggest demand that does not exist at real prices.
4. **"Adjusted EBITDA" excluding ESOP, and ESOP rising as a share of revenue.** Endemic among Indian new-age platforms at 3–8% of revenue. Watch equally for "one-off" restructuring, store-closure or inventory-write-down adjustments that recur every single year.
5. **IFRS 16 / Ind AS 116 EBITDA flattery.** Headline EBITDA growth driven by the lease standard rather than trading; or a company presenting lease-inflated EBITDA against a pre-IFRS 16 enterprise value; or comparing post-FY20 D/E and ROCE with the pre-FY20 series as though they were the same metric.
6. **Strong total revenue growth while LFL is flat or negative** — growth entirely from new openings. Paired with falling sales per sq ft as the estate expands, this means new stores are entering weaker catchments or cannibalising existing ones, and the runway justifying the multiple is already spent.
7. **Stretching supplier payment terms to manufacture FCF, and undisclosed reverse factoring / supply-chain finance / channel financing** that keeps economically debt-like obligations inside trade payables. Check the cash-flow-statement classification, the payables ageing and any auditor comment. A 15–25 day jump in payable days without a stated procurement change is not efficiency.
8. **Inventory building ahead of an inevitable markdown.** Days-inventory rising while sales are flat; deterioration in the >90/>180 day ageing buckets; quiet loosening of the provisioning policy; or recognising shrinkage only at the annual physical count so the first three quarters look clean.
9. **Channel stuffing into franchisees, dealers or distributors** — common in Indian franchise-led apparel and footwear. Primary sales (to the franchisee) are booked as revenue while secondary sales (to the consumer) stagnate. Demand secondary-sales and franchisee-inventory disclosure; a widening primary–secondary gap plus rising related-party receivable days is the signature.
10. **Contribution margin defined to exclude real variable costs** — rider incentives, platform fees paid out, packaging, gateway charges, reverse-logistics on returns. Companies reach "contribution positive" by moving a cost line into corporate overhead. Rebuild the number from the segment notes yourself.
11. **Take-rate expansion from unilateral fee hikes rather than value delivered** — visible as rising take rate with decelerating GMV, falling active-seller counts, or seller agitation and regulatory noise. In India, add the overhang on marketplace fees, deep-discounting under FDI rules and platform-neutrality scrutiny, any of which can reset take rates abruptly.
12. **Disclosure withdrawal or metric-switching in cohort and user data** — dropping monthly transacting users for annual, redefining a cohort, moving from "orders" to "transactions", or ceasing to publish retention curves. Companies stop disclosing a metric precisely when it turns. Treat removal of a previously reported KPI as a negative disclosure event in its own right.
13. **Marketing cut sharply to hit an EBITDA-breakeven promise while GMV growth decelerates in the same quarters** — harvesting rather than building. Genuine operating leverage shows margin improvement *with* sustained growth; artificial improvement shows margin up and growth down together.
14. **Rising returns and RTO with inadequate provisioning, and a rising COD mix.** Revenue is booked net of expected returns, so an understated return assumption inflates current profit and creates a future reversal. High COD plus rising RTO can make an entire category structurally loss-making while GMV still looks healthy.
15. **Lease sweeteners flattering early store economics** — sale-and-leaseback gains, rent-free periods, landlord fit-out contributions, step-up leases. Reported occupancy cost is artificially low in years 1–3 and steps up later. Also check whether gains on property disposals or store closures sit inside operating profit rather than below the line.
16. **Quick-commerce / dark-store specific:** reporting dark-store **count** as the growth metric while blending immature and mature stores to hide per-store economics; or claiming store-level profitability on a definition excluding central warehousing, line-haul and rider fixed costs. Demand mature-cohort economics and orders-per-store-per-day by vintage.
17. **Aggressive capitalisation** — of customer acquisition or content costs, software and technology development, pre-opening store expenses, or last-mile assets. Cross-check the gap between reported EBITDA and (operating cash flow − capex); a persistently widening gap at a company reporting improving "adjusted" profitability is the classic signature.
18. **Treasury income presented as core earnings.** A large post-IPO cash pile generating other income that turns consolidated PAT positive while the operating business still burns cash. Strip other income out and read operating EBIT; a company earning more from treasury than from operations is not yet a business.
19. **Gift-card and loyalty-point breakage, and deferred-revenue release from subscriptions**, used to smooth or manufacture quarterly profit — a low-visibility lever with material year-end effects.
20. **Governance and concentration warnings** — CFO or auditor churn (especially at Indian new-age names), heavy related-party transactions with promoter-owned suppliers, logistics arms or franchisees, GMV concentrated in a few sellers or a single category, and continuous large promoter or pre-IPO-investor selling into strength.
---
## Cycle and structural context
**Where you are in the cycle changes which metric leads.** Retail is a levered play on real disposable income. In a consumption upcycle, LFL runs ahead of inflation, footfall grows, markdowns fall and operating leverage flatters margins; in a downturn, ticket holds up (price/mix) while footfall falls first — which is why the footfall-vs-ticket decomposition is the earliest cycle read available. Watch **trade-down**: value formats and private label gain share in downturns while premium and department stores lose it, so the same macro is a tailwind for one peer group and a headwind for another. Never treat a sector-wide LFL number as a company signal without this split.
