How to Build a DCF Model: 6 Steps With an India Example

What Is a DCF Model?

A DCF (discounted cash flow) model values a company as the present value of all the cash it will generate in the future. Instead of asking what the market pays for similar stocks, it asks what the business itself will produce — and what that cash is worth in today's rupees.

You build one in six steps: project free cash flows for five years, calculate a discount rate (WACC), estimate a terminal value, discount everything to present value, bridge from enterprise value to equity value per share, and sanity-check the output with a sensitivity grid.

This guide runs all six steps on a fully illustrative listed Indian consumer company — ₹1,000 crore revenue, deliberately round numbers — and lands at about ₹173 per share. The only real-world figures we use are the discount-rate inputs: India's 10-year G-Sec yield of about 6.7% (early July 2026) and Damodaran's 7.08% India equity risk premium (January 2026).

DCF is the right tool for businesses with forecastable cash flows — mature consumer companies, IT services, utilities, manufacturers. It is the wrong tool for banks and NBFCs, where debt is raw material rather than financing, and for pre-revenue startups with nothing yet to project.

The Intrinsic-Value Logic

Two ideas power the whole machine: a rupee next year is worth less than a rupee today (today's rupee can be invested; next year's might never arrive), and risk has a price — the shakier the promise, the higher the return you demand for waiting.

The discount rate carries both ideas. Discount every future cash flow at the return you require, sum the present values, and you have intrinsic value — the analysis an analyst covering a Titan or an Asian Paints runs before publishing a target price.

DCF vs Relative Valuation

Relative valuation (comps) prices a company off what similar businesses trade at — P/E or EV/EBITDA multiples. It is fast and market-anchored, but it can never tell you whether the entire sector is mispriced.

DCF makes the opposite trade-off: anchored to the company's own cash generation, indifferent to market mood, hostage to your assumptions. Practitioners run both — the DCF builds the thesis, the comps reality-check it. Our guide to the types of financial models maps where each fits.

Step 1: Project Free Cash Flows (FCFF)

The first step of a DCF model is projecting free cash flow to the firm (FCFF) — the cash left for all capital providers after the business pays operating costs, taxes, and reinvestment. The convention is a five-year explicit forecast, built driver by driver from revenue downward, using this formula:

FCFF = EBIT × (1 − Tax rate) + D&A − Capex − ΔNWC

Each term earns its place. EBIT × (1 − tax) is after-tax operating profit as if the company had no debt — interest is deliberately excluded, because financing cost lives in the discount rate; counting it here would double-charge it. Add back D&A because it is non-cash; subtract capex and incremental working capital because that cash genuinely leaves.

Resist typing a growth percentage from thin air. Decompose revenue into drivers you can defend — volumes × realisation for a manufacturer, stores × revenue per store for retail — and let the percentage fall out. For our illustrative company we keep the drivers deliberately simple:

• Revenue: ₹1,000 crore base, growing 10% a year
• EBIT margin: 20% of revenue
• Tax rate: 25%
• D&A: 4% of revenue
• Capex: 6% of revenue — above D&A, because growth must be paid for
• Working capital: incremental NWC absorbs 2% of revenue each year (in a full model you would hold NWC as a percentage of revenue and take the year-on-year change — see our three-statement build guide)

Run them forward:

All figures illustrative and rounded to one decimal, so a recomputed cell can differ by 0.1.

In Excel, keep every assumption in its own input cell (blue font for inputs, black for formulas), then roll forward: revenue as =B5*(1+$B$2), FCFF as =EBIT*(1-$B$3)+DA-Capex-dNWC. One driver, one row — an auditable model beats a clever one.

Sanity checks before moving on: FCFF here settles at 11% of revenue every year because every driver scales with revenue — in a real model that ratio should drift for reasons you can narrate. And reinvestment must support growth: 10% growth with capex below D&A is a model quietly claiming free lunches.

Step 2: Calculate the Discount Rate (WACC)

FCFF belongs to both lenders and shareholders, so it is discounted at the weighted average cost of capital (WACC) — the blended annual return your capital providers require. Cost of equity comes from CAPM, cost of debt is your after-tax borrowing rate, and the two are weighted by market values:

Cost of equity = Risk-free rate + Beta × Equity risk premium

Here we switch from illustrative to real inputs. The risk-free rate for a rupee DCF is the 10-year G-Sec yield — about 6.7% in early July 2026 per Trading Economics; bond yields move daily, so pull the live figure the day you build. For the equity risk premium, the standard reference is Aswath Damodaran's country risk premium table at NYU Stern, which in its January 2026 update puts India's total ERP at 7.08% — a 4.23% mature-market premium plus a 2.85% India country risk premium.

