Ecommerce KPIs

4.4.1 The model, and why "revenue" needs a decoder

An e-commerce company aggregates demand on a screen and fulfils it from a warehouse network. Before any analysis, settle the revenue-recognition question from M1: is it the merchant (1P) or the agent (3P)? A 1P (inventory-led) seller books the full basket as revenue. A 3P marketplace books only its take, meaning commission + fees + ads. On the same ₹1,000 order, 1P revenue is ₹1,000 and 3P revenue is ₹250. GMV, gross merchandise value, is the value of goods transacted on the platform, ideally net of returns and cancellations (but definitions vary, so read the footnote). It's the platform-scale metric that spans both models, and it is not revenue. Value a mixed 1P/3P business on EV/GMV without take-rate context, or on EV/"revenue" without checking the mix, and you'll pay 3× too much or short something that's 3× too cheap. So use EV/Sales on net revenue, with the 1P/3P mix stated.

4.4.2 The ladder: GMV → net revenue → CM1 → CM2 → CM3 → EBITDA (fully worked)

Read the e-commerce P&L as a ladder where each rung asks one question. Definitions vary from company to company. Nykaa, Zomato, Delhivery and their peers each cut it slightly differently, so always rebuild from their definitions. Here's the standard we'll use:

• CM1 = net revenue − cost of goods (product margin). Is there a real trading margin?
• CM2 = CM1 − fulfilment (warehousing, packaging, last-mile, payment gateway, returns processing). Does the margin survive the truck?
• CM3 = CM2 − marketing. Does it survive the customer-acquisition machine?
• EBITDA = CM3 − fixed costs (tech, corporate). Does scale exist to pay for the head office?

Worked example: GlowCart's contribution-margin ladder

Worked example: GlowCart (synthetic beauty/fashion e-tailer, ₹ cr). Gross GMV is 6,000. Returns + RTO take 10%, leaving net GMV of 5,400, and at an AOV of ₹900 that's 6.0 cr (60 million) orders. On mix, 1P is 60% of net GMV, so ₹3,240 is booked as product revenue. 3P is the other 40%, and ₹2,160 of GMV × a 25% take rate = ₹540 of commissions/fees/ads. Net revenue = 3,240 + 540 + 120 of 1P-brand ad income = ₹3,900 cr. That's 72% of net GMV, high because of the 1P tilt.

State your base every time: % of net revenue, % of net GMV, or per order. Sell-side decks switch bases mid-slide and hope you won't notice. Here, 14.6% of net revenue = 10.6% of net GMV, and both describe the same ₹570 cr.

Where does a loss-making e-commerce company lose money?

Reading the ladder. GlowCart loses money, so ask where it loses money. CM2 is +27%, so the trading machine works after the truck is paid. CM3 is +7%, so even the marketing engine covers itself. The whole loss comes from fixed costs, and those are, by definition, the rung that scale fixes. Rerun it at 1.5× (net GMV 8,100, net revenue 5,850). Hold CM2% flat, let marketing fall to 16% of net revenue as the repeat mix rises, and let fixed costs grow only 25%. Then CM2 = 1,591, CM3 = 1,591 − 936 = 655, and EBITDA = 655 − 588 = +₹67 cr (+1.1%). That's a credible path to profitability.

Now picture the mirror image, with CM2 negative because fulfilment eats the product margin. No amount of scale fixes negative unit economics. More volume just digs the hole faster. Losses by choice (positive CM, heavy fixed/marketing investment) vs losses by physics (negative CM) is the most important distinction in loss-making e-commerce.

