Time to put everything together in the first of two flagship builds. "SwiftMart" is a synthetic value grocer modelled on the shape of Avenue Supermarts: owned stores, EDLP, cash sales. The figures are illustrative, so cross-check every line against the actual DMart annual report and earnings-call commentary.
Store-level unit economics example: a DMart-style store
One mature store, per year (₹ cr):
| Line | ₹ cr | Notes |
|---|---|---|
| Revenue | 120.0 | ~₹33 lakh/day through one store |
| − COGS (gross margin 14.5%) | 102.6 | thin by policy: the EDLP engine (see the flywheel effect) |
| Gross profit | 17.4 | |
| − Store staff, electricity, consumables, maintenance | 6.0 | costs that exist because this store exists |
| Store EBITDA | 11.4 | 9.5% of revenue |
| − Depreciation (building + fit-out) | 2.0 | owned property, so capex instead of rent |
| Store EBIT | 9.4 | 7.8% |
Next comes the part per-unit decks leave out, the store's balance sheet. Land + building + fit-out ≈ ₹42 cr, plus inventory net of payables ≈ ₹6 cr, gives capital per store ≈ ₹48 cr. Store-level pre-tax ROCE = 9.4 / 48 = 19.6%. After tax (25%) it's ≈ 14.7%, comfortably above the ~12% INR hurdle from the cost-of-capital build, and that's before corporate costs. Two numbers carry the thesis. Each new box compounds capital in the high teens, and the reinvestment runway is decades long, with India's grocery market set against ~400 stores. The DCF identity g = reinvestment rate × ROIC stops being abstract when the "project" is a building you can walk into.
Advertisement
How does store maturity change the economics?
Maturation. A store doesn't open mature. A typical ramp runs year 1 ≈ ₹55 cr revenue, year 2 ≈ ₹80, year 3 ≈ ₹100, year 4 ≈ ₹115, mature ≈ ₹120+. The cost base is mostly in place from day 1, so store EBITDA margin climbs from ~3–4% in Y1 to 9.5% at maturity. That has two consequences, and you should compute both, never guess:
- Simple payback. Set cumulative store post-tax cash against the ₹48 cr. On this ramp it takes roughly 7–8 years, which is slow. Now compare the leased alternative. Rent is ₹4.5 cr/yr in place of ₹42 cr of property, and capital falls to ~₹12 cr (fit-out + working capital). Store EBITDA, now paying real rent, = 11.4 − 4.5 = 6.9. EBIT ≈ 6.1, so pre-tax ROCE = 6.1/12 ≈ 51%, with payback in ~3 years. So why own? Rent escalates ~5%/yr forever, landlords re-price success, and EDLP dies the day costs stop being the market's lowest for good. Ownership deliberately trades reported ROCE and speed for cost immunity and permanence. That's a moat decision, and you'll name it properly when you get to moats. You can only see it at the unit layer.
- The maturity-mix effect. Take a chain of 100 mature stores (₹120 cr @ 9.5%) plus 40 young ones (avg ₹70 cr @ 4.0%). Revenue = 100×120 + 40×70 = ₹14,800 cr. Store EBITDA = 1,140 + 112 = ₹1,252 cr, a blended 8.46% against 9.5% for a mature store. Fast growth depresses reported margins by ~1 point purely through mix, and nothing has deteriorated anywhere. The flip side follows. When a retailer stops opening stores, margins "expand" with zero operational improvement. When it speeds up, margins "compress" while the machine is fine. If you can't decompose maturity mix, you'll mistake the throttle for the engine.
How do you check the store model against reported results?
The reconciliation test (first pass). Units × unit P&L has to land on the reported company. 140 stores' EBITDA of ₹1,252 cr − corporate/HQ costs (say ₹210 cr) ≈ ₹1,042 cr company EBITDA (7.0% of revenue). If the company reports 7.0%, the unit story is real. If it reports 4.5% and the bridge is a vague "other", the "store economics" slide is curated fiction. No check you have is more powerful than this one, unit × count + corporate ≈ reported, and the games section at the end turns it into a standing routine.