How to Analyse an ER&D Engineering Services Company

On its January 2026 call, KPIT was asked the only question that really decides an ER&D stock: what are its customers' R&D budgets doing? The answer was blunt. "From a passenger car perspective, their spend has gone down by 20% to 25%. It's a dramatic… their volumes have gone down substantially, their profits have gone down even more," said joint MD Sachin Tikekar (KPIT Q3 FY26 concall). An analyst on the same call put it plainly: "mobility ER&D budgets have fallen sharply last year." (Illustration of how to read the sector, not a view on the stock.)

Sit with that number, because it is the whole sector in one sentence. An engineering research-and-development (ER&D) services firm doesn't sell generic IT — it sells the outsourced product-development of its clients: the software inside a car, the avionics in a jet, the firmware in a medical device. Its revenue is, almost literally, a slice of someone else's R&D line item. And an R&D line is the first thing a squeezed manufacturer cuts. So an ER&D firm carries a cyclicality that a plain IT-services company — selling cost-takeout and run-the-business work that clients can't switch off — simply does not. KPIT's own revenue still grew through that 20-25% client budget cut, because it gained wallet share; but the gravity it was fighting is the thing a generic margin-and-growth screen never shows you.

We'll admit our own early mistake, because it's the one that catches most readers. When we started tracking ER&D managements, we read them like IT-services firms — watching the headline constant-currency growth and the EBITDA band. That misses the point. The growth and the margin in this sector are outputs of two things a P&L doesn't carry: which verticals the firm is exposed to, and where each of those clients' R&D budgets sit in their own cycle. Read those, and the financials stop surprising you.

This piece is about the handful of operating numbers that actually drive an ER&D business, where they hide (mostly in the concall and the investor deck, not the income statement), and the one disclosure habit that should make you nervous. One boundary up front: you cannot forecast a client's R&D cycle from outside. You can read which way it's turning and test whether management is honest about it. That is the circle this stays inside.

What actually drives the economics — and why it's not plain IT services

A plain IT-services firm is a labour-arbitrage machine selling billable hours to run a client's existing systems. A SaaS firm sells a product and lives on recurring revenue. An ER&D firm is neither: it is a verticalised, project-based R&D contractor. It wins a programme — design the software-defined-vehicle stack, certify an aircraft sub-assembly, build the medical-device firmware — staffs deep domain engineers against it, and bills as the programme runs. Three consequences fall out of that structure, and they are what you analyse.

First, the vertical is the business. An IT-services firm is broadly diversified across BFSI, retail, healthcare. An ER&D firm is concentrated by engineering domain — and each domain has its own cycle. Tata Elxsi's transportation vertical was more than 55% of revenue in Q3 FY26 (Tata Elxsi Q3 FY26 concall); KPIT is essentially a pure automotive ER&D play. When autos cut R&D, there is nowhere for that revenue to hide.

Second, the work is discretionary and cyclical. New-product engineering is exactly what a manufacturer postpones when its own margins compress — the 20-25% passenger-car budget cut KPIT described is the mechanism. This is why ER&D earnings swing harder than IT-services earnings even when both report "softness."

Third, the moat is domain depth, not scale. You don't win a powertrain or avionics programme on price; you win it on engineers who already speak the client's regulatory and physical constraints. That depth is why these firms can command a P/E premium — and why losing a marquee vertical client hurts more than losing a generic account.

The homely version: a plain IT firm is the utility company keeping your lights on — boring, sticky, hard to switch off. An ER&D firm is the architect you hire to design the new wing. In a good year you build; in a tight year the wing waits, and the architect's order book goes quiet while the utility keeps billing. Everything below is a way of telling which year the architect is in.

The metrics that matter — and where they hide

These are sector-specific. A generic ROE-and-revenue-growth read will mislead you, because the forces that move an ER&D firm live one layer below the income statement.