**Inflation cuts both ways.** Moderate food inflation *helps* grocers' reported LFL and absorbs fixed costs; sharp inflation compresses basket volumes and shifts mix to essentials. Deflation is the harder regime — LFL turns negative on unchanged volumes and operating leverage runs in reverse. Always separate price-led from volume-led LFL before calling a trend.
**Rent and wage cycles set the floor on fragility.** Occupancy ratios are agreed years in advance with 5% escalations, but sales are not. A retailer that signed leases at peak rents and then sees flat LFL will breach fixed-charge cover before it breaches any P&L covenant. In India, minimum-wage revisions and rising gig-worker payouts feed directly into quick-commerce and delivery unit economics.
**Secular threats to physical retail:** marketplace commoditisation of assortment (the reason private label matters), quick-commerce cannibalisation of top-up grocery baskets and of general trade, mall traffic decline in weaker catchments, and format obsolescence (large-box department stores globally). Test any physical retailer for what proportion of its basket is genuinely defensible against 10-minute delivery.
**Secular threats to platforms:** customer-acquisition cost inflation as ad auctions mature; regulatory resets to take rates and to gig-worker classification; the shift of discovery to social and AI-mediated interfaces, which can disintermediate a marketplace's own search; and brands going direct (D2C) to reclaim margin. The counterweight is advertising monetisation and logistics density, which is why those two metrics dominate the terminal-value case.
**Regulation — India specifically.** FDI policy distinguishes single-brand from multi-brand retail and restricts e-commerce marketplaces from owning inventory or influencing prices, which is why the marketplace/1P structuring is often legally driven rather than economically driven. Press Note 2 style restrictions, deep-discounting scrutiny, platform-neutrality and seller-exclusivity issues, CCI investigations, ONDC as a policy attempt to unbundle discovery from fulfilment, and gig-worker social-security proposals are all live take-rate and cost risks. GST rate changes shift ticket sizes; the e-commerce TCS/TDS regime affects seller cash flows. **Globally:** antitrust and self-preferencing actions against large platforms, DSA/DMA-type obligations in the EU, state-level gig-worker classification rules in the US, and de minimis import thresholds that determine whether cross-border ultra-low-price players are viable.
**Seasonality is structural, not noise.** Indian apparel, gifting, jewellery and electronics concentrate 30–40% of annual profit in the festive quarter (Q3, Oct–Dec, around Navratri/Durga Puja/Diwali, with a wedding-season tail); Western retail concentrates in Q4 holiday. Never annualise a quarter, always compare year-on-year on the same quarter, and check whether a festive-date shift moved sales between quarters before reading a "collapse" or a "surge".
---
## India vs global notes
| Dimension | India (NSE/BSE, Ind AS) | US / global (10-K, US GAAP / IFRS) |
|---|---|---|
| Units and reporting | ₹ crore/lakh; quarterly results plus a detailed investor deck and concall; annual report with notes, CARO annexure and related-party schedules | $ millions; 10-Q/10-K on EDGAR; 8-K earnings release; store counts and comparable-sales definitions in MD&A |
| Lease accounting | Ind AS 116 from FY20 — all leases on balance sheet; rent vanishes from opex | IFRS 16 from 2019 (all leases on balance sheet); **US GAAP ASC 842 keeps operating-lease expense as a single opex line in the income statement** while still recognising the ROU asset and liability — so US EBITDA margins were *not* restated the way IFRS/Ind AS ones were. This is a major cross-border comparability trap |
| Where KPIs live | LFL, sales density, store count, GMV, take rate and contribution margin are in the investor presentation and concall transcript, not the financials | Comparable sales, square footage, segment revenue and fulfilment costs are in the 10-K/8-K itself; GMV and take rate in shareholder letters and supplemental decks |
| Ownership | Promoter holding, pledge disclosure, promoter-affiliated suppliers and logistics arms; SEBI related-party rules; look at promoter and pre-IPO investor selling | Dispersed ownership; dual-class structures common at founder-led platforms; insider sales via Form 4 and 10b5-1 plans |
| Audit and controls | CARO 2020 reporting on inventory verification, related-party loans and undisclosed income; auditor resignations are a strong governance flag | SOX 404 internal-control opinion; material weakness disclosures; auditor changes on Form 8-K Item 4.01 |
| Non-GAAP conventions | "Adjusted EBITDA" almost always excludes ESOP; pre-Ind AS 116 EBITDA commonly presented alongside reported; "Adjusted EBITDA as % of GOV" is the food-delivery/quick-commerce convention | Non-GAAP reconciliations mandated by Reg G; SBC exclusion is also common and equally objectionable; "free cash flow" definitions vary widely on lease and finance-lease treatment |
| Payments and returns | COD and RTO are material, distinctly Indian cost lines; UPI has collapsed COD share; card/EMI penetration lower | Card-dominant; RTO negligible; returns cost concentrated in apparel; "returnless refunds" now common at scale |
| Regulation | FDI rules on multi-brand retail and marketplace inventory ownership; deep-discounting and platform-neutrality scrutiny; CCI; ONDC; gig-worker social-security proposals; GST and e-commerce TCS | Antitrust/self-preferencing actions; EU DMA/DSA; state gig-worker classification; de minimis import thresholds; FTC rules on subscriptions and pricing disclosure |
| Valuation convention | High absolute P/E and EV/EBITDA for quality retail; SOTP with net cash added separately for new-age platforms; sell-side presents pre- and post-Ind AS 116 multiples in parallel | EV/EBITDA and EV/Sales dominant; SOTP used for conglomerate platforms (retail vs cloud vs ads); property adjustments common for owned-real-estate big-box |
| Real estate | Mall revenue-share leases common (12–18% of sales); some large grocers own the majority of their stores, making opco/propco essential | Predominantly leased in specialty; owned real estate significant in big-box and warehouse clubs; sale-leaseback activity is a recurring earnings-quality issue |
---
## Checklist
- [ ] Classify the business: 1P principal, 3P agent, hybrid, franchise, or quick-commerce inventory-led — and state it before quoting any margin.
