The Tech Deal series - 2. The LOI: what to lock down before the SPA

Letter of intent (LOI): the decisions that are fixed before the lawyers start drafting

Previously in this series. Post #1 argued that in a tech deal most of the value sits in assets you cannot touch, may not fully own and sometimes cannot transfer. It introduced the four companies this series follows, among them Nevelo, a Dutch B2B SaaS scale-up with €12 million of ARR. It also set out the four arenas where those risks surface. The LOI is where parties first put commercial terms on paper, and where risks and benefits are allocated or left open.

By the time one of the parties’ lawyers starts drafting the share purchase agreement (SPA), most of the deal economics are already fixed, in a document everyone kept calling non-binding.

The corporate M&A lawyer usually arrives after the letter of intent has been signed. A price range has been agreed, exclusivity has been granted, and the mandate is to turn the key deal terms into a long-form agreement. From both a negotiation and a legal perspective, drafting the LOI is a series of strategic decisions: which subjects do you settle and which do you deliberately leave alone, and how you do settle them: do you do so in detail and exhaustively, or on main lines and on general principles?

This post is about those decisions. The relevant question throughout is this: if I do not settle something now, or do not settle it concretely, what does that do to my own negotiating position, and to my counterparty’s, later in the process?

Each is a decision about risk and benefit taken under uncertainty, and it is worth taking deliberately rather than waving through to reach the next phase.

What the LOI is for, and the tension inside it

An LOI has two purposes: it records what the parties have agreed on the substance of the deal, and it sets the rules of the process that follows. It does the same job in a software deal as in a deal for a logistics business, and the same goes for a term sheet, heads of agreement or memorandum of understanding. The name on the document is irrelevant in and of itself.

What it leaves both parties with is a contractual basis to work from while diligence runs and the SPA is negotiated, common ground on the deal principles, and enough trust to carry the rest of the process. Without that, parties are generally reluctant to spend additional time and cost to negotiate a long-form document before diligence has even started. Settle everything in it, though, and that is what it becomes: a de facto SPA negotiated before diligence has started.

Which side benefits if a key point is left open

A point left open does not sit neutrally between the parties, and it rarely falls the same way for both.

The seller’s requirements are known on the day the LOI is drafted: deal certainty, the money in the bank, and as little liability as it can negotiate. Diligence will not change them. Settling the commercial deal in full therefore costs a seller little and improves its position afterwards, because every point the purchaser raises from then on arrives as a request, to be granted in exchange for something else.

Where the seller rolls over into the purchaser’s structure, this negotiation dynamic softens. A seller who is staying in needs things the purchaser controls, from the participation terms and the leaver provisions to minority protection in the new structure, and those usually only crystallise after the LOI is signed. That gives the purchaser something to trade for what diligence later produces.

When negotiating the LOI, the purchaser’s concrete requirements are not visible yet. What it will need protection from is produced by the diligence that has not started, which is why the looser LOI, the one that agrees principles rather than spelling out the commercial deal in detail, often serves a purchaser better. That only works if the purchaser also colours them in, which is why a purchaser does well to include in the LOI that it shall hold the pen on the first draft of the SPA.

Price and risk allocation as communicating vessels

A deal is essentially a price and a risk allocation, and the two are often communicating vessels. A higher price warrants a heavier protection package; a lighter package should be reflected in what the purchaser pays. Negotiated together, each side can concede on one and recover on the other.

An LOI that settles the price and leaves the protection package for later takes one vessel out of the negotiation. The trade is no longer available, and what follows is closer to a zero-sum fight: the purchaser wants as much protection as it can get at a price already agreed, the seller wants to give as little as possible, and neither has much left to offer in return. The timing compounds it. The purchaser learns what protection it needs after the price has been settled, so every risk point it raises from then on is a price cut in disguise, which is exactly why sellers tend to resist them.

Non-binding means you can walk away, not that you can renegotiate

One more piece of the frame, because it is routinely misread by less experienced parties. An LOI headed non-binding does broadly what the label says: as a rule, under Dutch law, you have the freedom to walk away from the deal, subject only to narrow exceptions. What it does not give you is the freedom to reopen a point the other side treats as settled. Walking away costs you the deal; renegotiating costs you a concession on another point, and sometimes the relationship.

The price side: how to lock it down before diligence

Most LOIs fix the price and qualify it: agreed, subject to confirmatory due diligence. That phrasing implies the principles behind the number are already settled and all that remains is to verify them against the underlying figures. Usually, those principles are not settled. Often they are not spelled out at all, or spelled out to a degree that still leaves room to negotiate away a material part of the agreed price later on.

