TL;DR: Business valuation in India is the work of fixing a defensible value range for a company or its shares, for a named purpose such as a fundraise, a sale or a regulatory filing. The method produces the range, while the regulation behind the transaction sets the floor and decides who may sign the report. Macro growth sits under every forecast, and World Bank data[1] reports annual GDP growth of 7.6% for India in 2025, a national figure no single company should borrow as its own. A CFO should fix the purpose, the regime, the signer and the valuation mechanism before anyone opens a spreadsheet.
Business valuation in India is a defensible range, not a single number
Business valuation is the estimate of what a company or its equity is worth under a stated standard of value, on a stated date and for a stated purpose. Change any one of those and the number moves.
- Standard of value
- the definition of “value” a report adopts, such as fair market value for a regulatory filing or investment value to one particular buyer in a sale.
In Astravise’s experience, three purposes cover most of what Indian finance teams face. A primary fundraise prices new shares and sets dilution, while a sale prices control at what the business is worth in the buyer’s hands. Compliance valuation, the third purpose, proves that a price clears a statutory floor, and its real audience is the regulator. One company can carry three different numbers in the same quarter, and each can be correct; the disputes start when they are mixed.
Three methods produce the number, and each answers a different question
Valuers in India work from three families of method: income, market and asset. A competent report usually runs at least two and explains why one carries the weight.
The income approach is where most of the argument happens, because two assumptions do most of the work: the discount rate and the terminal growth rate. A terminal growth rate pinned to national ambition rather than outturn is the most common error we see in founder-built models. The Hoover Institution’s 2026 Survey of India[2] describes India as aiming for developed-nation status by 2047, striving for growth of about 8 percent. That is a stated aspiration, set against an actual 2025 outturn of 7.6% on World Bank figures, and neither number is a company’s perpetual growth rate.
| Method | What it measures | Where it fits best | Where it breaks |
|---|---|---|---|
| Income approach (discounted cash flow) | Present value of forecast free cash flows at a risk-adjusted discount rate | Businesses with a credible forecast: growth-stage companies, captive centres, steady services firms | Forecasts written backwards from a target price; a terminal value that carries most of the result |
| Market approach (comparable companies, transactions and prior rounds) | Value implied by revenue or earnings multiples of similar businesses | Fundraises, and sectors with listed peers or recent deals | Thin comparables in niche Indian sectors; a prior round priced on terms that no longer apply |
| Asset approach (net asset value) | Fair value of assets less liabilities | Holding companies, asset-heavy or loss-making businesses | Services and software firms, where value sits in people and contracts rather than the balance sheet |
A valuation is only as defensible as its weakest assumption, and in a discounted cash flow that is almost always the terminal value.
Which regulation decides the floor, the method and who signs?
The regime behind a transaction matters more than the method, because a regime can rule a price out whatever the model says. Foreign capital makes this concrete: World Bank data[3] shows net FDI inflows equal to 1.0% of India’s GDP in 2025, and a foreign purchase of unlisted Indian shares from a resident passes through a pricing rule.
Astravise reads that rule, Rule 21(2)(b) of the Foreign Exchange Management (Non-Debt Instruments) Rules, 2019, as a floor rather than a formula. A resident-to-non-resident transfer cannot be priced below fair market value, worked out on an internationally accepted pricing methodology at arm’s length. The certificate must come from a chartered accountant, a SEBI-registered Category I Merchant Banker or a practising cost accountant. Under the explanation to the rule, a non-resident cannot be guaranteed an assured exit price, so pre-agreed, floor and IRR-linked exit prices do not survive.
Timing is the constraint many teams miss. In Astravise’s practice, authorised dealer banks want a valuation dated within roughly six months of the transfer, and Form FC-TRS is due within 60 days of transfer or receipt of consideration, whichever is earlier. A report commissioned at term sheet and used a year later is a second valuation waiting to happen.
Company-law valuations, including those that call for a registered valuer, run on a separate track whose signer and report requirements turn on the specific corporate action. We have not reproduced that rule text here. Map it with counsel before engaging anyone, because the signer list narrows the field of valuers you can use.
