EBITDA in IT: why everyone measures by it and how to read a company fast

What the metric actually shows, how it is calculated, where it gets inflated, and how to size up a company from just two numbers

When people in IT talk about a company, two numbers almost always come up: headcount and EBITDA. Not net profit, not revenue on its own — EBITDA. Valuations at exit are built on it, companies are compared with it, and it dominates conversations with investors and between owners.

Yet the metric itself is understood very differently. Some treat it as a synonym for operating profit, others as a way to hide real costs, and many simply take the number from a deck without asking how it was calculated.

Let's work through it: what EBITDA really is, why it became the standard, how IT companies inflate it, and how two numbers — EBITDA and headcount — let you understand a business in about a minute.

What EBITDA is and how it is calculated

EBITDA stands for Earnings Before Interest, Taxes, Depreciation and Amortization — profit before loan interest, taxes, depreciation and amortisation.

There are two ways to calculate it, and both should give the same result.

Bottom-up, from net profit:

Net profit
+ Income tax
+ Loan interest
+ Depreciation and amortisation
= EBITDA

Top-down, from revenue:

Revenue
– Cost of sales (COGS)
= Gross profit
– Operating expenses (excluding depreciation)
= EBITDA

An example for an IT services company:

Line item	Annual amount
Revenue	$12,400,000
COGS: salaries of engineers on projects	$7,100,000
COGS: infrastructure, client-specific licences	$380,000
Gross profit	$4,920,000
Gross margin	40%
Salaries: admin, sales, HR, management	$2,340,000
Rent, office, internal IT	$410,000
Marketing and sales (excluding salaries)	$290,000
Legal, accounting, other	$180,000
EBITDA	$1,700,000
EBITDA margin	13.7%
Depreciation and amortisation	$210,000
EBIT	$1,490,000
Loan interest	$120,000
Taxes	$190,000
Net profit	$1,180,000

EBITDA is $1.7M, net profit is $1.18M. The $520,000 gap is depreciation, interest and taxes — three items that have nothing to do with how well the operating side of the business runs.

Why EBITDA and not net profit

Net profit is the final number, so it should be the more important one. But for comparing companies it is almost useless, and here is why.

Taxes depend on jurisdiction and structure. A Ukrainian IT company built on sole-trader contractors, a Polish spółka z o.o., a Cypriot holding structure and a US C-Corp pay fundamentally different taxes at identical operating efficiency. Comparing them on net profit means comparing tax regimes, not businesses.

Interest depends on how the company is financed. Two operationally identical companies: one runs on its own cash, the other took a $3M loan to expand. The second one's net profit is lower by the interest. That's a difference in financial structure, not in business quality. A buyer arriving with their own money to repay the debt does not care.

Depreciation depends on accounting policy and purchase history. A company that bought equipment and an office three years ago carries heavy depreciation. A company that rents everything and runs on employee laptops carries almost none. In IT this distorts the picture especially badly, because fixed assets are secondary here, and amortisation of intangibles depends on whether development was capitalised.

By stripping out those three items, EBITDA leaves what is genuinely about operations: how much the company earns from its core activity, regardless of where it is registered, how it is financed, and what it bought three years ago.

That is why it became the standard for comparison. Not because it is more accurate, but because it is comparable.

What EBITDA shows and what it does not

It shows operating efficiency — how far revenue exceeds the cost of producing it. In IT services this mainly answers one question: is the gap between the client rate and the cost of an engineer big enough to cover all non-billable costs?

It does not show three important things.

Capital requirements. Depreciation is excluded, but money for hardware, licences and infrastructure is really spent. For IT services this is not critical — capex is small. For product companies that capitalise development, the difference between EBITDA and real cash flow can be several times over.

Working capital. EBITDA can be $2M with no cash in the bank, because $2.5M is sitting in receivables. In outsourcing with net-45 or net-60 terms this is standard. A company with good EBITDA and poor receivables management can hit a liquidity gap every quarter.

The cost of growth. A company growing 40% a year spends far more on recruiting, onboarding and sales than one standing still. The flat company will show higher EBITDA even though it is the worse business.

EBITDA in valuation: how multiples work

The most common way to value an IT company at exit is to multiply EBITDA by a multiple.