ARR valuations: the metric is the price

In software deals the enterprise value of the target is usually expressed as a multiple of annual recurring revenue (ARR). In practice the parties agree both the multiple and the ARR figure, subject to confirmatory due diligence. In most cases, the multiple then stays where it is. The ARR definition is what remains open, which makes it the thing that moves the price between the LOI and signing.

Nevelo again. Its AI assistant is priced per use rather than per subscription and brings in €750,000 a year, and the seller’s €12 million revenue headline number includes it. The purchaser will want it out of the ARR line, because nothing commits a customer to next month’s volume. The seller will resist, because the revenue is real and growing and it is part of what is being sold. Both positions are reasonable, which is why the answer is a choice rather than a finding: the module sits inside the ARR figure to which the multiple applies, or inside it at a discount, or outside it and paid for separately.

Settling that now is the same negotiation with less entrenchment on both sides. Three months in, the €750,000 has not moved but the parties have: the purchaser has paid advisers, run diligence and taken the deal to an investment committee, the seller has had its business picked over and its team distracted, the price is settled so there is nothing left to trade, and neither side is keen to walk. The merits are unchanged. What has changed is what it costs each side to lose, and both of them know it.

A handful of questions drive the ARR figure, and they are answerable before the LOI is signed: where the line falls between what recurs and what does not, how contracted but not yet live revenue is treated, what happens to customers on notice, at what date it is measured, and whose numbers govern. Post #4 of The Tech Deal, ARR valuation: negotiating the definition, takes that anatomy apart question by question. The obvious retort is that a purchaser at this stage has not seen the financials and cannot negotiate a definition in the dark, but a definition is a rule, not a number: the parties fix the ruler now and measure against the financials later, in diligence. And a seller who resists a particular rule has told the purchaser something useful about how it came to its ARR number.

If it turns out during diligence that the ARR figure is materially below what the multiple was built on, the multiple itself is back in play. A purchaser paying a premium for durable, sticky revenue is entitled to say that the premium was for revenue that turned out not to be there. Worth making that explicit in the LOI, because the alternative is a seller arguing that only the ARR definition, and not the multiple, was subject to confirmatory due diligence.

Deferred revenue is what catches the first-time software seller

The agreed enterprise value is almost never what the seller receives. It is typically stated cash-free and debt-free, assuming a normal level of net working capital: for every euro in the target’s bank account the price goes up by the same amount, and for every euro of debt it goes down. The grey areas in between leave plenty of room for discussion.

Deferred revenue is the one that catches the software seller doing this for the first time. A private equity purchaser with a standard equity bridge finds it every time; the seller who has never been told that cash in the bank arrives with an obligation attached does not. It is the item I saw catch a seller from the other side of the table, as head of legal at a listed IT company.

Take Nevelo. Its customers pay annually in advance, so there is real money in the bank for services still to be delivered. About a third of its €12 million of ARR, call it €4 million, sits on the balance sheet as deferred revenue (cash collected for work not yet performed). The LOI says cash-free, debt-free and everyone nods. Three months later the purchaser’s equity bridge arrives, treating that balance as a debt-like item, and the argument starts over how much of it should come off the price. Neither side is wrong. The target must still deliver those services, and the purchaser says it should not pay for cash that arrives with obligations attached; the seller says prepayment is simply how the business model works. Neither is really arguing about the €4 million. They are arguing about what serving those customers actually costs, and on a deal of this size the gap between their two answers runs into seven figures. We take that fight apart in Post #5, The equity bridge: pricing deferred revenue. What matters here is that both sides should sign knowing a seven-figure price discussion is still waiting for them.

The LOI cannot finish the ARR figure or the deferred revenue discussion. What it can do is fix the basis on which both will be had: a short pricing annex setting out the convention and the intended treatment of deferred revenue, with a half-page ARR schedule alongside it. Two pages, and it does not require a single figure the parties do not already have.