- [ ] Normalise everything to GMV/GOV or gross profit before comparing across models; never treat GMV as revenue.
- [ ] Pin the GMV definition (returns, cancellations, RTO, delivery fee, taxes) and check it has not changed mid-series.
- [ ] State your lease basis (post-IFRS 16 or lease-adjusted 8× rent) and apply it identically to EV, EBITDA, D/E and ROCE for every peer.
- [ ] Restate EBITDA pre-Ind AS 116 to compare an Indian retailer against its own pre-FY20 history.
- [ ] Compute FCF **after** lease payments; model the negative-working-capital unwind at lower growth.
- [ ] Get LFL/SSSG on a TTM basis and decompose it into footfall × conversion × ticket × items.
- [ ] Compare mature-store versus new-store sales density; if density falls as the estate grows, the runway is closing.
- [ ] Rebuild the occupancy cost ratio from the lease note; flag anything above ~15% of sales for a specialty format.
- [ ] Rebuild contribution margin per order yourself from the notes, including rider incentives, discounts, gateway and reverse-logistics costs.
- [ ] Check take rate alongside GMV growth and active-seller counts — rising take rate with falling GMV is extraction.
- [ ] Reconstruct discount spend as % of GMV regardless of where it is booked, and re-net it against revenue and take rate.
- [ ] Read the cohort/retention disclosure; treat any withdrawn or redefined KPI as a negative event.
- [ ] Test whether fulfilment cost per order is falling with volume — flat cost per order means no density moat.
- [ ] Size advertising and ancillary monetisation as % of gross profit; it usually is the terminal margin story.
- [ ] Get new-store/dark-store payback, capex per store and mature-cohort store EBITDA; compute return on **incremental** capital.
- [ ] Read the inventory ageing note and provisioning policy; rising days-inventory on flat sales predicts a markdown.
- [ ] Check payable days, reverse factoring / channel financing and the cash-flow classification before crediting negative working capital.
- [ ] Add ESOP charges back into EBITDA and model option-pool dilution per share.
- [ ] Strip treasury/other income out of consolidated profit and read operating EBIT alone.
- [ ] For franchise-led chains, demand secondary sales and franchisee inventory; watch the primary–secondary gap.
- [ ] Run lease-adjusted net debt/EBITDAR and fixed-charge cover; retail failures are rent failures.
- [ ] For owned-property retailers, run an opco/propco split with market rent charged before judging ROCE.
- [ ] Use TTM, never annualised quarters — festive/holiday concentration makes single-quarter annualisation misleading.
- [ ] For Indian new-age platforms, value SOTP by segment and add net cash separately.
- [ ] Invert the valuation: what terminal margin, store count or GMV share does the price require, and has anyone ever achieved it?
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| add-product-e2e-evalpaperclipai/paperclipAdd or extend a Paperclip full-stack runner E2E workflow, fixture, matcher, or report evidence path for local or Daytona execution. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| add-runner-evalpaperclipai/paperclipAdd or extend a Paperclip Runner protocol evaluation definition, roster, assertion, or report fixture with provenance and narrow validation. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| diagnose-why-work-stoppedpaperclipai/paperclipCompatibility entry point for failed, stalled, looping, or over-recovered Paperclip work. Use fix-failed-paperclip-run for the full diagnosis, regression, repair, real verification, and green-PR workflow; honor an explicitly requested diagnosis-only scope. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| fix-failed-paperclip-runpaperclipai/paperclipDiagnose and fix failed, stuck, slow, or incorrectly blocked Paperclip tasks and runs. Use with a task/run URL or failure report to trace the root cause, prove a regression red then green, check second-order effects, verify the real workflow, and deliver a PR with passing checks. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |