Understanding HMRC EMI Valuations: A Founder's Guide

An EMI valuation is not a generic company valuation; it is a tax valuation, governed by HMRC Shares and Assets Valuation (SVM) practice and underpinned by s.431 of the Income Tax (Earnings and Pensions) Act 2003. The valuation establishes two numbers: the Actual Market Value (AMV) of a share carrying the EMI restrictions, and the Unrestricted Market Value (UMV), the same share if all restrictions fell away. Both numbers matter and HMRC expects to see both.

The practical consequence is that the number you agreed with an investor is almost never the number HMRC will agree for an option grant, and that is not a contradiction. It is the point.

Why your EMI number is not your funding round number

The single most common misunderstanding is that a company has one value on a given day. It does not. It has a value for each question being asked of it, and an EMI grant asks a narrower question than a funding round does.

A priced round values a preference share: one carrying a liquidation preference, often an anti-dilution ratchet, information rights, and a board seat. An EMI option is over an ordinary share carrying none of that. It sits behind the preference stack, it is a minority holding, it cannot be sold, and its holder cannot influence an exit.

So the same company on the same afternoon can support a £20m post-money round and an ordinary share AMV that implies a materially lower figure. Founders often assume the gap is a negotiating position. It is not; it is the difference between two genuinely different instruments, and it is defensible precisely because it can be explained that way.

Where founders get into difficulty is asserting the gap rather than evidencing it. HMRC does not object to a discount. It objects to a discount that appears without a derivation.

AMV and UMV: why both matter

AMV is what an arm's length buyer would pay for the restricted share. UMV is what the same buyer would pay if no marketability or transfer restrictions applied. The gap between the two is the marketability discount: commonly in the region of 10 to 30 per cent for an unlisted SaaS share, depending on cap-table dynamics, the rights attached to the share class, and the realistic exit horizon. The range is wide because the answer is fact-specific rather than conventional. There is no standard discount, and the higher the figure, the more the evidence has to carry it.

The two numbers do different jobs, which is why neither can be dropped.

AMV sets the exercise price. Grant at or above AMV and there is no income tax charge when the option is exercised, which is the whole benefit of the scheme. Grant below it and the discount is taxable as employment income at exercise, with National Insurance alongside it where the shares are readily convertible assets.

UMV governs the limits. The individual cap of £250,000 of unexercised options, and the company-wide limit, are both measured on unrestricted value. A company that quietly uses AMV for the headroom calculation can find it has granted more than it was entitled to, and the excess is not an EMI option at all.

That company-wide limit changed this year. It rose from £3m to £6m for options granted on or after 6 April 2026, while grants made before that date remain tested against the old figure. A company that last reviewed its headroom in 2025 may therefore have considerably more room than its own records suggest, and one carrying grants either side of the date is testing them against two different ceilings.

The s.431 election is the connective tissue between the two. Signed within 14 days of acquisition, it elects to treat the shares as though the restrictions did not exist for income tax purposes, moving future growth into the capital gains regime. It is a short form, it is routinely forgotten, and forgetting it converts what should have been a capital gain into employment income years later.

The discount stack, and how to defend it

The marketability discount is where most EMI valuations are won or lost, because it is the number with the least external evidence behind it and the most influence on the answer.

A defensible stack is built, not chosen. It typically runs in three layers:

• Lack of control. The optionholder cannot direct a sale, block one, or influence dividend policy. This is evidenced from the articles and the shareholders' agreement, not assumed from the size of the stake.
• Lack of marketability. There is no market. Transfer is restricted, pre-emption applies, and any realistic exit sits years out. The length of that horizon does real work here.
• The preference stack. Where preferred shares sit ahead of the ordinary in a waterfall, the ordinary share absorbs the shortfall first at lower exit values. This is modelled, not estimated.

The discipline that separates an accepted valuation from a challenged one is showing each layer separately with its own reasoning. A single blended percentage, however reasonable the total, invites the question of how it was reached, and having to answer that question after submission is a worse position than answering it in the report.

How the three sectors differ

The framework is identical across sectors. What changes is where the scrutiny falls.

SaaS

For a SaaS business, the forecast carries the valuation, and HMRC knows it. Recurring revenue makes a multi-year projection genuinely meaningful, which is a strength, but it also means the assumptions are the argument. Net revenue retention, churn and gross margin progression each need a stated basis and a sensitivity, because a valuation resting on 115 per cent NRR that is not evidenced from cohort data is resting on an assertion.