What it costs to leave the pricing annex blank

A Dutch case from the summer of 2025, on the sale of an IT service management company, shows the shape of the problem (Amsterdam District Court, interim relief judge, 6 August 2025, ECLI:NL:RBAMS:2025:5776, annotated by Ruygvoorn in JOR 2026/64). The two shareholders agreed to sell for €5 million: €3 million in cash and €2 million in shares of the purchaser, the purchaser to be valued cash-free and debt-free in accordance with Annex 1. Annex 1 was to contain the method for valuing those shares and did not exist when the term sheet was signed. The purchaser’s value rested on a €30 million indicative bid from a private equity house at a multiple of twelve, which the sellers thought implausible for the sector. When they asked for the underlying figures and did not get what they wanted, they walked, more than twelve months in, after the parties had organised a joint “meet and greet” with each other’s personnel, and close to the date they had set for closing.

The court held the break-off unlawful: it was too abrupt and too unexpected, given how far the negotiations had advanced and that the parties’ actions until then had been directed entirely at completing the transaction. The court ordered the parties back to the table, holding that the missing valuation annex was itself part of what had to be negotiated. Ten weeks later a second judge lifted that order, finding further negotiation pointless (Amsterdam District Court, 20 October 2025, ECLI:NL:RBAMS:2025:7736). The purchaser won the first case, lost the second, and ended up with no deal.

That outcome is the exception: the Dutch threshold for an unlawful break-off is high, and most negotiations end without consequence. What the case illustrates is not the litigation risk but the cost of leaving a valuation method blank while it governs a significant part of the price. The sellers were taking €2 million of that price in the purchaser’s own shares, so what the blank annex cost them was confidence in the party they were about to become shareholders in.

The risk side: the warranties and liability package the price depends on

The other vessel gets the same treatment, and it is the one parties often leave to “customary”. Where the price side is narrow, a number and the definitions underneath it, the risk side is as broad as the business, and what carries the value moves from deal to deal: the technology, the data, the essential customer contracts, the licences the product runs on, the people who built it. Naming the two or three that matter in this deal is usually worth more than any general formula.

Fundamental status is cheap in the LOI and expensive in the SPA

“The parties will agree customary representations and warranties” is commonly found in LOIs and looks settled. In practice it does rough duty as a middle ground: nobody signing that line expects the purchaser’s heaviest package or the seller’s lightest. But no party knows exactly what the other believes customary to be.

Parties usually negotiate the warranty catalogue, and the liability architecture of caps, thresholds and survival periods, during the SPA phase, and usually that is soon enough. It is not soon enough when the valuation rests on a single thing being true. If a purchaser is paying for a company on the premise that it owns its core technology, its data or its models, that warranty is not an ordinary business warranty to be capped and time-barred with the rest.

Take Deverel, the founder-owned software house from Post #1 of The Tech Deal. Nine freelancers wrote modules of its core product between 2003 and 2012, none of them ever signed an assignment deed, and under Dutch copyright law those nine people still own what they built. Cap and time-bar that warranty with the rest, and most of the downside risk shifts to the purchaser.

Wait and raise fundamental status in the SPA, and it is a concession the seller will resist and you may have to buy with something in return. State it in the LOI and it is a premise the seller accepted going in. Why ownership warranties earn that status is a question for later in the series; the point here is when to claim it.

Lock in the founders early

Where the technology lives substantially in a founder or a small key team, which is often the case in tech deals, the purchaser knows from the outset whether keeping those people is part of its investment thesis. The question is how it makes sure they stay. Roll-over, incentive plans, retention packages, earn-outs, notice periods and non-competes belong in the LOI in outline. Outline is usually enough; the mechanics can wait. What the purchaser wants is their commitment before it spends a lot of additional time, energy and money on the deal.

Who stands behind the warranties: the seller or a W&I policy

The other item that materially changes the risk profile is whether the seller stands behind the warranty and liability package itself, or whether the purchaser takes out warranty and indemnity (W&I) insurance.

In case parties choose the first, they usually resort to a customary warranty set (as discussed above), with recourse on the seller to be secured via escrow, a keep well arrangement or a parent company guarantee, depending on the type and financial status of the seller.

When parties opt for W&I insurance, several questions need to be considered at the outset. First, what residual liability will the seller bear? If the seller bears zero recourse behind the policy, and every seller would ideally push for that, the question is who covers the gaps in the policy, because there will always be gaps. Some are standard exclusions, others are areas the diligence did not adequately reach, because insurers expect the warranty set in the SPA to have been adequately diligenced. And then there are the matters the diligence did reach, but which have been fairly disclosed as a matter not covered by the warranty, which then also fall outside the W&I coverage, as “you can’t insure a burning house”. The carve-back to ask for is the warranties the valuation rests on, to the extent the policy will not cover them.

