A DCF model estimates what a business is worth today by forecasting its future cash flows and discounting them back to present value. In investment banking, private equity, and M&A, it is one of the clearest ways to test valuation, pressure test assumptions, and frame what a business can realistically be worth.
A good DCF model does not give you one perfect answer. It gives you a valuation range based on operating assumptions, capital intensity, working capital needs, and the discount rate. That is why it is so widely used in financial modelling, fairness work, deal evaluation, and internal investment committee discussions.
This walkthrough shows how to build a DCF model from scratch, what each section should do, and where people usually get it wrong.
If you want to learn the mechanics using a clean Excel file instead of building everything from a blank sheet, you can use my advanced DCF model. It is built to help you understand the logic and apply it to on-the-job scenarios.
A discounted cash flow model converts future cash generation into a present value today. In most cases, you forecast unlevered free cash flow, discount it using WACC, arrive at enterprise value, and then bridge from enterprise value to equity value.
The logic is simple:
If that flow is wrong, the rest of the model will be wrong too.
Before opening Excel, decide what exactly you are valuing.
If you are building a standard corporate DCF, you will usually value the firm using unlevered free cash flow and WACC. That gives you enterprise value first. You then move from enterprise value to equity value.
If you are valuing equity directly, you would use levered cash flow and cost of equity. That approach is less common in investment banking and private equity work. Most interview tests, sell side models, and internal valuation files use unlevered free cash flow.
Takeaway: Match the cash flow to the discount rate. Unlevered free cash flow goes with WACC. Levered free cash flow goes with cost of equity.
A DCF is only as good as its inputs. Start with the historical financial statements, then collect the operating, market, and capital structure data that will support the forecast.
You will usually need:
For public companies, common sources include annual reports, investor presentations, earnings transcripts, and filings. For private companies, this often comes from management accounts, CIMs, quality of earnings reports, and lender materials.
Do not collect data just because it is available. Collect data that supports the forecast drivers you intend to use.
Revenue is the starting point of most DCF models because nearly every other line item depends on it.
There are two broad ways to forecast revenue. The first is a top down approach, where you start with market size, market growth, and share assumptions. The second is a bottom up approach, where you build revenue from unit volumes, prices, customers, contracts, stores, subscriptions, or segments.
For a serious model, bottom up is usually stronger. It forces you to explain where growth is actually coming from.
Questions to answer:
A flat average of past growth rates is rarely good enough on its own. Use history as context, not as autopilot.
Once revenue is in place, forecast the major cost lines and build to EBIT or EBITDA.
Typical lines include:
Some costs move closely with revenue. Others are fixed, semi fixed, or driven by headcount, footprint, or strategy. That distinction matters. A business with operating leverage should not be modelled as if every cost is a flat percentage of sales.
This is also where you should think about margin trajectory. Does the company have room for scale benefits? Is there pressure from wages, input costs, or pricing? Does the forecast assume margin expansion that management has never delivered before?
A DCF often rises or falls on margin assumptions more than on revenue growth.
This is where weak models usually start to break.
A company can show strong EBITDA growth and still destroy value if it requires heavy capital expenditure or large working capital investment to support that growth. That is why a DCF has to move past earnings and into cash generation.
Forecast maintenance capex and growth capex separately if possible. Maintenance capex keeps the business running. Growth capex supports expansion. Lumping them together can hide what the business truly needs to sustain its cash flow.
Depreciation should tie back to the asset base and capital spending over time. A simple percentage of revenue can work in a rough model, but a more detailed schedule is better for mature work.
Forecast accounts receivable, inventory, accounts payable, and other operating working capital items based on days or turnover ratios where appropriate.
Watch for the direction of change. A growing business often consumes cash through working capital. That cash outflow needs to be reflected in the DCF.
This is also the part of the process where many people realize theory is not enough. Our DCF model includes editable assumptions, projections, sensitivity tables, and dashboard style outputs so you can see how these moving pieces connect inside a full model.
This is the heart of the model.
For a standard DCF, the usual formula is:
Unlevered Free Cash Flow = EBIT × (1 − Tax Rate) + D&A − Capex − Change in Net Working Capital
This formula matters because it measures cash flow available to all capital providers before interest payments. That is why it pairs with WACC.
A common mistake is to start with net income and then discount the result with WACC. That mixes equity cash flow with a firm level discount rate. The output may look neat, but the logic is off.
If you are building a full DCF model in Excel, each forecast year should clearly show:
Keep that bridge clean. If someone cannot follow how you moved from operating profit to cash flow, the model will be hard to defend.
The weighted average cost of capital reflects the blended required return of debt and equity investors.
The formula is:
WACC = (E / (D + E)) × Re + (D / (D + E)) × Rd × (1 − Tc)
Where:
This is usually estimated through CAPM:
Cost of Equity = Risk Free Rate + Beta × Equity Risk Premium
That means you need to decide on the risk free rate, beta, and market risk premium. For private companies, beta is often estimated using public comps and relevered to match the target capital structure.