What moves the number in a fundraise
Investors price a round on four things: the quality of reported earnings, the credibility of the forecast, the size of the reachable market and the governance behind the numbers.
Market size is where evidence discipline separates a defensible deck from a hopeful one. A consumer-internet business might cite the Kepios analysis in DataReportal’s Digital 2026: India report[4], which puts India at 1.03 billion internet users in October 2025, equal to 70.0 percent of the population. That figure is a modelled estimate dated October 2025, not 2026, and it counts users rather than buyers. An investor who sees the caveat handled honestly trusts the rest of the model more.
Quality of earnings is a reconciliation exercise, not an opinion. A diligence team will strip out one-off income, normalise founder salaries and related-party charges, and test revenue recognition against contracts. Any adjustment they make that you did not make first gets priced at its worst case.
Governance shows up as a discount rather than a line item. Late monthly closes, a board that meets without papers and an audit trail that depends on one person all widen the haircut an investor applies to the forecast. “We do not publish numbers we cannot attribute” is a house line at Astravise, and it is the right test for every figure in a data room.
How a sale or a BOT transfer changes the valuation
A sale is priced on control, which changes both the method and the incentives. The buyer values synergies and the right to change strategy, the seller wants to be paid for them, and FEMA only asks that the price clears fair market value.
Currency adds a layer when the buyer is foreign. A US-listed parent values an Indian business in dollars while the seller’s model runs in rupees, and the gap between the two discount rates reflects expected inflation and currency movement. World Bank data[5] shows consumer price inflation of 2.4% for India in 2025, and a buyer will use a low reading like that to argue for a thinner rupee premium. A seller should ask whether one year’s figure is a fair basis for a five-year forecast.
Build-operate-transfer (BOT) is a model in which a provider builds and runs a captive centre, such as a global capability centre, and transfers it to the client after an agreed period. Astravise works both sides of these deals, and our position is that a BOT contract is an M&A transaction signed two to four years before it closes. The contract should therefore fix a valuation mechanism, not a price: on a three-year BOT, a signing-date valuation is worthless paper by the time it is needed.
The route moves the number too. A share purchase carries the company’s history with it, while a slump sale transfers the business as a going concern; under the Income-tax Act, 2025, which governs from 1 April 2026, Astravise reads Section 77 as computing net worth at book value with revaluation ignored. People liabilities can dominate either route. If employee transfer terms are drafted carelessly, as Astravise has put it, “the acquirer is not buying a GCC. It is funding a statutory severance event across the entire workforce.”
AI-built valuation models need a human check on every input
Finance teams increasingly draft models and memos with AI assistants, and the published research on those tools argues for checking every computed figure. A June 2025 Apple Machine Learning Research paper[6] found that reasoning models showed limitations in exact computation, failing to use explicit algorithms and reasoning inconsistently across puzzles. Valuation is a chain of exact computations, so one silent error in a discount factor carries into every year that follows.
The model vendors concede the same weakness in their own systems. Neither the Apple paper nor any vendor study measures valuation work specifically, and no published figure exists for how often AI-drafted financial models contain errors. What the developers do say is telling: OpenAI acknowledged in June 2024[7] that as its models become more accurate their mistakes become more subtle and harder for human reviewers to spot, calling this a fundamental limitation of RLHF.
Our rule is that an AI assistant can draft structure and narrative, and a named person signs off every number, formula and source.
The six steps Astravise runs before a valuer is engaged
Astravise has run finance, governance and people workstreams through complete builds, including a Hyderabad global capability centre set up for a US-listed logistics company. A valuer works from what the company hands over, so this sequence comes first.
-
Name the purpose and the standard of value. Write down whether the valuation supports a fundraise, a sale, a transfer or a filing, and which standard applies. That one line settles half the later arguments.
-
Map every regime the transaction touches. FEMA pricing, income-tax computation and company-law requirements can all apply to one deal, each with its own floor, signer and timing. List them before choosing a valuer.
-
Close the books and reconcile the earnings. Normalise one-off items, related-party charges and founder costs yourself, and document each adjustment. Unreconciled numbers are the most expensive input in any valuation.
