How to Analyse an EPC Infrastructure Company

In April 2025, KEC International — a 75-year-old engineering-and-construction house — told analysts it would lift its operating margin from "the current 7%" to "8% to 8.5%" through FY26 (KEC Q4 FY25 concall). It was a reasonable plan from a competent management. By the October 2025 call the guidance had quietly slid to "7% to 7.5%" (KEC Q2 FY26 concall), and the nine-month margin printed at 7.1% (KEC Q2 FY26 concall). The order book over the same stretch was never the problem — it grew from ₹34,409 crore to ₹36,725 crore (Inve data, Q1→Q2 FY26). What broke the margin was not demand. It was the working capital that sat between winning the job and getting paid: net working capital stuck at 122–128 days against a 100-day target (Inve data, Q4 FY25 / Q1 FY26), and interest costs eating roughly 2.9–3.2% of revenue off a margin that started near 7% (Inve data, Q3 FY25 / Q2 FY26). (Illustration of how to read the numbers, not a view on the stock.)

That gap is the whole subject of this guide. An EPC (engineering, procurement and construction) company looks, from the outside, like a demand story — a giant order book, a roaring capex cycle, headlines about ₹3 lakh crore of inflows. But the backlog is the easy part. The thing that decides whether that backlog turns into shareholder value, or into a slow bleed of borrowed money, is what happens to cash between the order and the payment. Across the management commitments Inve tracks, EPC names are well represented among those that guide confidently on order inflow and margin, then quietly revise the working-capital and debt targets that actually govern returns.

This is how to read an EPC contractor the way a credit-aware equity analyst does: the operating numbers that decide the outcome, where the important ones are buried, the concall questions that separate a real backlog from a cash trap, and the one red flag that has cost infrastructure investors more than any missed order.

A note on the boundary first. You will not model a 200-project book from the outside. What you can do is read the direction of the order book, the cash conversion and the debt, and check whether management's account survives the Q&A.

What actually drives the economics of an EPC company?

Strip an EPC contractor down and it is a working-capital business wearing an order-book costume. It bids fixed-price contracts to build something — a transmission line, a metro, a water-supply network — buys steel and cement and labour up front, executes over two to four years, and gets paid in stages against certified milestones, with a slice (retention money) held back until the job is signed off. The order book tells you about future revenue. It tells you almost nothing about whether that revenue will come back as cash.

That framing fixes the two beginner errors. First, a bigger order book is not automatically better — a contractor can win ₹80,000 crore of work at thin margins, in states that pay slowly, and load its balance sheet with receivables that won't convert for years. Second, operating margin alone flatters the picture, because below the EBITDA line sits the interest on the debt that funds all that stuck cash. A 9% margin business that runs at 120 working-capital days and borrows to bridge it can earn a worse return on capital than a 7% margin business that gets paid in 70 days. The question is never "how big is the book?" but "how fast does the book turn into cash, and who is funding the gap until it does?"

The homely version: an EPC firm is a caterer who agrees to feed a wedding for a fixed price, buys all the food on his own credit card, and gets paid in instalments — the last one only after the in-laws stop complaining. A full order book of weddings means nothing if the cheques arrive six months late and he's paying 10% on the card.

The metrics that matter — and where they hide

For a lender you watch the spread and the staging (the discipline is similar to reading an NBFC). For an EPC contractor, eight numbers carry the story. The first two live in the income statement and the order-book release; the rest you often have to dig out of the concall transcript and the investor presentation, because they're nowhere in the standard financials.

Order book — the visibility, not the verdict

The order book (or backlog) is the value of contracts signed but not yet executed. It's the demand engine and the headline everyone quotes. Read it as a coverage ratio: order book ÷ trailing revenue tells you how many years of work are locked in. NCC carried an order book of ₹79,571 crore at end-December 2025 (Inve data, Q3 FY26) against roughly ₹20,721 crore of trailing revenue (Inve data, Q3 FY26) — about 3.8 years of visibility. KPIL stood at ₹63,287 crore (KPIL Q3 FY26 concall) on ~₹27,143 crore revenue, roughly 2.3 years. More coverage is comforting, but it is visibility, not quality — a fat, slow-paying book is a liability dressed as an asset. Other listed EPC names worth studying on the same lens include Rail Vikas, NBCC and IRB Infrastructure.