Revenue by vertical — the cycle map

The split across auto/transport, aerospace, medtech, industrial and hi-tech/telecom is the single most important disclosure, because each vertical is on its own R&D cycle. Where it hides: the investor deck's segment slide and the concall, not the P&L. Why it matters here: concentration cuts both ways — Tata Elxsi's >55% transportation tilt (Tata Elxsi Q3 FY26 concall) means an auto upturn lifts it hard and an auto R&D freeze hits it hard; a firm spread across auto, aero and medtech is buying itself cycle diversification. What good looks like: either genuine diversification, or deep concentration in a vertical that is structurally growing (SDV, electrification). Real number: Tata Technologies' aerospace business — a counter-cyclical hedge to its auto exposure — grew 19% QoQ in Q3 FY26 and was tracking ~$40m for FY26 (Tata Technologies Q2/Q3 FY26 concalls), even as its automotive clients tightened.

SDV / EV / ADAS exposure — the structural growth inside the cycle

Software-defined vehicles, electrification and driver-assistance are the secular wave inside an otherwise cyclical auto vertical. A firm levered to software content per vehicle can grow even while overall auto volumes and budgets fall. Where it hides: concall Q&A and deck — almost never quantified in the financials. What good looks like: new-programme wins explicitly tied to SDV/EV, ramping into revenue. Real number: Tata Elxsi's transportation business grew 7.7% QoQ in Q3 FY26 "led by accelerated ramp-ups in SDV-led OEM deals" (Tata Elxsi Q3 FY26 concall) — growth coming from the software-content shift, not from auto volumes, which were flat to down. The tell to listen for: is the growth from more cars (cyclical) or more software per car (structural)?

Deal TCV and the order book — the leading indicator

Total contract value of large deals won, and the order book, tell you what revenue is coming before it arrives. Where it hides: concall and deck. Why it matters here: ER&D programmes are multi-year, so a strong TCV clip is a more durable signal than a single quarter's revenue. What good looks like: a rising large-deal clip and an order book growing faster than revenue. Real number: LTTS booked $855m of large-deal TCV in FY26, up 40% year-on-year, and has stated its aspiration to move from a "$200 Mn clip to a $300 Mn clip" per quarter (LTTS Q4 FY26 and Q2 FY26 concalls). Watch the clip, not the headline annual figure — a rising quarterly run-rate is the real expansion.

Utilisation — the margin lever you can read a quarter early

The share of billable engineers actually deployed on programmes. Every benched engineer is a domain specialist paid with no revenue against them, so utilisation is the most direct margin lever in the sector. Where it hides: concall, occasionally the deck. What good looks like: rising toward the firm's historical peak with headroom left. Real number: Tata Elxsi was at ~75% utilisation in Q3 FY26 with management noting historical peaks of "85%, 86%" — so a ~10-point lever sitting unused, worth roughly "200 basis point uplift" to margin from operating leverage (Tata Elxsi Q3 FY26 concall). Falling utilisation is the earliest warning that a vertical's demand is softening; the revenue line confirms it a quarter or two later.

Offshore mix and EBIT/EBITDA margin — the cost engine

Offshore mix (work done from low-cost India centres versus onsite) and the operating margin are the cost side. ER&D margins typically sit above plain IT services for the high-domain verticals and well below for commoditised work. Where it hides: margin is in the P&L; offshore mix and segment margins are in the deck/concall. What good looks like: offshore mix rising and the spread between verticals understood. Real number: LTTS's Q4 FY26 EBIT margin was 15.2% blended — but underneath, its Sustainability segment ran at 28.7% while its Tech segment ran at 12.6% (LTTS Q4 FY26 concall). The blended number hides everything; the segment spread tells you which verticals are actually carrying the firm.