One currency warning. Damodaran's January 2026 data update derives that 4.23% mature-market ERP against a 4.18% US T-bond rate — a dollar risk-free rate. It belongs in dollar DCFs only; rupee cash flows take the G-Sec. Mixing them is mistake #2 later.

The company-specific inputs stay illustrative: beta of 1.0, a 9% pre-tax cost of debt (6.75% after tax at 25%), and market-value weights of 75% equity / 25% debt. The arithmetic:

• Cost of equity: 6.7% + 1.0 × 7.08% = 13.78%
• After-tax cost of debt: 9% × (1 − 0.25) = 6.75%
• WACC: 0.75 × 13.78% + 0.25 × 6.75% = 12.02% — call it 12%

In Excel this is four input cells and two formulas: =rf+beta*ERP for cost of equity, =We*Ke+Wd*Kd*(1-tax) for WACC. The built-in sanity check: WACC must land between the after-tax cost of debt and the cost of equity — ours does.

Step 3: Estimate Terminal Value

Terminal value captures everything beyond the five-year forecast — usually the majority of a DCF's value, so treat it with respect. The standard tool is the Gordon growth (perpetuity) formula, which values a cash-flow stream growing at a constant rate g forever:

Terminal value = FCFF_{Year 5} × (1 + g) ÷ (WACC − g)

For our illustrative model, take g = 5%: TV = 177.2 × 1.05 ÷ (0.12 − 0.05) = ₹2,658.0 crore. Note this is the value as at the end of Year 5 — it still has to be discounted back to today in Step 4.

The discipline is in choosing g. A company growing faster than the economy forever would eventually become the economy, so terminal growth must sit below the long-run nominal GDP growth of wherever the company earns. Keep it nominal (your cash flows include inflation), keep it modest, and never let it crowd the WACC — as g approaches WACC, the denominator collapses and the formula manufactures infinity.

The alternative is an exit multiple — value the Year-5 business at a comps-based EV/EBITDA. It is the banker's habit, but it smuggles market mood into an intrinsic model. Better: use Gordon growth, then cross-check the implied multiple. Our TV of ₹2,658.0 crore against Year-5 EBITDA of ₹386.5 crore (EBIT 322.1 + D&A 64.4) implies about 6.9× EV/EBITDA — if comps trade nowhere near that, your g is claiming something the market doesn't believe.

Step 4: Discount Everything to Present Value

Now convert every future rupee into today's rupees: divide each cash flow by (1 + WACC)^t, where t is the year it arrives. The terminal value sits at the end of Year 5, so it is discounted five years, not six — and yes, it must be discounted; the undiscounted TV is the most common beginner error we see.

Illustrative figures. PVs computed at full precision then rounded to one decimal, so multiplying by the 4-decimal discount factors shown can differ by ₹0.1 crore.

Read the split: the five forecast years contribute ₹521.2 crore of value; the terminal value contributes ₹1,508.2 crore — about 74% of enterprise value. That is normal for a five-year DCF, but it tells you where your risk lives: the valuation leans harder on perpetuity assumptions than on the forecast you sweated over.

Two practitioner notes. First, the mid-year convention: cash arrives through the year, not at midnight on 31 March, so many analysts discount at t − 0.5 — which multiplies every discount factor by (1.12)^0.5 ≈ 1.058, lifting value about 5.8% here. Year-end is fine for a first model; just be consistent and label it.

Second, =NPV(rate, range) assumes the first cell sits one full year away — easy to misuse. Safer to build an explicit discount-factor row, =1/(1+$WACC)^t, and multiply. Quick check: at 12% the Year-5 factor is ≈ 0.567 — a rupee five years out is worth about 57 paise.

