by Scott Kupor
But, as we saw before, that introduces other potential complexities. It creates very different financial incentives between the various classes of preferred investors in the event of an acquisition where the price does not exceed the full amount of preferences. Also, as a common shareholder, you would certainly prefer to have the total amount of the liquidation stack reduced to create a greater likelihood that you and your employees will be able to receive some proceeds from an eventual acquisition offer. This new investor likely wants this, too, as she wants the employees motivated to work hard toward a good financial outcome, rather than everyone feeling as though they are working with no prospect of ever realizing a financial return.
As you can imagine, this made it extremely difficult to get a new investor comfortable with even issuing a term sheet and nearly killed the company’s financing prospects. The various investors eventually worked it out, but it took months of back-and-forth negotiations, all while the company remained idle without the needed cash infusion to continue building its business.
There may be times when giving different classes of preferred stock makes sense—and you do see this oftentimes as companies get more mature and raise larger amounts of capital at higher prices. But, doing so for XYZ Company at such an early stage is generally not recommended. Once you set the precedent of giving individual series of preferred stock their own votes, it’s very hard to take that back. So for a Series A term sheet like we have here for XYZ Company, having the capital “P” Preferred vote on voluntary conversion is a smarter path to pursue.
Antidilution Provisions
Whenever VCs invest in a company, they hope that the valuation of the company keeps increasing at every round. If that happens, there is no need to worry about antidilution protection. But hope is not a strategy, so better to be safe than sorry.
Antidilution protection provides some element of safety in the event that the company raises money at a valuation below that at which a VC invested. We call this a “down round,” since the valuation is in fact down from the prior round. That’s not a fun place to be for either the VC or the founder/employees since the effect of a down round can be highly dilutive to everyone’s ownership stake. This is because the company will need to issue a significant number of shares per dollar of capital it seeks to raise, and the addition of those shares to the company’s capital stock means that all existing shareholders will own proportionately less of the company than they did prior to the down round.
Every new financing of course has some element of dilution, since in all cases—whether the price is high or low—the company has to issue new shares. But down rounds are particularly painful for two reasons. First, the lower price means that more shares need to be issued to raise a fixed amount of capital than if the price were higher. Second, in the case of an “up round,” even though new shares are being issued, everyone is generally happy, notwithstanding the dilution, because the value of the company has increased. So while you may have been diluted by 10 percent as a result of the issuance of new shares, the value of your ownership position in the company should be much higher as a result of the higher valuation assigned to the company.
So the VCs have engineered what might be safely regarded as “schmuck insurance”—we think the company is worth five dollars per share on the day we invest in the company, but if, in the future, the valuation turns out to be two dollars per share, the insurance provides a price adjustment to minimize the otherwise dilutive effect of the two-dollars-per-share financing round.
The extent of the price adjustment depends on the precise type of antidilution protection.
In our VCF1 term sheet we have broad-based weighted average antidilution protection. That’s a mouthful, and the formula, which I am not going to print here (you can Google it), is equally daunting. In simple terms, think of the broad-based weighted average as an intermediate form of antidilution protection. VCF1 does not get to reset its original purchase price fully to the new, lower purchase price, but it does get a blended price in between these two that is weighted by the amount of capital raised in the different financing rounds.
Contrast that with the true insurance policy that provides the VC a complete price reset—that’s called a “full ratchet.” In a full ratchet, using our five-dollars-per-share/two-dollars-per-share example from above, our VC would essentially ignore its original five-dollar price and reset its stock holdings based upon the two-dollar price. Mathematically, this means the number of shares that the VC will now hold based on its original investment in the company increases by roughly two and a half times (5/2). As you can see, the full ratchet therefore protects the VC from getting diluted by this down round of financing.
But what about the founder and employees? Tough luck—they don’t have such a mechanism, so in effect they are subsidizing the VCs’ antidilution protection by eating the dilution on their end. Before you get too worked up over this, as a practical matter, many VCs when faced with a situation where the antidilution protection will kick in may be willing to offset some of that founder/employee dilution by increasing the option pool and granting additional options to these folks. It won’t solve the problem completely, but will put a dent in the dilution suffered by the common shareholders.
If you take this to its logical extreme, you get into a circular problem: the more the VCs get antidilution protection, the more the common shareholders get diluted, and thus the more the VCs are inclined to increase the option pool and grant them more shares, and the more in turn this dilutes the VCs. There is no perfect equilibrium for this problem, but sometimes the VCs will waive or modify their antidilution protection to prevent this circular spiral.