Book-to-bill — is the backlog growing or being eaten?

Book-to-bill is annual order inflow ÷ annual revenue. Above 1.0x, the company is winning work faster than it bills, so the backlog grows; below 1.0x, it's living off the existing book. KEC guided to "around ₹30,000 crore" of inflow for FY26 (KEC Q4 FY25 concall) against ~₹24,000 crore of revenue — a book-to-bill near 1.25x, healthy. KPIL guided ₹26,000–28,000 crore of inflow for FY26 (KPIL Q4 FY25 concall, revised up) on ~₹27,000 crore of revenue — close to 1.0x, meaning the book is being topped up but not surging. A book-to-bill drifting below 1.0x for several quarters is the early sign growth is about to roll over, long before revenue shows it.

Order inflow — the leading indicator the headline lags

Inflow is the flow that feeds the book. It's lumpy by nature — one large order can swing a quarter — so read it on a year-to-date and trailing basis, not quarter by quarter. The number to distrust is a management that keeps an annual inflow target alive while the YTD run-rate quietly falls behind it; that's a guide overtaken by reality. KEC, for instance, "well positioned to exceed" its ₹25,000 crore FY25 inflow guidance at the Q2 FY25 call, ended up with that order-inflow commitment marked missed in Inve's record (Inve data, Q2 FY25).

Execution rate — can it actually build what it sold?

A huge book is worthless if the firm can't execute it. The execution (or burn) rate is revenue ÷ opening order book — how fast the backlog converts to billing. Stalled execution shows up as revenue growth lagging order-book growth quarter after quarter: the book balloons, the top line doesn't, and the gap is projects stuck on land acquisition, clearances, or client-side delays. This number isn't reported directly; you compute it, and you cross-check it against management's "workflow visibility" commentary. L&T told investors its "order book is so strong that we have almost 3 years of workflow in our hand" (LT Q1 FY26 concall) — useful, but visibility is only good if the work is actually moving.

Working-capital days — the metric that decides everything

This is the one. Net working-capital (NWC) days measures how long cash is tied up in receivables, retention money and unbilled revenue, net of what suppliers fund. It is the single number most predictive of an EPC company's return on capital, and it is almost never in the income statement — you find it in the concall or the investor deck, or you back it out of the balance sheet. KEC ran 122 days at end-FY25 and 128 days a quarter later (Inve data, Q4 FY25 / Q1 FY26) against a stated 100-day target — and that 28-day gap is precisely why its margin guidance kept sliding. KPIL, by contrast, drove consolidated NWC down to 79 days by Q3 FY26, with management noting "net working capital days improved by 15 days, reaching 79 days at the consolidated level" (KPIL Q3 FY26 concall). The contrast is the lesson: same sector, same demand, 49 days of difference in how fast cash comes home.

Net debt — the working capital made visible

Working capital that won't come home has to be funded, and that funding is net debt. Watch the trend against the order book: rising debt while the book is flat means the company is borrowing to carry receivables, not to grow. KEC's net debt rose to ₹6,806 crore at Q2 FY26, and management revised its net-debt target up from ₹4,500 crore (Q1 FY26) to ₹5,500 crore for March 2026 (Inve data, Q2 FY26); the earlier ₹4,000–4,500 crore net-debt guidance from Q2 FY25 is marked achieved_diluted in Inve's record, and the FY26 ₹4,500 crore guide diluted (Inve data). KPIL ran the opposite way — consolidated net debt fell 29% QoQ to ₹2,240 crore (KPIL Q3 FY26 concall). For the deeper version of this trap, see how to spot a debt-trap stock.