Enterprise value (EV) = EBITDA × Multiple

Indicative multiples for IT services businesses:

Company EBITDA	Typical multiple
under $1M	2–4x
$1–3M	4–6x
$3–10M	5–8x
over $10M	7–12x

Why the spread is so wide, and what determines the actual number.

Client concentration. A company with $2M EBITDA where one client provides 70% of revenue is valued far below a company with the same EBITDA and twenty clients. Losing a single contract can wipe out half the business.

Contract type. Long-term contracts with automatic renewal, dedicated teams and subscription models are valued higher than project work started from scratch every three months.

Owner dependency. If sales rest on the owner's personal relationships, the business sags after they leave. The multiple drops substantially.

Growth rate. A company growing 35% a year earns a higher multiple than a flat company with the same EBITDA. The buyer pays for the future, not the past.

Tech stack and domain. Teams in narrow high-margin domains — fintech, healthtech, ML — are valued above general web development.

Attrition. 35% annual turnover means the buyer inherits vacancies, not a team. That hits the multiple directly.

Why product IT companies are often not valued on EBITDA

For SaaS and product companies EBITDA is frequently not used as a valuation base at all, because it is structurally understated or negative there.

A product company in growth mode invests more in development and marketing than it earns today, betting on future revenue from the customers it acquires. EBITDA of –$3M at $12M ARR growing 60% a year is a normal picture, not a warning sign.

So valuation runs off revenue instead:

EV = ARR × Multiple (typically 3–10x)

The multiple depends on growth, retention and unit economics. The baseline check for a healthy balance between growth and profitability is the Rule of 40:

Revenue growth % + EBITDA margin % ≥ 40

A company growing 60% with a –15% EBITDA margin scores 45 — acceptable. A company growing 10% with a 12% margin scores 22 — weak, and the question becomes why profitability is also low at such slow growth.

Practical takeaway: if someone quotes a product company's EBITDA without ARR and growth rate, the number alone means almost nothing.

How to tell when EBITDA is inflated

IT has several specific ways of making EBITDA look better than it is. Here is what to look at.

Capitalising development costs

The most common technique in product companies. Salaries of developers building the product are capitalised as an intangible asset and amortised, instead of hitting operating expenses.

The effect is double: the costs leave the operating section, and the amortisation that replaces them is excluded from EBITDA by definition. The company spent $2M on development and EBITDA contains none of it.

How to check: compare capitalised development on the balance sheet with personnel costs in the P&L. If the company has 40 developers but only 15 developers' worth of salary in operating expenses, the rest is capitalised. Ask how the split was made.

Adjusted EBITDA with a long list of add-backs

"Adjusted EBITDA" is a number from which management has additionally excluded whatever it considers non-typical. Sometimes justified, often not.

Typical IT add-backs and what's wrong with them:

Stock-based compensation. In option-granting companies this can be 10–20% of payroll. Formally non-cash, so it gets added back. But it is a real cost of retaining people — without it they would demand more cash salary.

"One-off" recruiting costs. If the company is growing, recruiting is not an event but a permanent line item. Excluding it overstates EBITDA by exactly the cost of growth.

Team relocation costs. In 2022–2023 many Ukrainian companies excluded these as force majeure. The question is whether they actually ended, or became the new normal.

Losses on failed projects or clients. "We lost $400,000 on a project that went wrong, it's atypical." If such projects happen every year, it is typical.

The check: look at add-backs across three years. If "one-off" costs appear annually, they are not one-off — they are operating reality.

Owner salary missing from the P&L

In companies under 100 people, owners often work actively: selling, running key accounts, making technical calls. Yet they may take no formal salary, pulling money out as dividends.

As a result the P&L carries no cost for a role that will have to be handed to someone after the sale. The buyer hires a CTO or Head of Sales at $120,000–180,000 a year, and EBITDA drops by that amount immediately.

How to check: ask what functions the owner performs and what it would cost to hire people into those roles. Subtract that from EBITDA.

A contractor structure that hides real personnel cost

In Ukrainian IT most engineers are engaged as sole traders. It is a legal and widespread practice, but it creates a distortion when comparing with Western companies where the same people would be employees with full benefits and employer taxes.

A buyer from another jurisdiction may raise questions about reclassification risk and about what the cost base would look like under a different employment model. It does not make EBITDA wrong, but it can affect the multiple.

Revenue recognition on long contracts

If the company works on fixed-price contracts with milestone recognition, there is room to manoeuvre: recognise more revenue in the current period while shifting costs out. The year looks better; the next one suffers.