SaaS also tends to show the widest gap between the last round and AMV, because SaaS cap tables carry the heaviest preference stacks. That gap is defensible, and it is exactly the one that must be modelled rather than asserted.

Marketplace

For a marketplace, the first question is what is being valued at all. Gross merchandise value is not revenue, and a valuation anchored on GMV rather than net revenue after take rate will attract scrutiny before anything else in the report is read.

The second question is take-rate durability. A take rate raised recently to defend revenue is not the same evidence as one held through a cycle, and the difference reaches the terminal value.

eCommerce

For an eCommerce or D2C business, the asset base does more work than in either of the others. Inventory, plant and brand carry real balance-sheet value, so an asset-based floor is a meaningful cross-check rather than a formality, and its absence is noticeable.

Seasonality also makes the valuation date unusually consequential. A business valued on its position in November is describing a different company from the same business in February, and the report has to say which one it valued and why that date was chosen.

What HMRC SVM actually reviews

Three things, in order: the methodology disclosure (have you triangulated DCF, comparable companies and precedent transactions?), the comparable-company set (are they current, sector-aligned, and selection-justified?), and the discount stack (is the marketability discount supported by evidence rather than asserted?). A valuation that ticks all three is materially less likely to be rejected.

The underlying principle is that SVM reviews reasoning rather than arithmetic. Two valuers can reach different figures from the same accounts and both be accepted, provided each shows its route. The submission that fails is the one that presents a conclusion without the path to it. Our triangulated approach exists for this reason as much as for commercial work.

The five most common rejection reasons

• Stale comparables: public companies whose multiples have moved more than 20 per cent since the valuation date.
• Asserted but undefended marketability discount.
• A DCF terminal value that materially diverges from the implied multiple at exit.
• No churn or retention sensitivity in the forecast for a SaaS business.
• Single-method valuation with no triangulation.

Four of the five are failures of evidence rather than of judgement. That is worth noticing, because it means most rejections are avoidable at the drafting stage rather than arguable afterwards.

Timing, and the deadlines that actually bind

From a clean data room we prepare an EMI valuation in 10 working days, plus the HMRC SVM clock, which can run from 4 to 12 weeks depending on workload.

Three dates then matter, and they are frequently discovered in the wrong order:

• The agreed valuation has a shelf life. HMRC agreement is given for a limited window, commonly 90 days, and it lapses if something material changes in the meantime. A funding round, a large contract or a significant acquisition inside that window can invalidate an agreement you are relying on.
• Grants must be notified to HMRC. Miss the notification deadline and the options are not EMI options, whatever the paperwork says. The relief is lost outright, not deferred.
• Disqualifying events start a short clock. Losing independence, breaching the gross-asset or employee limits, or a change in the qualifying trade gives holders a narrow window to exercise before the tax treatment changes.

Work backwards from the intended grant date, not forwards from today. Founders routinely start this in the week of the board meeting at which they intend to grant, and by then the sequence cannot be compressed.

ValuCap delivers signed EMI valuations from £899, scoped inside a signed business valuation rather than issued beside one, because the basis and the date are agreed with HMRC and have to be evidenced the same way.

Who does what: your accountant and your valuer

Engaging your accountant on the tax side and your valuer on the methodology side is the cleanest split, and the line between them is easy to state even though it is often blurred in practice.

Your accountant advises on the tax. Whether the company and the individuals qualify, how a grant sits alongside the rest of the remuneration structure, the notification itself, and the s.431 election.

Your valuer establishes and evidences the numbers. AMV, UMV, the methodology, the discount stack, and the submission that carries them.

We do not provide tax advice; your accountant does. We work alongside them on the valuation submission. The reason the split is worth stating is that HMRC tests the two sides differently: a tax position is either right or wrong, whereas a valuation is defensible or it is not, and defensibility is entirely a function of the evidence attached to it.

What it costs to get wrong

The scheme is generous, and the failure modes are correspondingly expensive.

An option granted below AMV creates an income tax charge on the discount at exercise. A missed notification removes the relief entirely, so gains that should have been taxed at capital gains rates are taxed as employment income, with employer's National Insurance where the shares are readily convertible. A miscalculated UMV can put grants outside the individual or company limits, which is discovered during due diligence on the exit that the options existed to reward.

None of these surface at grant. They surface years later, in a data room, at the moment the company can least absorb them, and the people affected are the employees the scheme was designed to reward.

That is the real argument for evidencing a valuation properly at the outset. The cost of doing it well is known and modest. The cost of doing it loosely is unknown, deferred, and paid by someone else.