01
Step one: fix the question and the roles
Begin with one written question the collaboration must answer, plus a list of who contributes what. In a worked example, a lab specialist, a product designer and an operator want to test a water-sensing tool for small farms. The question becomes: will growers in one region pay for weekly soil reports? Each partner names their contribution, time limit and decision area. Writing this on one page prevents drift later. Include how disagreements are resolved and who speaks to outside parties. Agreeing early on the question and the roles matters more than polishing any plan or pitch document.
02
Step two: design a small test with evidence
Choose the cheapest test that could prove the question wrong. In the farm example, the team might interview twenty growers, offer a paper sample report to five, and record who asks for a second one. Set a measure of success before the work starts, such as how many growers request ongoing reports. Assign one partner to collect evidence and another to review it independently. Keep notes in a shared log with dates. This step turns enthusiasm into observations, and it gives every partner the same facts when the group meets to decide whether to continue, change direction or stop.
03
Step three: review, decide and document terms
At the end of the test, hold a decision meeting using the shared log. Each partner states what the evidence showed and what they will commit next. In the example, the group might agree to build a basic version for one region, with a clear budget and a further review date. Record the outcome, updated roles and any ownership or contribution terms, then have qualified advisers review anything binding. A short written summary keeps the venture honest and gives new partners context. If the evidence is weak, documenting an orderly pause is a useful result too.