In the final week of a financing negotiation, the distribution of argument across the clauses tends to be a reliable indicator of what the parties actually treat as consequential. The valuation figure absorbs hours; whether the option pool is calculated pre-money or post-money produces extended correspondence; the liquidation preference is contested line by line, its multiple and its participation feature each defended on separate grounds. The provision enumerating which decisions will require investor consent, by contrast, is frequently read once and accepted on the representation that it reflects market practice. Two distinct standards of scrutiny operate inside the same document, beneath the same signature, and the distance between them reflects less a disagreement about which rights carry value than an asymmetry in knowing when each right will make itself felt.
The same pattern is observable on the institutional side of the table, though running in the opposite direction. In the internal memoranda prepared for an investment committee, the economic assumptions are stress-tested through sensitivity analysis, dilution is modelled across successive rounds, and the preference stack is run against a range of exit values. The governance provisions travel through the same memorandum as a checklist: if the customary items are present, the section is treated as satisfied. What emerges is a set of provisions that goes unchallenged on one side because it is standard and unexamined on the other side for precisely the same reason. The portion of the document that produces the most durable effect thereby takes effect outside the analytical attention of both parties.
The mechanism underlying this behaviour is described in negotiation practice as term sheet asymmetry — the tendency of investment terms to distribute economic and governance power at different speeds and with markedly different visibility. It draws on two sources. The first is a difference in measurability: valuation, liquidation preference and anti-dilution enter a table, resolve into a multiple, and become comparable across offers, whereas a consent right exists as a probability distribution and reveals its cost only when the underlying event occurs. The second is a difference in repetition: the fund side negotiates this document dozens of times a year, while the founding side negotiates it a handful of times in a career, so the accumulated intuition about which provision binds under which circumstances sits on one side of the table only.
Characterising this tendency as an error would be misleading. The standardisation of governance terms genuinely lowers transaction cost; if every fund designed a control architecture from first principles in every transaction, both negotiation time and legal expense would rise by an order of magnitude. Standard provisions function as a substitute for trust at a stage where the parties do not know each other and information asymmetry is high, and in that function they are rational. The difficulty lies not in the shortcut itself but in the shortcut remaining fixed after conditions change: as the company grows, as the number of shareholders increases, and as the irreversibility of decisions rises, a consent threshold that was reasonable at the seed stage becomes a constraint that directly limits operating speed in a mature business.
Contrary to the common expectation, the institutional cost of the asymmetry does not surface in the first round. In that round the founder invokes almost none of the provisions accepted, because the company has not yet reached the scale at which the triggering decisions arise. The cost accumulates across the rounds that follow. In a structure where budget approval, key hires, the incurrence of debt, the formation of subsidiaries and the disposal of assets are each subject to investor consent, what governs decision speed is the length of the approval chain rather than the formal composition of the board. That chain lengthens when a new investor joins in the next round; where the consent of two separate classes of preferred stock is required, a single strategic decision becomes dependent on two distinct internal committee calendars.
The second and less frequently noticed surface of the cost is the contraction of the alternative set. In the negotiation of a subsequent round, an existing investor's consent right is priced by the incoming investor as an item of uncertainty, and that uncertainty is absorbed either as a discount, or as a condition precedent, or as an extension of the negotiation timetable. In an M&A process the same logic operates through the preference stack, whose distribution among founders and employees shapes the acquirer's offer architecture directly, while drag-along thresholds have quietly determined, at the moment of signature, the price range within which a sale can realistically be executed. The term sheet thus ceases to be a document governing one round and becomes a structural instrument defining the boundary conditions of every future liquidity scenario.
The third surface is institutional memory itself. Where no record is kept of why a control provision was accepted, the party seeking to renegotiate that provision two years later must reconstruct the logic of its own prior position from inference. Turnover within the founding team makes the loss sharper still: once the individual who signed the document has departed, what remains is the provision, not the consideration for which it was granted. A concession given in exchange for something specific, if that exchange is never recorded, comes over time to look like an unreciprocated surrender, and the party carrying it loses the ability to argue that the original bargain has already been paid for.
This tendency cannot be managed through individual vigilance, because the problem originates not in inattention but in two layers of the same document advancing at different speeds within a single process. The structural intervention separates into three components. The first is treating the economic and governance layers of the term sheet as though they were distinct instruments, each assigned to its own owner: the finance function carries the economic layer, the governance function carries the control layer, and each approves independently. The second is binding every control provision to a scenario — which concrete decision triggers it, on what time horizon that decision is expected, and what alternative remains once it is triggered. The third is a rationale record, in which the consideration obtained for accepting a provision is written down at the moment of negotiation rather than at the moment of signature.
A fourth component institutionalises a role that most teams do not carry: a counter-argument function seated at the table and explicitly not accountable for closing the transaction. In a team operating under closing pressure, any question capable of delaying execution appears implicitly costly, which is why the asking of that question must be assigned to someone whose mandate benefits from it being asked. The purpose of the role is not to reject provisions but to ensure that every provision accepted is accepted deliberately. If a provision is examined and then adopted unchanged, the intervention has not failed; failure consists in the provision passing without examination at all.
The transaction architecture practice BEIREK runs across capital-intensive and financed projects operates this separation as a matter of structure rather than judgement. Working on a term sheet, we resolve the economic provisions into a value bridge and the governance provisions into a separate decision-authority map, the latter setting out which decision is subject to which approval at which threshold — measured not against the company's present scale but against its projected scale at three and five years. Running that same map under a subsequent financing round and under an exit scenario makes visible which consent threshold, costless today, is capable of blocking a specific transaction later, and it does so while the provision is still negotiable.
Alongside this we maintain a concession register for each round of negotiation: which provision was granted for what consideration, which alternative structures were discussed, and on what grounds each was set aside, recorded on the day the decision is taken. The register is not an archival document produced after signature but the negotiating ground of the following round; when a new investor arrives at the table, it becomes possible to distinguish which elements of the existing structure derive from a historical concession and which still rest on a live structural rationale. Closing discipline attaches to the same record, with the final pre-closing review testing not the economic table but whether the decision-authority map corresponds to the text about to be executed.
The real subject of a term sheet is not what the company is worth today but who is able to take which decision, at what speed, independently of its founder; and unlike the valuation figure, that question is not open to renegotiation after signature. The asymmetry arises accordingly not from the counterparty's firmness but from two sets of rights sitting at the same table with unequal visibility. The question facing the party taking a seat at that table is therefore not whether the terms are fair, but whether it is known, at the moment of signature, which future decision each term has already bound.