Use the company’s current borrowing cost, market yield, or an estimated spread based on credit risk. Then apply the tax shield if you are using WACC.
Do not spend time making WACC look precise to two decimal places if the inputs are weak. A false sense of precision is still false.
Once you have forecast unlevered free cash flow and estimated WACC, discount each forecast period back to present value.
The basic formula is:
PV of FCF = FCF / (1 + WACC)^t
Where t is the year number in the forecast period.
Some models use year end discounting. Others use a mid year convention. Mid year convention is often more realistic because cash flow is generated throughout the year, not only at year end.
If you are building a simple model, year end discounting is acceptable. If you want a cleaner professional output, use mid year convention and be explicit about it.
Terminal value usually accounts for a large part of total DCF value. That is exactly why it needs to be handled with care.
There are two standard methods.
Terminal Value = FCF in Year n+1 / (WACC − g)
This assumes the business continues beyond the explicit forecast period and grows at a stable long term rate.
The growth rate must be conservative. If you set terminal growth too high, the valuation can become overstated very quickly. In most cases, the perpetual growth rate should sit below or around long term nominal GDP growth, depending on the business and geography.
This applies a market multiple, usually EV / EBITDA or EV / EBIT, to the final forecast year.
This method is common in private equity and deal work because it links the DCF to observed market pricing. Still, it should be consistent with trading comps, precedent transactions, and the company’s quality at exit.
A useful discipline is to calculate terminal value both ways and compare the implied result. If they tell very different stories, one of your assumptions is probably too aggressive.
After discounting the forecast cash flows and terminal value, you sum them to get enterprise value.
Enterprise Value = Present Value of Forecast FCFs + Present Value of Terminal Value
From there, bridge to equity value:
Equity Value = Enterprise Value − Net Debt − Other Debt Like Claims + Non Core Assets
Depending on the company, this bridge may also include:
Do not assume equity value is just enterprise value minus debt plus cash in every case. Some businesses have more going on.
A DCF should always show a valuation range, not a single point estimate.
The most common sensitivity table flexes:
You can also sensitize:
This is where the model becomes genuinely useful. A DCF is about seeing which assumptions are doing the heavy lifting.
If a one turn change in exit multiple or a 50 basis point move in WACC changes value materially, that should shape how you talk about valuation risk.
If your goal is to improve fast, a structured model helps far more than reading formulas in isolation. My DCF model includes WACC, terminal value methods, IRR outputs, sensitivity analysis, and a clean presentation layer so you can see how professionals lay this out in practice.
Before trusting the result, ask a few basic questions.
Does the implied valuation line up with public comps and precedent transactions?
Does the forecast imply margins, growth, or capital efficiency that the business has never shown before?
Is terminal value too large as a percentage of total enterprise value?
Are working capital and capex assumptions realistic for the sector?
A DCF should support your valuation view, not replace judgment.
Even strong candidates make the same errors repeatedly.
If you use unlevered free cash flow, use WACC. If you use levered free cash flow, use cost of equity.
A 5 year projection filled with flat percentages is not a real forecast. Revenue, margins, capex, and working capital should each have a reason.
Small changes in terminal growth can move value more than people expect. Be conservative.
Preferreds, minorities, leases, and excess cash can all change the final equity value.
A DCF is one valuation method. It should sit alongside comps, precedents, and commercial judgment.
A clean DCF model usually has the following tabs:
A reviewer should be able to follow the model from top to bottom without hunting for hard coded numbers.
Some people dismiss the DCF because terminal value often drives a large part of the output. That criticism is partly fair. A sloppy DCF can tell you whatever you want it to tell you.
But a good DCF still earns its place. It forces you to make the operating story explicit. It shows how value changes when assumptions change. It exposes weak thinking on margins, reinvestment, and cost of capital. In deal work, that is useful.
A comparable companies analysis shows how the market prices similar businesses. A precedents analysis shows what buyers have paid. A DCF shows what the company has to deliver for a given valuation to make sense.
A strong DCF model is not built by memorizing formulas alone. It is built by linking business drivers to cash flow, linking cash flow to valuation, and testing whether the final answer is credible. If you can explain each step clearly, from revenue build to terminal value, you are already ahead of most people who treat the DCF as a black box.
If you want to move from theory into actual modelling, you can get the advanced DCF model here. It includes lifetime access, instant download, a top tier firm style structure, key Excel formulas, public company valuation logic, and the assumptions and outputs you would expect in a professional DCF template.
For anyone preparing for internships, analyst roles, or buy side interviews, this is a practical way to sharpen financial modelling skills using a format built for real use rather than classroom examples.
P.S. – don’t forget to check our Premium Resources where you will find valuable content to help you break into the industry.
