How to Read a Term Sheet, Explained Through a Middling Sale

A term sheet explained in one sentence: its economic terms decide who is paid how much when the company is sold, and its control terms decide who can make the sale happen. The two meet in a middling sale, where the liquidation preference binds and the investor with the deciding votes can prefer a certain exit to a better gamble.

Most of the money terms have converged, at least in one large law firm’s deals. Of the 166 venture financings Cooley handled from April to June 2026, 95.8% carried a 1x liquidation preference and 96.4% used non-participating preferred stock (Cooley, 2026). What still needs reading is how the terms interact: the range of sale prices over which an investor is paid the same amount, and the votes that decide whether a sale inside that range goes ahead.

Term sheet explained: the preference splits a sale before the price does

In the model term sheet of the NVCA (National Venture Capital Association), only the no-shop and confidentiality clauses bind (NVCA, 2020); the economics land in the charter. There, a merger or a sale of substantially all the assets counts as a “Deemed Liquidation Event”, so the preference applies to an acquisition as well as to a wind-down.

The model offers two main versions. Non-participating preferred receives its purchase price first “(or, if greater, the amount that the Series A Preferred would receive on an as-converted basis)”. Participating preferred receives its purchase price, and “Thereafter, the Series A Preferred participates with the Common Stock pro rata on an as-converted basis”. The first is a choice between a fixed claim and a share; the second takes both.

The waterfall shows what that choice does. The company in the table below is hypothetical. Founders and employees hold 10.0M common shares; a $4.0M seed round at $1.00 a share and a $16.0M Series A at $2.50 are both 1x non-participating and paid side by side (pari passu). The Series A owns 31.4% of the 20.4M shares, and its round valued the company at $51.0M after the money.

The term sheet explained in dollars: a hypothetical company’s sale proceeds, $M

Sale price	Series A	Seed	Common	Common’s share	Common, if the preferred participated
10	8.00	2.00	0.00	0%	0.00
20	16.00	4.00	0.00	0%	0.00
25	16.00	4.00	5.00	20.0%	2.45
30	16.00	4.00	10.00	33.3%	4.90
40	16.00	6.86	17.14	42.9%	9.80
51	16.00	10.00	25.00	49.0%	15.20
100	31.37	19.61	49.02	49.0%	39.22

Source: hypothetical company; CX Cash calculation. Each series takes the greater of its preference and its as-converted share, given the other series’ choices, as the NVCA model charter provides. The seed converts above $30M and the Series A above $51M.

For the Series A, every sale price from $20M to $51M pays the same $16.0M, a range of 2.55 times (our arithmetic). Over that band of prices the common stock goes from nothing to $25.0M. With pari passu preferences, the band runs from the total of all the preferences to the price at which converting pays more. For the latest round, that upper bound is the round’s own post-money valuation, provided the earlier series, bought at lower prices, have already converted (our algebra).

Flat payout for the latest round: total liquidation preferences → the round’s post-money valuation

Participation removes the band and charges for it at every price. At a $30M sale, participating preferred would cut common’s take from $10.0M to $4.90M, and at $100M from 49.0% to 39.2% of the proceeds.

But clean terms still split preferred and common in a middling sale

Venture terms were harsher a generation ago. Steven Kaplan and Per Strömberg studied 200 investments in 118 companies by 14 venture firms between 1987 and 1999. Participating preferred appeared in 72 rounds, cumulative dividends in 46% and redemption or put rights in 84% (Kaplan & Strömberg, 2000); among Cooley’s Q2 2026 deals the figures were 3.6%, 3.0% and 5.4%. The two samples differ, 14 venture firms then and one law firm’s clients now, but a gap that wide is hard to read as anything other than a change in the market.

Clean terms still leave the flat range. Jesse Fried and Mira Ganor set out the consequence in a 2006 law review article that the Delaware Court of Chancery quoted in 2013. “Because of the preferred shareholders’ liquidation preferences, they sometimes gain less from increases in firm value than they lose from decreases in firm value”. The distortions “are most likely to arise when, as is often the case, the firm is neither a complete failure nor a stunning success” (In re Trados, 2013).

The waterfall shows why. Below $20M the Series A takes 80 cents of every dollar; from $20M to $51M it takes nothing more; the common takes every extra dollar once the preferences are covered. Suppose the company is offered $28M today, against staying independent with a 50% chance of a $70M sale and a 50% chance of nothing. The sale pays the Series A $16.00M against an expected $10.98M from staying, while the common’s $8.00M compares with an expected $17.16M. Staying is worth 25% more to the company as a whole ($35.0M against $28.0M, ignoring the time value of money), and only the Series A would sell. Participation would not change that choice; above $51M the conflict fades, because the Series A converts.

Therefore the votes decide who chooses inside the band

Kaplan and Strömberg’s central finding is that venture contracts “separately allocate cash flow rights, voting rights, board rights, liquidation rights, and other control rights”. The allocation moves with performance: “If the company performs poorly, the VCs obtain full control”. When things go well, “the VCs retain their cash flow rights, but relinquish most of their control and liquidation rights”. The flat range is the zone in between, and three control terms decide what happens there.