We talked earlier about whether you should always maximize valuation in a given round, and here is where the rubber really meets the road—in the context of a down round. Not only do you have to deal with the lack of momentum and upset feelings of your employee base, but down rounds have real economic consequences in the form of antidilution protection for the VCs.
This is largely avoidable if you try to structure your current financing to maximize the likelihood of success for your next round of financing.
Voting Rights
This section doesn’t really say too much, as the meat of the voting is going to come in the protective provisions section below. Important to note here, though, is that each share—both common and preferred—has one vote. As we briefly discussed above, when some startup companies go public, they have different classes of shares with different voting rights. While it is not unheard of when the companies are still private, it’s pretty unusual. In our term sheet, the voting follows the normal convention of one person, one vote.
As dual-class stock has started to proliferate among some public technology companies (e.g., Facebook, Google, Snap), some startups have been thinking about whether to adopt these structures as a private company.
There are two flavors of this that we’ve been seeing of late. First, some founders would like to have a high-vote stock that applies to their shares only. The theory is that, in addition to likely controlling the board, as we’ll talk about in the next section, they also want to ensure that any time a corporate matter requires a shareholder vote, they have enough voting power to retain control over those actions. For example, if the company were to be acquired and a shareholder vote was required to approve the transaction, if the founders have, say, ten times the number of votes per share relative to any other shareholder, they are likely to be able to control the outcome of that vote. There are not any instances of this type of voting structure that I am aware of having actually been implemented in startup companies.
What has happened in a very small number of cases is that some founders have asked certain investors to enter into what’s called a “voting proxy” in connection with their investment. This means that the investor hands the voting authority for her shares to the founders, such that the founders get to execute that v
oting authority in favor of whatever corporate actions come to a vote. While this is also very unusual, we do sometimes see this with very-late-stage, passive investors who are interested in the financial investment opportunity only and not in participating in the governance of the company.
The second variety of dual-class structures, which is more common than the first (although still in a minority of deals), is to put in place what’s known as a springing dual-class structure in anticipation of an IPO. This means that we keep the one person, one vote construct while the company is private, but that immediately prior to an IPO, the dual class springs into being. The usual implementation of this is to have all existing shares at the time of the IPO convert into supervoting shares, including both common and preferred shares. Then the shares that are issued in the IPO have the regular one person, one vote. The theory here is that over time the venture investors will exit the company by selling their stock in the public markets and, when they do so, the supervoting attached to those shares disappears. Thus, likely within the first few years of the company’s being public, the founders are left with significant governance control because they still have their supervoting shares, while the general public holds shares with only the single-vote structure.
CHAPTER 10
The Alphabet Soup of Term Sheets: Part Two (Governance)
Now we are starting to get to the heart of the governance structure of our company. This basically means: who gets a say in what happens in the company? Nobody likes to focus on these sections very much, but they turn out to be very important. For example, the board composition matters a lot. After all, the board gets to hire or fire the CEO and vote on major corporate actions—raising money, selling the company, etc.
And the protective provisions, which also determine what corporate actions the preferred shareholders (i.e., the VCs) get to have a say in, matter a lot. Ultimately, these are checks on the CEO’s ability to undertake significant corporate actions. The auto-convert, drag-along, and voting sections are also part of this governance bucket.
So although these are less sexy to talk about, if you are a founder, make sure you don’t just ignore these and focus exclusively on the economic issues. They can definitely come back to haunt you later in your company’s life!
Let’s make sure that doesn’t happen to you.
The Board of Directors
We’ll talk more about the role of the board of directors, but probably the most foundational thing that a board does is to hire (or fire) the CEO. Understandably so, many founder CEOs have been paying much more attention to the composition of the board of directors given historical concerns over VCs being quick to replace the founder CEO.
In our term sheet, we have a three-person board. (There are no requirements for an odd-size board, but many people prefer this to avoid situations where the board vote may be deadlocked.) One person is appointed by the holders of the Series A preferred (this is VCF1); it is pretty typical for the major investor in an early-stage financing to have a board seat.
We simplified our term sheet by having only one venture investor in this round of financing, but it is often the case that there are multiple investors. In that case, there is typically what’s known as one “lead” investor who is driving the negotiation of the term sheet with the CEO and, as a result, is typically investing at least half of the total amount of the round. Given this lead position, that investor will likely be the board representative for that set of preferred investors.
The second seat is reserved for the common shareholders and designated to be the CEO. Note that the seat is reserved for the CEO, not for the founder per se. This means that whoever is sitting in that CEO seat at the time is entitled to the board seat. Sometimes the founder CEO herself asks to have the board seat designated to her directly (versus to the then-existing CEO).
