Mykola Cheremisin describes a pre-launch partnership session as a filter, not a ceremony. In his practice, he says, about 20% of prospective partner groups that do this work before trading decide not to proceed. [CD-01] No sample size or underlying case list was supplied, so that figure is a practice estimate, not a market rate. The useful result is narrower: a structured conversation can produce a clear stop before shared money and operating dependence make stopping expensive.
What the ranking says
The highest-cost omissions are exit, decision authority and money. If co-founders cannot produce a shared sentence for those three rows, polishing a mission statement or debating job titles adds little. The ranking below orders eight decisions by the likely cost of postponing them. It does not score whether the people like or trust one another. It tests whether they can turn trust into rules that still work when incentives diverge.
The eight-decision table
| Rank | Decision to settle | Question that forces a choice | Observable pass | Stop signal |
|---|---|---|---|---|
| 1 | Exit and deadlock | What happens if one person wants out, cannot continue, or blocks a decision? | A named sequence, valuation basis, deadlines and ownership of the next action | “We will work it out when it happens” |
| 2 | Decision authority | Which decisions belong to one role, which require consultation, and which require every partner? | A decision map with an owner and an escalation path for each major domain | Consensus is required for everything, or nobody has the final call |
| 3 | Money | Who is paid first, when can profit be distributed, and who must provide more capital? | Salary, distributions, reserves and new funding are separated and triggered by named conditions | Different assumptions are hidden inside the word “fair” |
| 4 | Equity and contribution | What does each stake reward: cash, past work, future work, access, risk or control? | Every stake has a stated basis and a response if the promised contribution does not arrive | Percentages are fixed before contributions are defined |
| 5 | Role and workload | What result does each partner own, and what level of time or performance counts as delivery? | Outcomes, minimum commitments and a review cadence can be observed by the group | Titles substitute for outputs |
| 6 | Information | What can every partner see, how often, and who controls accounts, contracts and reporting? | Access rights and reporting rhythm do not depend on one person’s goodwill | One partner has exclusive visibility into money or commitments |
| 7 | Renegotiation | Which events force the group to revisit the arrangement? | Dates and triggers are named: funding, relocation, illness, missed targets, new business line or changed workload | The agreement is treated as permanent despite a changing business |
| 8 | Related parties and side projects | Can partners hire relatives, invest elsewhere, compete, or use shared assets outside the company? | Disclosure, approval and conflict-handling rules are explicit | Each person assumes their own unwritten norm is obvious |
The table is deliberately stricter at the top. A vague answer in row eight can create friction. A vague answer in row one can make every later disagreement harder to resolve. The order is therefore about reversibility, not morality: settle first what becomes hardest to repair after the company has customers, staff and cash flow.
In my pre-launch practice, about 20% decide not to enter the partnership. That is a good result: the session has found the problem before the business has to pay for it. [CD-01]Mykola Cheremisin
How to use the table in one meeting
Give every participant the same eight rows before the meeting. Each person writes a preferred answer and an unacceptable answer without consulting the others. In the meeting, compare wording rather than intentions. “We both want growth” is an intention. “The commercial lead may approve spending up to an agreed threshold; above it, both partners decide” is a rule.
Mark each row green only when the group can state one answer and one trigger for revisiting it. Mark it amber when the direction is shared but an owner, threshold or deadline is missing. Mark it red when the answers conflict or someone will not state a position. The meeting succeeds when the colours are honest, not when every row is green.
Cheremisin argues that participation in wording is itself diagnostic. In one case from his practice, one member of a proposed three-person group declined to sign after the session and draft had made the commitments concrete. [CD-02] A single case cannot establish a rate. It does show the type of information the test is designed to reveal: willingness to accept a specific commitment can differ from enthusiasm for a general partnership.
What the numbers do and do not mean
The 20% figure is not evidence that one in five co-founder groups should split. It is not a forecast for a new client, and it is not a benchmark for advisers. It is Cheremisin’s estimate for the subset of his own cases that arrived before launch. The denominator, period covered and selection criteria were not published. [CD-01] The refusal by one of three participants is a case, not a frequency. [CD-02]
Sources and scope
- Expert material
- Mykola Cheremisin’s DATUM questionnaire, received 30 July 2026; his public description of partnership sessions and agreement topics at cheremisin.partners.
- Scope limit
- The ranking is an editorial synthesis of recurring co-founder decisions. It is a meeting tool, not a predictive score. Practice figures remain attributed to the expert wherever they appear.