The board comes first. Michael Ewens and Nadya Malenko traced the boards of 7,780 venture-backed startups from 2002 to 2017. After the first round, 48% of boards were controlled by the entrepreneurs, 32% shared and 20% held by investors; by the third and fourth rounds investors controlled 52% and 63% (Ewens & Malenko, 2024). Protective provisions come second. Under the NVCA’s model charter the company may not “liquidate, dissolve or wind-up the business and affairs of the Corporation or effect any Deemed Liquidation Event” without the consent of the Requisite Holders, by default a majority of the preferred (NVCA, 2025). In the waterfall above, the Series A holds 61.5% of the preferred, so a majority threshold would let it block any sale on its own.

The drag-along comes third, and it works the other way: it obliges every holder to vote for a sale approved by named groups. The NVCA’s 2026 voting agreement notes that “the objective of the drag-along is not generally to grant the investors the unilateral right to force a sale”. It also warns that “Including the Board as one of the parties necessary to trigger the drag along can introduce fiduciary duty risks in light of the Trados decision” (NVCA, 2026). A veto lets an investor stop a sale; a drag-along with the right triggers lets a group of holders impose one.

Trados shows the money and the votes acting together

Trados, a maker of translation software, raised venture capital in 2000, and its investors placed directors on the board. Revenue grew every year but disappointed them. In July 2005 SDL bought Trados for $60M. The preferred stock’s preference, including accumulated dividends, was $57.9M, or 96.5% of the price (our arithmetic). A management incentive plan approved by the board took the first $7.8M, the preferred received $52.2M, and the common received nothing.

The court found that “six of the seven Trados directors were not disinterested and independent”, and that they had not followed a fair process. They won only because “the common stock had no economic value before the Merger”. The opinion restated the Delaware rule for boards of companies with preferred stock: “generally it will be the duty of the board, where discretionary judgment is to be exercised, to prefer the interests of the common stock”. Under that rule, a conflicted board that sells inside the flat range must be able to show why the common was worth no more.

Redemption adds a lever rather than cash. ThoughtWorks took $26.6M from SV Investment Partners in 2000 for preferred stock redeemable after five years “for cash out of any funds legally available therefor”. By April 2010 the claim had grown to $66.9M (SV Investment Partners v. ThoughtWorks, 2010). Over sixteen quarters the board redeemed $4.1M, and the court held that funds legally available did not simply mean surplus. The NVCA’s notes now treat a put mainly as support for a decision to sell, and add that “Redemption provisions are uncommon in early-stage financings.”

Does a good partner make the clauses irrelevant?

Silicon Valley Bank’s guide argues that it does: “Experienced venture investors and founders consistently say that the most important factor isn’t any single legal term — it’s who you’re working with” (SVB). SVB is right that good outcomes switch the terms off, and Kaplan and Strömberg found exactly that. The middle is another matter. Trados’s investors included Sequoia, and the court traced the push to sell to the structure of the business rather than to bad faith. In its words, “VCs also operate under a business model that causes them to seek outsized returns and to liquidate (typically via a sale) even profitable ventures that fall short of their return hurdles”. A 1992 study the court cited found that 20% of its sample of venture investments became the “living dead”, so a good partner meets the same arithmetic often.

With the term sheet explained this way, the money terms fix who is paid in a middling sale, and the board, the protective provisions and the drag-along decide whether that sale happens. In the hypothetical company, the Series A is paid the same at every price from $20M to $51M, and under a majority threshold its 61.5% of the preferred can block any sale in that range on its own.

Notes and sources

• Cooley LLP (2026). Q2 2026 Venture Financing Report, 17 August 2026; also the Q1 2026 and Q4 2025 reports.
• Ewens, M. and Malenko, N. (2024). Board dynamics over the startup life cycle. NBER Working Paper 27769 (September 2020, revised May 2024).
• In re Trados Inc. Shareholder Litigation, 73 A.3d 17 (Del. Ch. 2013).
• Kaplan, S. N. and Strömberg, P. (2000). Financial contracting theory meets the real world: an empirical analysis of venture capital contracts. NBER Working Paper 7660; published in Review of Economic Studies 70, 281–315 (2003).
• National Venture Capital Association. Model Term Sheet for Series A Preferred Stock Financing (2020); Model Amended and Restated Certificate of Incorporation (October 2025); Model Voting Agreement (June 2026).
• Silicon Valley Bank. Term sheets for startups: key components and how to read them (web page).
• SV Investment Partners, LLC v. ThoughtWorks, Inc., 7 A.3d 973 (Del. Ch. 2010), aff’d, 37 A.3d 205 (Del. 2011).

Deal-term frequencies come from Cooley’s reports, which cover its own clients’ financings, and from Kaplan and Strömberg’s 2000 working paper; the Fried and Ganor and 1992 “living dead” findings are quoted as the Trados opinion cites them. Payouts, ratios and expected values with no citation are worked out from the stated terms, and the company in the waterfall table, with its two rounds and its $28M offer, is hypothetical. The piece describes Delaware law and US practice and is not legal advice.