The second question ties to the first one, as it drives what the scope of the diligence will be. The paradox is that a seller pushing for zero recourse will need to allow extensive diligence to maximise the purchaser’s potential coverage under the policy, but by doing so more warranty limitations may surface in the disclosure process, which then effectively increase the gaps the purchaser was trying to minimise. Which will then reinforce the residual recourse discussion.

The third is the question of who bears the premium. The premium is worth raising now as part of the price discussion, though in practice it tends to wait until the policy is negotiated and both sides know the size of the number, which largely depends on the scope of diligence performed, the policy coverage and the exclusions. The purchaser will frame it as the seller’s costs as it reduces the seller’s risk profile, whereas the seller will argue the contrary as the purchaser receives a counterparty with deep pockets. And in roll-over situations, the purchaser does not have to bring a claim against its own management team or co-shareholder.

Process: agree the roadmap to completion

So much for the substance. The LOI’s second purpose is the process: how the parties get from here to completion, through the diligence, the timetable and the conditions. Two points on that roadmap are worth spelling out in a technology transaction, because it heavily impacts the diligence exercise and the transaction timing.

Technical diligence requires concrete arrangements

The LOI usually lists the diligence workstreams at a high level: financial, tax, legal. Listing a technical workstream alongside them is not enough, for one reason. Most of it runs like any other: records go in the data room, questions get answered, experts are made available. But a dependency and licence scan runs against the source code itself, not against documents describing it, and sellers rarely hand source code to a purchaser before signing. In practice it is run by an independent auditor who reports findings to the purchaser, by named reviewers with time-limited, read-only access, or by the seller itself on tooling and a configuration the purchaser has agreed to. Where the purchaser competes with the target, including through a platform company in a buy-and-build, access may also need a clean team. None of that is quick to negotiate, and it is the part that will not fit into a timetable agreed without it. So settle in the LOI the scope and depth of the technical workstream, including who runs the scan and who gets to see the code.

The same applies to anything else that cannot be run out of the data room: a freedom-to-operate assessment by a patent attorney, penetration testing, a visit to a site or a piece of hardware. Each needs access or a third party the seller has to agree to, and naming it in the LOI is what lets the purchaser insist on it later without receiving pushback for “over-asking”. It also matters for W&I, because the policy’s cover follows the diligence scope.

Put the screening question on the radar before the timetable is fixed

If there is any chance the target counts as sensitive technology, which today covers hardware, semiconductors and photonics, with artificial intelligence and several further technology fields expected to join from 2027, the LOI is where you put it on the radar, not where you analyse it in full. Unless it is plainly out of scope, flag it: a line recording that the parties will carry out the triage together, that an investment-screening (Vifo) notification may be required, and that completion is conditional on clearance. That does two things. It commits both sides to doing the analysis and, if it is needed, the filing. And it builds the triage and any notification into the timetable, rather than discovering them when the closing date is already fixed.

In summary: the points that move the needle in a tech LOI

Whether each of these is settled in the LOI or left to the SPA, and in detail or in principle, depends on which side of the table you are on. And it is worth deciding intentionally before signing.

• The price basis. Enterprise value, cash-free and debt-free, assuming a normal level of net working capital, and the treatment of deferred revenue, in a one-page pricing annex.
• The metric. If the price is a multiple of ARR, attach a half-page definition: what recurs, what does not, at what date it is measured, and whose numbers govern.
• The fundamental warranties. Name the ones the valuation rests on, being ownership of the code, the data, the models and the inbound licences, as fundamental rather than ordinary.
• Who stands behind them. Whether recourse runs against the sellers or a W&I policy, and if the LOI opens on zero recourse, a carve-back for the (fundamental) warranties, which the seller stands behind to the extent the policy will not.
• The founders and key staff. Whether they roll over, what incentivises them to stay and their commitment to the process before the purchaser spends another three months on it.
• The conditions to closing. A satisfactory diligence outcome, agreement on the transaction documentation, the internal and external approvals required, financing, and a Vifo clearance condition if the target might be sensitive technology.
• Diligence and technical diligence. The workstreams and the timetable, and for the technical workstream the scope and depth: the dependency and licence scan, who runs it and who sees the code, and anything else that cannot be run out of the data room. Scope drives what the W&I policy will cover.

Next up

With the LOI signed, we turn to what the purchaser thinks it is buying. Next post: the investor’s view, in conversation with a private equity investor about selecting, valuing and acquiring targets in the software and AI business.