At the beginning, this seems like a benign request, since she is in fact the CEO. But what happens if the founder gets removed as CEO or decides to quit the company of her own volition? If we hadn’t written the board provision the way it is in the term sheet, the founder would continue to hold the board seat. We call this “ruling from the grave” (or sometimes “dead hand control”), neither of which is a good position for the company to be in.
It doesn’t make a lot of sense for a founder who is no longer with the company to be sitting on the board. In cases, therefore, where the founder wants the seat designated directly to her, the VCs will often insist on some service requirement attached to her continued occupation of the board seat. That is, she can keep the board seat for as long as she is CEO (or maybe some other senior officer in the company), but she loses the seat when she ceases to hold one of those positions.
The third seat is reserved for an independent; that is, someone not otherwise affiliated with the company by virtue of being an investor or officer. The selection process here calls for the independent to be approved by the other two board directors.
If you take a step back, this is a pretty fair and even configuration of the board—the common shareholders are represented by the CEO, the preferred are represented by VCF1, and we have an ostensibly neutral third party who has no pecuniary interests in the company. Most corporate governance experts would view this as a balanced board.
But this isn’t what the boards always look like. In more recent times, some founders have insisted on having what’s called a “common-controlled” board, meaning that there are more board members representing the common shareholders than other classes of shareholders.
The reason for this is obvious: If common controls the board, then the VCs can’t really fire the founder CEO, because they likely don’t have the votes to do so. They would of course need to convince at least some of the common directors to go along with them, but in most cases the common board seats are controlled by the founders (since they have the most stock and therefore the most votes). Thus, removing a founder CEO will prove difficult. Some have argued that these board structures are at the heart of why there have been some high-profile CEO-board governance challenges of late in Silicon Valley.
The Uber case is illustrative here. During Travis Kalanick’s tenure as CEO, Uber had established a board of directors with up to eleven seats, only seven of which were filled at the time. It’s not unusual, by the way, to have some unfilled board seats in anticipation of filling the slots as the business needs develop. Travis effectively controlled three of the seven seats, because they were filled by him, his cofounder, and a third early employee of the company likely sympathetic to Travis. He also had the right to fill the remaining four unfilled board seats at his discretion.
So had the board tried to force a vote to remove Travis from his CEO role, he could have quickly filled the four open seats and thus would have been able to win such a vote. Ultimately, the board applied enough pressure, including a lawsuit filed by one of their major VCs, to convince Travis to resign. This avoided the need for a formal vote by the board.
The other thing to think about is what happens to the board configuration as XYZ Company goes through subsequent rounds of financing. Our perfectly balanced board is likely to be upset. If a new VC leads the next round of financing, they may be likely to ask for a board seat, in which case now the VCs will have two board seats to the one common and one independent.
There are no magic solutions to this problem, but sometimes the founder CEO will ask to add a second independent to provide more balance to the VCs’ two seats. Other times, the founder CEO will ask for a second board seat for common, in other words, to match the additional VC board seat with an offsetting common seat. Any configuration is permissible; where we end up is simply a function of the negotiating positions of each of the parties.
Protective Provisions
Recall that the voting rights section didn’t really tell us too much about who needs to vote on what things, other than to let us know that each share of stock (both common and preferred) has one
vote. But what we really care about is who needs to vote, and in what proportions, to pass various corporate actions.
Delaware law governs the default voting for corporate actions (most startups are incorporated in Delaware because it has the most well-developed set of laws and legal opinions on corporate governance and shareholder rights). It specifies the baseline of whether the common and preferred vote together or separately for various corporate items.
But the protective provisions are really an overlay on top of Delaware law. As long as they don’t lessen the foundations of Delaware law, the protective provisions give the parties to a transaction the ability to create some stricter rules of the road. In this term sheet—and in most other venture financings—the protective provisions grant the preferred shareholders (generally the VCs) additional say in various corporate matters.
Just as we saw in the auto-conversion section above, the protective provisions in our term sheet designated the capital “P” Preferred as the class of shareholders entitled to vote on the defined corporate actions. Recall that we talked about the wisdom of this in anticipation of later-stage financing rounds. In general, you want to avoid smaller minority investors in later rounds having greater governance control than they have economic interests. The way to do this is to lump all the independent series of preferred stock together into a single voting class versus allowing each series of preferred to have their own votes. Doing the latter means that each series of preferred can block the other—as we said, that’s not a good place to be.