Put the Deal in Writing Before the Handshake Hardens

Every partnership runs for a while on a shared memory. That memory is not a contract.

You do not need an acquisitionOne company buying another company, or buying its assets., one company buying another, on the table to belong on this page. Every founding team, every two-person consulting shop, every side project with a friend, runs for months on an understanding that lives only in conversation. It works fine, right up until it doesn't: a client shows interest, an investor asks for the paperwork, someone leaves, and suddenly the thing everyone "agreed on" turns out to have three different versions in three different heads.

The team behind Greenfield Analytics lived this in real time while building a data business with a technical co-builder who owned the underlying database, a rancher partner supplying the raw data, and a slate of new co-founders joining mid-flight. Nothing about their situation was unusual. A verbal understanding, a growing list of people who needed to hear different things at different times, a method worth protecting, and a name to pick before any of it could be pitched to outside money. What follows is how they turned that mess into paper, in the order they actually did it.

If you run a business with no data, no AI, and no acquisition in sight, the pages ahead still apply. Swap "the technical co-builder's database" for whatever your version is: a recipe, a client list, a pricing method, a piece of code. The lesson is the same regardless of what you're protecting.

Write the deal before you pitch it.

Verbal agreementsA deal that exists only in conversation and memory, with nothing signed., deals that live only in conversation and memory, nothing signed, feel solid because nobody has tested them yet. The test comes the first time real money shows up and asks to see the paperwork. On the Greenfield deal, that moment arrived early: an investor conversation was approaching, and the acquisition of the technical co-builder's database and expertise was still a set of conversations, not a document.

Writing the deal starts with how you frame it, not just what you put on paper. On the same call, the team worked through how to describe what was actually happening: not a data purchase, but the acquisition of a company and everything it carried with it.

Framing decided, the next question is what actually goes on the page. The method that turned a vague understanding into something a lawyer could use: three lists, no more.

"I think what we would need, this is something that only [the technical co-builder] would know, is a list of all of those assets... I'll just write up something, send it to all of us... and then that would be what we give to the lawyer."

The three lists: tangible assetsThings a company owns that you can point at: equipment, inventory, cash., the things you can point at, intangible assetsValuable things you cannot touch: data, brand, methods, relationships, code., the valuable things you can't, and IPCreations you legally own: code, methods, brands, designs, writing. with the rights attached to each, and obligations with dates. Nothing about this requires a lawyer to start. It requires someone in the deal sitting down and writing what everyone already believes is true, so that belief can be checked before it gets expensive to be wrong about.

One more question belongs in this same pass, before any document is drafted: what is each person actually here for, past the current project. On the same call, the team asked it directly about the technical co-builder himself.

"What role do you see [him] playing, let's say, let's say a year, two years from now?... Is [he] helping getting the database up to date and then retiring or doing his thing? Or [he] is someone that we always come to because he has the knowledge and the understanding? What is the trajectory here?"

Not every piece of a deal needs solving today. The harder, lower-confidence pieces, the parts where you genuinely don't know what's fair yet, get optionedA right, agreed now, to buy something later at set terms without committing today., given a right, agreed now, to be exercised later, instead of sold outright, and priced as mutual risk rather than one-sided.

"I would put it on the back burner. To get seed forecast going before, and then just put in an option as part of the agreement. If we do this, will you give us a five-year option... to get this going first, and then we can talk about that later... that protects you, so we don't sell from underneath you."

The same instinct shows up later in the engagement, when the team is structuring how an outside contributor gets paid for early, unproven work: "Sometimes it's more like, hey, we are actually doing more work... because this is front edge kind of stuff that he can't get anywhere. So even though he's taking a risk, we are also taking a risk in doing that." Early relationships are not one party doing a favor for another. They are two parties taking a risk on each other, and the paperwork should say so.

Not legal advice. This page describes patterns observed in one engagement, not a template guaranteed to fit yours. Have a lawyer in your jurisdiction review anything before you sign it.

Sequence who hears what, and form the team without freezing it.

Writing the deal down does not mean announcing it to everyone at once. Part of protecting a deal in progress is controlling which stakeholdersAnyone with a real say or a real interest in a decision, inside or outside the company., anyone with a real say or a real stake in the outcome, hear what, and when. On the same call, the team decided the rancher partner, a sensitive external relationship, did not need to hear about the acquisition conversation before the internal deal was settled.