4.4.3 Repeat, cohorts, CAC: the demand side (bridge to M4.02/M5.04)

Tight fulfilment decides whether an order is profitable. Repeat behaviour decides whether the customer is. Here's the panel:

• Repeat rate / existing-customer share is the fraction of orders or GMV that comes from customers acquired in earlier periods. Mature, healthy platforms run 60 to 75%+. Beauty skews high, and Nykaa-style businesses have reported ~70%+ of GMV from existing customers (illustrative, check the investor deck). High repeat means the marketing line is buying growth instead of survival.
• Cohort repeat means taking everyone acquired in a quarter and tracking their spend over time. A healthy cohort retains, with 40 to 60% buying again in year 2 depending on category. It also deepens, as spend per retained customer rises. So cohort GMV holds at 70 to 100%+ of year-1 levels. Stack healthy cohorts and you get revenue growth even with zero new customers. Decaying cohorts are the treadmill, the e-commerce twin of retail's SSSG-zero flag.
• CAC and its trend. CAC = acquisition marketing spend / new customers acquired. For GlowCart, ₹585 cr of its marketing is acquisition and it brings in 1.3 cr new customers, so CAC is ₹450. Now payback. New customers order ~2.4× in year 1 at a CM2/order of ₹177. That's a year-1 contribution of ₹425, so payback takes ~13 months, just into year 2. It's acceptable if retention holds, and the LTV machinery from unit economics takes over from there. The red-flag combination is CAC rising while repeat stays flat. You're paying more for customers who behave no better. That's the signature of a saturating niche or of intensifying competition, with someone else's discounts bidding up your keywords.

4.4.4 Fulfillment cost % and the physical scale curve

Fulfilment is e-commerce's rent. GlowCart's ₹95/order ≈ 14.6% of net revenue. Amazon's fulfilment expense runs ~15 to 16% of net sales before outbound shipping (illustrative, ~FY24 scale, and the 10-K splits fulfilment from shipping). The cost falls with density: fuller trucks, closer warehouses, higher AOV. It rises with service ambition: faster promises, COD, high-return categories. Watch per-order fulfilment cost and fulfilment % of net revenue over time. If they're flat to falling while delivery speeds improve, a logistics moat is compounding. If they're rising, the growth is coming from geographies or categories worse than the average. In India, COD drives RTO (return-to-origin), where the parcel travels twice and sells nothing. Fashion returns run 25 to 30%+ including RTO, and beauty runs single digits. Category mix decides fulfilment economics.

4.4.5 Path-to-profitability, valuation, and anchors

Read any loss-maker in this order:

1. Is CM1 positive and stable?
2. Is CM2 positive, with per-order contribution trending up?
3. Is CM3 positive, with marketing % falling because repeat mix rises, and not because growth was cut?
4. Is fixed-cost growth ≪ revenue growth?

What does an Amazon-shaped path to profitability look like?

A company that passes 1 to 3 while investing heavily behind 4 is Amazon-shaped. Amazon ran thin-to-negative retail margins for two decades by choice. It had positive contribution economics and reinvested them relentlessly. The money went first into fulfilment density, then into the two profit pools that now fund everything: AWS at ~35 to 38% operating margins and ~$56bn of advertising (illustrative, ~FY24 scale). A negative CCC of roughly −20 to −30 days helped pay for it. Customers pay today, and suppliers get paid next month. Read the 1997 shareholder letter, which has been reattached to every annual report since. It's the unit-economics manifesto. It puts cash flow per share ahead of reported earnings, and market leadership metrics like repeat purchase and GMV per customer ahead of accounting optics. A company failing rung 2 with rising CAC is doing Amazon cosplay, and its losses are physics dressed up as strategy.

How do you value a loss-making e-commerce company?

Valuation. Before profits, use EV/Sales on net revenue. An EV/Sales of 1 to 5× is ordinary, and more is justified only with SaaS-like repeat economics and visible CM2 expansion. Sanity-check it with EV/net-GMV given the take rate. As EBITDA arrives, migrate to EV/EBITDA and eventually P/E. Every EV/Sales multiple is an implicit claim about terminal margin and growth, so reverse it with a reverse DCF. 4× sales at a 10% eventual FCF margin is 40× eventual FCF, before you wait years for "eventual." In India the anchor is Nykaa-style, with GMV ~₹15,000 cr and net revenue ~₹8,000 cr (1P-heavy beauty + 3P fashion). EBITDA is ~5 to 6% and climbing, with beauty profitable while fashion invests (illustrative, modelled on ~FY25 scale). Pull the annual report and the quarterly investor deck, which disclose the CM structure. In the US the anchor is Amazon, as above.

4.4.6 E-commerce red flags