Attrition and client concentration — the risk pair

Attrition matters more in ER&D than IT services because domain engineers are scarce and slow to replace; losing them loses programme knowledge. Client concentration matters because a single marquee OEM can be a large share of a vertical. Where it hides: concall; concentration sometimes in the annual report. Real numbers: KPIT ran ~7% attrition in Q1 FY26 — very low, but read it against the demand backdrop (LTTS ~14.7%, Tata Technologies ~16.2% LTM in Q4 FY26; all from respective FY26 concalls). And concentration risk is concrete here: when KPIT's passenger-car clients cut R&D 20-25%, it kept wallet share in its top-25 ("we have not lost our wallet share in any of our T25 clients," KPIT Q3 FY26 concall) — but a firm less embedded would have lost revenue outright. Embeddedness is the defence; concentration is the exposure.

How do you value an ER&D firm without overpaying for the cycle?

ER&D firms usually trade at a premium P/E to plain IT services — the market pays up for niche domains, SDV/EV secular growth, and higher-quality earnings. The trap is treating that premium as a constant. It isn't: the earnings underneath it swing with client R&D budgets, so a P/E that looks fair at peak budgets looks expensive the moment the cycle turns.

So read the multiple growth-adjusted, not absolute. The discipline: take the firm's normalised earnings power across a cycle — not this quarter's, which may be flattered by peak utilisation or depressed by a client budget freeze — and ask what growth rate the premium implies. Then check that growth against the only durable driver: structural software-content growth (SDV, EV, electrification) rather than cyclical client-volume growth. A P/E premium underwritten by SDV content per vehicle is defensible; the same premium underwritten by a one-off auto upcycle is borrowed time.

A live caution on why this matters: Tata Elxsi's quarterly net profit dipped to ₹109 crore in Q3 FY26 from ₹155 crore in Q2 FY26 — roughly a third lower — even as the transportation vertical grew 7.7% QoQ, dragged down by a one-time New Labour Code exceptional charge rather than any demand collapse; it then recovered to ₹220 crore in Q4 FY26 once that charge dropped out (Inve data; Tata Elxsi Q3 FY26 concall). A reader paying a premium multiple on a single quarter's earnings would have anchored on a number that a one-off accounting item — not the underlying business — could swing by a third. Same instinct as separating reported profit from durable cash earnings: in ER&D, ask which part of the earnings is the cycle and which part is the structure.

A worked case: KPIT and the 20% budget cut it grew through

Put KPIT's Q3 FY26 call together as one picture, because three forces converge on a single quarter — and the convergence, not any one number, is the lesson. (Illustration, not a view on the stock; figures from the KPIT Q3 FY26 concall and Inve data.)

Here is the uncomfortable part, and why it teaches more than a blow-up would: nothing here is dishonest. The wallet-share gains are real. The 21% margin is real. But laid side by side, they say something a single line never could — this is a firm running hard to stand still inside a vertical whose budget pool is shrinking, and even a disciplined, well-embedded operator missed its own near-term growth guidance when the cycle bit. The pivot KPIT described — from selling services to selling pre-built "solutions" so it captures more of a smaller budget — is the rational response. Whether it works is the owner's question; that the attempt is needed is the diagnosis.

This is also why the concall matters more here than the financials: the deck gives you the growth rate; the Q&A is where an analyst forces management to say what the client budgets are actually doing. Inve's concall summaries pull every such exchange — the question, the speaker, the quote — into one place per quarter, so the budget-cut admission doesn't sit buried 40 minutes into the Q&A.

The red flags specific to ER&D

• Heavy concentration in one vertical at the top of its R&D cycle. A >50% auto tilt is a strength in an upcycle and a trap when OEM budgets freeze. Always ask where that vertical's clients are in their cycle, not the ER&D firm's.
• Growth from client volumes, not software content. If the SDV/EV/electrification story is in the deck but the growth is really just an auto-volume upcycle, the premium multiple is borrowed from the next downturn.
• Utilisation already near its historical peak. If a firm is at 85% with no headroom, the easy margin lever is spent — further margin gains need pricing or mix, which are harder.
• A marquee client coming off peak run-rate with nothing ramping behind it. A thin pipeline behind a single big account is a single point of failure — when that programme rolls off, there is nothing to backfill the revenue gap.
• The disclosure that quietly disappears. When a firm that used to break out vertical mix, large-deal TCV, or an SDV win count stops — and nobody on the call presses to get it back — that silence is usually the metric that turned against them. The number a management stops volunteering is the one to ask about first.
• Rising attrition in scarce domain skills. In ER&D, an engineer leaving takes programme knowledge with them; persistent attrition above peers in the same vertical is a delivery risk, not just a cost.