1. Keep the sensitive party out until the internal deal is settled."We can just say that we're completing this acquisition in the paperwork and we don't have to wait for the paperwork... I think we can have the conversation with [the rancher partner]. And we can [say] this is in process right now. That's it. So we have clarity."
2. Sort every open founding decision into three buckets."I always divide decisions into, let's say, three categories. Decisions that are needed to be made now, and without it we can't move forward... decisions that can be made later... and decisions that are irrelevant if we make right now." Decide only what's blocking progress; leave the rest alone on purpose.
3. Don't overload the first few steps."So my take is don't overload the first few steps. As much as possible, let's make the necessary decisions. Let's highlight what we want and what we want to be considered for. And let's highlight as well our red lines."
4. Keep the starting structure flexible, especially on equity and board seats.Two financing instruments explain why. A priced roundAn investment where investors buy shares at an agreed company value, with full legal paperwork. is an investment where investors buy shares at an agreed company value, with full legal paperwork attached. A SAFEA simple investment contract where the money comes in now and turns into shares later, once the company is properly valued. is simpler: the money arrives now and turns into shares later, once the company is properly valued. The difference matters here because lock as little as possible before you have to. In US and Canadian venture practice, a priced round typically comes with restated corporate documents and a new voting agreementA contract in which shareholders agree ahead of time how board seats and key votes get decided. that sets board seatA voting position in the group with final oversight of the company's biggest decisions. composition anyway, so pre-funding governance fights over board seats are usually renegotiated the moment institutional terms arrive. SAFE-based seed rounds don't touch those documents at all until they convert. Either way, the fight is deferred, not avoided, and starting flexible means less of it. Practice varies by jurisdiction and instrument; confirm with your own counsel before locking anything.
5. Don't lock equity or org structure alone."Don't go too far down the road until we talk about the organizational structure and equityOwnership in a company, usually held as shares. — and implications of the decisions." Get someone who has actually structured a company before into the room before the cap tableThe record of who owns what share of the company. hardens.
6. Let the operating agreement, not the title, decide who decides."The operating agreementThe document that sets how a company's owners make decisions and split the rewards. — can define how decisions are made, separate from what the role is defined." A title tells people what you do. The operating agreement is what actually holds up when two people disagree about who gets the final call.

None of this is about moving slowly. Two co-founders on this same team hired specialized help part-time before committing to a full hire: "We would use them as a consultant, not that, we would eventually hire a staff person, but we'd use a consultant, somebody who really knows what they're doing, and maybe pay them at that rate for two days a week." Flexibility and speed are not opposites here. The flexible structure is what let them move at all.

Not legal advice. Financing-document norms and board-composition practice differ by country, and in the US and Canada by state and province. Confirm with your own counsel before treating any of the above as settled fact for your deal.

Protect the method. Pick a name built to last.

The rival raised money on borrowed language and still did not reproduce what took years to build. Protection is layered, not paranoid: the NDA matters, and it is not the only thing standing between your method and a fast follower. What actually protects you is the depth the method took to build in the first place. The paperwork just keeps things honest along the way.

Pick the NDAA signed promise to keep what the other party tells you secret. tier deliberately

Confidentiality about a deal is not one setting. It runs on a spectrum, and naming the tier out loud, before signing, avoids the argument later about what "confidential" was supposed to cover.

"The most restrictive is you will not disclose the existence of this agreement or the contents. There's one that sort of, you can disclose the existence of the agreement, but not the contents. You can disclose the existence of the agreements and high-level outcomes."

Not legal advice, practice patterns only. This three-tier description is a negotiation vocabulary, not a published legal taxonomy, and real NDAs add more dimensions (duration, compelled disclosure, carve-outs). Confidentiality and financing norms vary by jurisdiction. Have your own counsel review anything before you sign it. If your data comes from one or a few identifiable sources, remember that even "high-level outcomes" can re-identify them.

The same instinct extends past the NDA and into daily language. On one engagement, the team refused to say a client's name out loud internally, even among themselves: "We were not even allowed to say the [client's] name between the team internally. We would call it Project Banana." A codename costs nothing and closes an entire category of accidental leaks.

The same discipline applies to explaining your own product to the world. On a founding-team call bringing in outside counsel: "I can tell you what it does, because what it does is not a secret... How it does it, that's none of your business... keep it under 10 words. People don't read anymore." Outcome, not mechanism. That line is the whole method-protection habit compressed into one sentence.

Name the company for what it does, not for the hype cycle

Naming decisions on this team followed their own small discipline. Separate the company name from the product name: "We call the company [the new brand name], and then we name the software or something. An extension, yep."

Run every finalist through a trademarkLegal ownership of a name or logo in your market, so competitors cannot use it. search before falling in love with it: "No results found means no registered or pending trademark exist." Run the name gauntlet before locking anything in: "Everyone comes back with three names with, let's say, confirmed domains. And we rate them. And we debate them. And unless we're... at least over 8 out of 10 for kind of all of us, or we do the round again."

And pick a domain that survives the current wave of hype: "The challenge with AI is that let's say if AI dies tomorrow and everyone is expecting — those things unfortunately go away. So I would definitely kind of lean to the .com side, unless if it's not existent, we can go .io or .ai."