How to check: compare the revenue trend with cash collections. If revenue grows but cash flow does not, and receivables grow faster than revenue, there is a question to ask.

The one-minute read: EBITDA and headcount

Now the most practical part. You've been given two numbers: EBITDA and headcount. Here's what you can work out in a minute.

Baseline IT market benchmarks

Annual revenue per employee:

Business type	Revenue per person
Staff augmentation, Eastern Europe	$50,000–75,000
Outsourcing with project management	$70,000–110,000
Custom product development, narrow domains	$100,000–150,000
SaaS, mature	$130,000–250,000
SaaS, top performers	$250,000+

EBITDA margin:

Business type	Typical margin
Staff augmentation	12–20%
Outsourcing	15–25%
SaaS, growth stage	–30% to +10%
SaaS, mature	20–35%

Annual EBITDA per employee: in healthy outsourcing, $10,000–20,000. Below $7,000 is a thin margin. Above $30,000 means either a very expensive domain or a number that needs checking.

The one-minute algorithm

Step 1. EBITDA per person.

You're told: "EBITDA $4M, 200 people." $4,000,000 / 200 = $20,000 per person.

Step 2. Estimate payroll.

The full cost of an engineer in Ukraine or Poland including all payments is roughly $45,000–65,000 a year. Take the middle, $55,000. 200 × $55,000 = $11M of payroll.

Step 3. Reconstruct revenue.

In IT services payroll is usually 55–65% of revenue. Take 60%: $11M / 0.60 ≈ $18.3M of revenue.

Step 4. Check the numbers hang together.

Revenue per person: $18.3M / 200 = $91,500 — consistent with outsourcing that includes project management. EBITDA margin: $4M / $18.3M = 21.8% — the top of the normal range for outsourcing, but realistic.

One-minute conclusion: most likely a mid-sized outsourcing company with a good but not anomalous margin and revenue around $18–20M. At a 5–7x multiple, indicative value is $20–28M.

When the numbers don't add up

The same algorithm quickly catches unrealistic claims.

Example: "EBITDA $6M, 150 people."

EBITDA per person: $40,000 — already above the norm for a services business. Payroll: 150 × $55,000 = $8.25M. If that is 60% of revenue, revenue ≈ $13.75M. After salaries, $5.5M remains, and out of that you still need office, admin, marketing and licences. $6M of EBITDA is physically impossible from that.

So one of three things: it is not a services company but a product company with entirely different economics per person; a significant share of the work is done by contractors outside the count and "150 people" is not the whole team; or EBITDA has been adjusted with add-backs.

None of those means deception, but all three mean you need follow-up questions.

Example: "EBITDA $800,000, 300 people."

EBITDA per person: $2,700 — very low. Payroll: 300 × $50,000 = $15M, revenue ≈ $23–25M, EBITDA margin ≈ 3.3%.

Conclusion: a large company on a very thin margin. Likely causes are low rates, a large share of bench, a bloated admin function, or loss-making fixed-price contracts. Not necessarily a bad company, but it will be valued cheaply on an EBITDA multiple — and question number one is what's happening with rates and utilisation.

Two follow-up questions worth asking immediately

What is utilisation? The share of billable hours in engineers' total working time. In healthy outsourcing, 75–85%. If it's 60%, a third of the team is on the bench, and that is the main reason for the low margin.

Is that EBITDA or adjusted EBITDA, and what exactly was adjusted? In IT companies the gap between those two numbers is easily 20–40%.

Summary

EBITDA became the standard not because it is the most accurate measure of profitability, but because it removes three things that make companies incomparable: tax jurisdiction, financing structure and the history of capital purchases. What remains is genuinely about operations.

In IT services it works well and converts directly into valuation through a multiple. In growth-stage product companies it is structurally understated, and you should look at ARR, growth rate and the Rule of 40 instead.

In IT it is inflated mainly in four ways: capitalising development, add-backs in the "adjusted" version, a missing owner salary in the P&L, and manoeuvres with revenue recognition on long contracts. All four can be tested with two or three questions.

And the most useful part in practice: from EBITDA and headcount you can reconstruct payroll, revenue and margin in a minute. If those three numbers don't hang together, the next conversation should not be about valuation — it should be about how the numbers were calculated.