A management that volunteers its vertical-by-vertical R&D-budget read, its utilisation headroom, and its SDV-versus-volume growth split is telling you it knows where it stands. One that keeps the conversation on blended constant-currency growth is telling you where it would rather you not look.

Where this read can be wrong

The strongest case against everything above is that ER&D cyclicality is transitional, and an investor who treats SDV/EV as "just another auto cycle" will sell the structural compounders too early. The software content of a vehicle is rising secularly — by some counts toward half of a car's value over the next decade — and a firm deep in SDV may grow straight through auto-volume downturns the way KPIT partly did, because content-per-vehicle is decoupling from units-per-year. If that decoupling is real and durable, then reading these firms as cyclical contractors understates them, and the premium multiple is not borrowed time but a fair price for a structural shift. The metrics here can tell you whether this quarter's growth is cyclical or structural; they cannot tell you how long the structural wave runs, and that — not the utilisation rate — often decides the outcome.

So the honest claim is narrower than it looks. Reading vertical mix, SDV exposure, TCV, utilisation and the client-budget commentary against management's own guidance tells you whether the firm's growth story survives its own numbers, and whether the premium is underwritten by structure or by the cycle. It does not tell you who wins the SDV platform war, or how deep the next auto-R&D downturn runs — and in this sector, those frequently decide the return. We are confident about the method; on which firm compounds, far less so. We have been wrong on the timing of these cycles before — an auto R&D freeze that looked like one bad quarter has run for several.

A hard limit worth stating plainly: a clean concall and a deep order book do not immunise an ER&D firm against a client-side recession, a marquee programme cancellation, or clients pulling engineering back in-house. Reading management well lowers your odds of being surprised; it does not predict the R&D cycle.

For the metrics that don't surface on the concall — programme-level contingencies, related-party dealings with a parent OEM (relevant for the Tata-house and L&T-house firms), and segment detail — pair this with how to read an annual report like an analyst. And because an ER&D firm shares the labour and currency mechanics of plain IT services, the IT-services guide is the right companion for the TCV-quality and constant-currency discipline; the new-age internet guide shows the same "read the business model, not the generic ratio" habit applied to a very different one.

Inve's KPI Screener lines these metrics up across ER&D firms — vertical mix, TCV, utilisation and margin where disclosed — with YoY/QoQ trends and a data-confidence flag per number, so the peer comparison takes minutes. The Promise Tracker pins each forward commitment — a utilisation target, a margin band, an SDV-revenue aspiration — to the quarter and quote it was made in, then marks it as later calls confirm or quietly drop it.

Frequently asked questions

Reading an ER&D firm well is the discipline of looking past this quarter's revenue to the R&D budgets that fund it and the software content that's reshaping them. The financials tell you what the firm did; the vertical mix, the SDV exposure, the order book and the utilisation tell you what it will do — and they speak a quarter or two before the income statement confirms it. So invert the question. Don't ask "is this growing?" Ask: if this firm's growth were really just a client's good year — peak budgets, peak utilisation, a marquee programme at its run-rate peak — what would the numbers look like, and does the disclosure rule that out?

And the owner's question, the one to sit with before buying any of these: what must I believe about the next five years of my clients' R&D budgets — and about how much of a car, a jet, a device becomes software — for this firm to earn its premium? If the honest answer leans on a single vertical at the top of its cycle rather than a structural content shift the firm is genuinely positioned for, you've priced the good year, 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