Documents that agree with each other
A shareholders agreement and articles that work together, rather than two documents saying different things.
Commercial and corporate
Founder arrangements are made when everybody trusts each other and tested when they no longer do. By then nothing can be renegotiated, because the person who benefits from the gap has no reason to close it. What is agreed at the start is not paperwork. It is the entire set of options available to the company later.
The most damaging version is an equity split agreed at the beginning with nothing attached to it. A founder leaves early, for good reasons or bad, and keeps everything they were allocated for work they will now never do. Every conversation afterwards, including every conversation with an investor, is shaped by a block of shares held by somebody who is no longer in the building.
The next is a company with nothing written at all. Shares were issued, roles were understood rather than defined, and decisions have been taken by agreement because everybody agreed. Consensus works until it does not, and it is not a mechanism. When it fails there is nothing to fall back on except the default position, which nobody chose and nobody has read.
Then there is deadlock. Two shareholders, equally held, no casting vote, no route to break a tie and no way for either to leave. The company cannot act, and the more urgent the decision the more effective the deadlock becomes as a negotiating position for whoever prefers inaction.
The quietest version is the company that does not own what it is built on. A domain in a personal name, accounts registered to an individual, a brand or a codebase created before incorporation and never formally transferred. The company everybody owns shares in does not own the thing everybody is arguing about, and that discovery tends to arrive at the worst moment.
The articles and any shareholders agreement, read together, decide almost everything. Rights that people assume come with shares, such as a board seat, a veto, access to information or a share of profits, come from those documents rather than from the size of a holding by itself. Where the two documents say different things, which prevails is a question of how they were drafted and of what was done to make them work together, and the answer is often untidy.
Leaver terms and vesting decide what happens when somebody goes. Whether shares can be recovered, in what circumstances, and on what basis are questions that can only be answered by what was agreed at the outset. They cannot be imposed afterwards on somebody who has already decided to leave. Once a departure is live, every party's position is fixed by whatever was signed while they still trusted each other, which is why this is the single most valuable thing to get right early.
Control and ownership are different things and are frequently confused. Who can appoint and remove directors, which decisions need consent and from whom, and what the board can do without asking anybody are all separate from who holds what. Directors owe their duties to the company, which is not the same as doing what the shareholder who appointed them wants, and that distinction becomes very sharp during a disagreement. What protection a minority holder has is largely what was negotiated for them in advance.
Then there are the events nobody plans for. Death, serious illness, divorce, the insolvency of a holder, or somebody simply wanting to sell and get out. Whether shares can be transferred, to whom, and whether the others get the chance to buy first are decided by transfer and pre emption provisions. Without them, shares can end up held by people the remaining shareholders would never have chosen and cannot remove.
A shareholders agreement and articles that work together, rather than two documents saying different things.
Set at the start, when they can still be agreed, so a departure does not leave equity stranded.
Brand, code, domains and accounts held by the company rather than by whoever happened to set them up.
What the board can do alone, what needs consent, and a route out of deadlock before there is one.
What happens on a death, a divorce, an insolvency or a shareholder who wants out.
Working out what the existing documents actually permit, and finding the least destructive route through.
A single founder company with no outside investment does not need a shareholders agreement, because there is nobody to agree with. What it does need costs very little: the intellectual property held by the company, clean statutory registers, and accounts and domains in the company name. That work is worth doing on day one and becomes progressively harder to do afterwards.
It is also worth being realistic about what a document can fix once a relationship has broken. When a founder is already leaving badly, drafting something new does not help, because the other side has no reason to sign it. At that point the work is understanding precisely what the existing documents permit and what they do not, and negotiating within that. Money spent on an ideal set of terms that will never be signed is money spent on nothing.
Positions harden the moment the other side takes advice, and the quiet routes stop being available once a demand has gone out. While nothing has been sent, everything is still open.