Operating margin (OPM) — thin, and read against interest

EPC margins are structurally thin — high single digits is normal. KEC runs around 7% (Inve data, FY26), KPIL and NCC around 8–9% (Inve data, FY26). The trap is reading OPM alone. On a 7% margin, interest costs of ~2.9–3.2% of revenue (Inve data, KEC) consume nearly half the operating profit before a rupee of tax. So always read OPM net of interest intensity: a margin holding at 9% while interest creeps up is a return-on-capital story quietly deteriorating under a stable-looking line.

L1 pipeline — the order book before it's an order book

"L1" means lowest bidder — contracts a company has won on price but not yet been awarded. The L1 pipeline is the next layer of visibility beyond the firm order book, and it lives entirely in the concall. KEC reported a T&D "Order Book + L1" of ₹26,000 crore (Inve data, Q1 FY26) and a civil "Order Book + L1" of ₹10,000 crore. NCC described an "L1 position of Rs. 9,000–10,000 crores" on top of a "prospective pipeline of Rs. 2.45 lakh crore" (NCC Q3 FY25 concall). Treat L1 as a probability-weighted hint, not a promise — it converts, but slowly and not always.

How do you value an EPC company?

The core business is a cyclical, capital-intensive contractor, and the honest multiple is a P/E on normalised earnings, not on a peak-cycle year. Apply it through the lens of working capital and debt: two contractors on the same P/E are not equally priced if one runs 79 NWC days and ₹2,240 crore of net debt and the other 128 days and ₹6,806 crore, because the second is funding its growth with borrowed money that suppresses real returns. The multiple should compress for the debt-heavy, slow-cash name even when the order book looks identical.

The complication is that the bigger EPC houses are not pure contractors — they own assets. L&T holds stakes in a financial-services arm, a tech-services business, road and metro concessions, and a hydrocarbon engineering franchise. Valuing that on a single P/E is wrong; you need a sum-of-the-parts (SOTP): value the core EPC business on normalised earnings, then add the market or fair value of each asset-owning arm separately. L&T's group order inflow of roughly ₹3.3 lakh crore for FY25 (LT Q1 FY25 concall, ~10% growth guidance off the ₹3 lakh crore FY24 base) and FY26 group revenue near ₹2.86 lakh crore with ~₹19,000 crore of net profit (Inve data, FY26) sit inside a conglomerate where the listed subsidiaries carry their own valuations. Put the parent on a blended P/E and you'll either over- or under-value it depending on what the subsidiaries are doing. (Illustration, not a view on the stock.)

The rule of thumb: for a pure contractor, P/E on through-cycle earnings, discounted for working capital and debt; for an asset-owner, SOTP that values the concessions and subsidiaries on their own terms.

A worked case: when the order book grows and the margin still falls

Take KEC International through FY25 into FY26 — a well-run, genuinely capable contractor, which is exactly why it teaches better than a fraud would. (Illustration, not a view on the stock; figures from Inve data and the company's own concalls.)

In April 2025, management was explicit on the call: "On the margin, we have been talking about 8% to 8.5% from the current 7%" (KEC Q4 FY25 concall). The order book was healthy and growing — ₹34,409 crore in Q1 FY26, ₹36,725 crore by Q2 FY26 (Inve data). Demand was never in doubt. Yet by the October 2025 call the margin guidance had been cut to "7% to 7.5%" (KEC Q2 FY26 concall), with the candid admission: "We have said that we are at 7.1% for the 9 months… for the year will be between 7% to 7.5%" (KEC Q2 FY26 concall). In Inve's Promise Tracker, that FY26 EBITDA-margin guidance is marked revised_down (Inve data, Q4 FY25 cohort).