-
Rebuild the forecast from dated evidence. Every market and growth assumption should carry its source and the period it covers. DataReportal’s Digital 2026: India report[8], for instance, was published at the end of 2025 and reflects data available in October 2025, whatever its title says.
-
Agree the mechanism before the number. In a sale or BOT, the contract should name the method, the valuation date relative to closing and expert determination for disputes. A fixed price signed years ahead fails either the regulator’s floor or the buyer’s patience.
-
Stress-test the two assumptions that carry the value. Run the discount rate and terminal growth rate across a realistic range and show the board the spread, not a point estimate. If a one-point move changes the decision, the board should know before the counterparty does.
Frequently asked questions about business valuation in India
Frequently asked questions
- Which valuation method is best for an Indian startup raising money?
- For an Indian startup raising money, the market approach usually carries the most weight, built on comparable rounds and listed peers, with a discounted cash flow as a cross-check. Where a foreign investor buys shares from a resident, the startup’s price must also clear fair market value under FEMA pricing rules, which sets a floor regardless of method.
- Does India’s GDP growth justify a higher valuation multiple?
- India’s GDP growth supports investor interest but does not justify any one company’s multiple. World Bank data reports annual GDP growth of 7.6% for India in 2025, while the figure of about 8 percent often quoted is a stated aspiration. A company’s valuation multiple rests on its own earnings quality, forecast and governance.
- Who can sign a FEMA valuation for a share transfer to a foreign investor?
- Astravise reads Rule 21 of the Foreign Exchange Management (Non-Debt Instruments) Rules, 2019 as allowing a chartered accountant, a SEBI-registered Category I Merchant Banker or a practising cost accountant to certify fair market value on a resident-to-non-resident transfer of unlisted shares. Company-law valuations that call for a registered valuer follow separate requirements, which counsel should confirm.
- How old can a valuation report be when the share transfer happens?
- In Astravise’s experience, authorised dealer banks generally want a valuation dated within roughly six months of the share transfer. A valuation report commissioned at term sheet stage can therefore expire before closing, especially on deals with long approval timelines, so finance teams should budget for a refreshed valuation report in the transaction calendar.
- Can a foreign investor be promised a fixed exit price in India?
- A foreign investor cannot be guaranteed an assured exit price on unlisted Indian shares, as Astravise reads the explanation to Rule 21 of the FEMA non-debt instrument rules. Pre-agreed, floor and IRR-linked exit prices are treated as impermissible, so a shareholder agreement should set a valuation mechanism for the foreign investor’s exit rather than a number.
- Can I use ChatGPT to build a company valuation model?
- ChatGPT can draft the structure of a valuation model, but every number in the model needs a human check. A short communication in the Brazilian Journal of Science, discussing the GPT-3.5-based ChatGPT, noted that apparently correct answers may be factually inaccurate. Astravise’s practice is for a named finance professional to sign off every formula and source in the valuation model.
- Is a slump sale valued differently from a share sale?
- A slump sale is valued and taxed differently from a share sale because a slump sale transfers the business as a going concern rather than the company’s shares. Under the Income-tax Act, 2025, Astravise reads Section 77 as computing net worth at book value with revaluation ignored, which changes the seller’s after-tax outcome.
Sources
- India | Data World Bank (Projects & Operations summary on the India country page) ↩
- The Hoover Annual Survey of India 2026 | Hoover Institution The Hoover Annual Survey of India 2026 Hoover Institution (Survey of India 2026; chapter 4 by Arzan Tarapore) ↩
- India | Data World Bank (Projects & Operations summary on the India country page) ↩
- Digital 2026: India — DataReportal – Global Digital Insights DataReportal (Digital 2026: India) ↩
- India | Data World Bank (Projects & Operations summary on the India country page) ↩
- The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexity – Apple Machine Learning Research Apple Machine Learning Research ↩
- Finding GPT-4’s mistakes with GPT-4 | OpenAI OpenAI ↩
- Digital 2026: India — DataReportal – Global Digital Insights DataReportal (Digital 2026: India) ↩