Step 5: From Enterprise Value to Value per Share

Enterprise value belongs to everyone who financed the business — lenders included. Shareholders get what remains after debt-holders and other claimants are carved out, so the bridge from EV to a per-share number runs through the balance sheet:

• Subtract net debt: gross borrowings minus cash and equivalents — ₹250 crore in our illustrative company, taken from the latest balance sheet.
• Subtract minority interest: ₹50 crore here. Consolidated cash flows include subsidiaries you don't fully own; this removes the slice that isn't yours.
• Add non-operating assets where genuine — listed investments, surplus property — anything producing value that never flowed through your FCFF forecast.

The arithmetic: ₹2,029.4 − 250.0 − 50.0 = ₹1,729.4 crore of equity value. Divide by 10 crore shares outstanding and you get ₹172.9 — call it ₹173 per share. Use the diluted share count in real models; ESOPs and convertibles quietly expand the denominator.

Interpretation is a comparison, not a command. If the stock trades at ₹140, your model says cheap; at ₹220, expensive. Either way the gap is a hypothesis, not a verdict — it survives only if the assumptions survive Step 6.

Step 6: Sanity-Check With Sensitivity Analysis

A single-point DCF is false precision. Two dials dominate the output — WACC and terminal growth — so the professional habit is a two-way sensitivity grid showing value per share across plausible combinations of both. Here is our illustrative model's grid:

Illustrative. Each cell reruns the full model — five discounted FCFFs plus terminal value, less ₹300 crore of net debt and minority interest, divided by 10 crore shares.

Read what the grid is telling you. Nudging two assumptions by a percentage point each swings the answer from ₹132 to ₹246 — an 86% spread from the same spreadsheet. If your buy case only works in the optimistic corner, you don't have a thesis; you have an assumption wearing a thesis costume. Present the range, and defend the cells you believe.

In Excel, don't build 25 models — build one and let a Data Table sweep it: put =value_per_share in the grid's corner cell, g across the top row, WACC down the first column, then Data → What-If Analysis → Data Table — g as the row input, WACC as the column input.

Beyond the grid, run the standing sanity checks: terminal value share of EV (74% here; above ~85–90%, your forecast period is saying almost nothing), the implied exit multiple from Step 3 (6.9×), FCFF margin drift, and whether reinvestment still supports assumed growth.

The 7 Mistakes That Break Most First DCF Models

We see the same seven failures on repeat in student and analyst models. Audit any DCF — yours or someone else's — against this list first:

• Terminal growth outrunning the economy. A g at or above WACC breaks the formula outright; a g above long-run nominal GDP growth breaks it philosophically. Keep it comfortably below both.
• Currency and inflation mismatch. Rupee cash flows discounted at dollar rates (like Damodaran's 4.18% US T-bond figure) or real flows at nominal rates. Match currency to currency, nominal to nominal.
• Growth nobody pays for. Revenue compounding at 15% while capex and working capital stay flat. Reinvestment is the price of growth — a model skipping the bill is fiction.
• Cash-flow / discount-rate mismatch. FCFF pairs with WACC; free cash flow to equity pairs with cost of equity. Crossing them double-counts or drops the debt-holders' claim.
• A sloppy equity bridge. Stale net debt, forgotten minority interest, basic instead of diluted shares — EV right, per-share wrong.
• The single-number answer. No sensitivity grid means no idea how fragile the output is. One cell is an opinion; the grid is analysis.
• An unexamined terminal value. At 74% of EV, ours is normal; above ~90%, the "forecast" is decoration. Extend the explicit period or admit you are really pricing the perpetuity assumptions.

These seven are interview ammunition — "walk me through a DCF" is a standard opener, and our financial modeling interview questions guide drills exactly this territory.

How Do You Learn to Build DCF Models Properly?

DCF competence comes from repetitions, not reading. The path that works: rebuild this exact model in Excel from scratch — all six steps, no template. Then pull a real listed company's annual report and construct FCFF from its actual statements, where depreciation hides in the notes and working capital misbehaves. Then vary drivers and watch the grid respond.

If you are brand new, start one level earlier with what financial modeling actually is — a DCF sits on a three-statement foundation. The skills pay, too: valuation modeling is core to the roles in our financial modeling salary breakdown — equity research, investment banking, corporate development.

The fastest route is guided reps. QuintEdge's Financial Modeling and Valuation course builds DCFs live on real Indian companies — you model alongside the instructor, submit your file, and get it torn apart kindly. Defend every cell out loud and you are interview-ready.