Now read the two numbers that explain it. Working capital never fell to the 100-day target — it sat at 122 days at end-FY25 and 128 days a quarter later (Inve data). The Q1 FY26 call laid out the plan plainly — a "reduction in Net Working Capital (NWC) from 128 days to 110 days" (KEC Q1 FY26 concall) — but the working-capital-days guidance from Q2 FY25 is marked missed in Inve's record (Inve data). And the cost of carrying that stuck cash showed up as interest at roughly 2.9–3.2% of revenue (Inve data, Q3 FY25 / Q2 FY26) and net debt rising to ₹6,806 crore — well past the earlier ₹4,500 crore target, which management revised up to ₹5,500 crore for March 2026 (Inve data, Q2 FY26).

Here's the said-versus-did in one line. The order book did its job; the cash did not. Management guided the metric the market watches — margin — and watched it sink under the metrics the market underweights: days and debt. An investor who anchored on the growing order book and the confident April margin guide would have been surprised twice. An investor watching NWC days and net debt would have seen the margin cut coming a quarter early.

Red flags specific to EPC contractors

• Order book growing, revenue flat. The book balloons while the top line stalls — projects stuck on clearances, land, or client delays. Visibility you can't execute is not value.
• NWC days rising while management calls it "improving." The most common EPC tell. A 100-day target reiterated for three straight quarters while the actual print climbs to 128 is guidance overtaken by reality. Distrust the adjective; track the number.
• Net debt rising with a flat order book. Borrowing to carry receivables, not to grow. Read the debt trend against the book, never alone.
• Margin defended only by "other income" or one-off claim settlements. When the operating margin holds because of arbitration awards or treasury income rather than execution, the core economics are weaker than the headline.
• Concentration in slow-paying clients. Heavy exposure to a single state government or one stressed sector means the receivables clock runs on someone else's cash position. NCC has flagged specific large receivables (e.g. an "Andhra Pradesh Outstanding Receivable" and JJM water-scheme receivables) in its commentary (Inve data, NCC) — the kind of concentration worth pricing.
• Aggressive bidding to fill the book. Winning low-margin work to keep the L1 pipeline full buys headlines and buries returns two years out.

Frequently asked questions

The discipline

The craft is refusing to be impressed by the backlog. The order book is the costume; the working capital and the debt are the body underneath. So invert the question you bring to an EPC result. Don't ask "how big is the order book?" Ask: if this management were quietly funding a slow-paying, low-return book with borrowed money to keep growth alive, what would the numbers look like — and does this balance sheet rule that out? Rising NWC days under a reiterated 100-day target, and net debt climbing against a flat book, do not rule it out; they are the pattern itself. This is the sequence Inve's Promise Tracker is built to surface — each margin, working-capital and net-debt guide pinned to the quarter it was made, with a verdict as later calls come in. (Illustration, not a view on the stock — a read on how management communicated through one cycle, not a lifetime verdict.)

Where this lens can be wrong. The strongest case against everything above is that working-capital days are partly outside management's control. An EPC contractor is hostage to its clients' cash — a government that defers payments, a stalled JJM water scheme, a state under fiscal stress — and even an excellently run firm will see days balloon in a bad client year through no fault of its underwriting. So a reader who always punishes rising NWC days will sometimes punish good management for a client's sins, and will miss the contractor whose days spike for one cyclical year and then normalise. The honest claim is narrower than it looks: reading days and debt against the order book tells you when growth is being bought with borrowed cash. It does not, by itself, tell you whether that's the management's failing or the client's — for that you need the concall, the client mix, and a few years of pattern. We haven't modelled the project-level claims and arbitration awards that can swing a year, and we won't pretend a transcript substitutes for that.

And the owner's question, the one to sit with before you buy a single share of any contractor: across the next down-cycle — not this quarter's order win — what must I believe about how fast this company collects its cash and how little it has to borrow to keep building, for it to still be compounding on the other side? If the honest answer leans on a growing order book rather than on days and debt coming home, you've read the costume, not the business.

Related reading

• How to Analyse a Bank Stock (NIM, CASA, GNPA, ROA)
• How to Analyse a Building Products Company (Pipes, Tiles)
• How to Analyse a Cables and Wires Stock