Instead of one large request, a sponsor asks for a first tranche tied to the few assumptions that must prove true, with checkpoints before more money, so approval costs little and each stage buys evidence.
An example
Small companies with under two million dollars in revenue were the market the head of small-business banking at a regional bank wanted to reach with online lending. A full launch would cost eleven million dollars, and the credit committee had rejected two similar proposals as too risky. She broke the plan down by the assumptions that had to hold: that applicants would finish an online application, that default rates on the bank's existing small-business loans would carry over, and that winning each customer would cost under four hundred dollars. She asked for 1.2 million dollars to test the first two, with a checkpoint after six months. If they held, she would ask for the next tranche to test customer costs at scale. The committee approved the first tranche in one meeting. Six months later completion rates beat the plan, defaults were in line, and the second tranche was approved.
Why it works
Large requests for uncertain ventures ask deciders to commit before anything is known, which makes them easy to refuse. Breaking the plan into stages tied to the assumptions that must hold makes approval easier: the first stage is small, the risk is visible, and each stage buys evidence before more money goes in. The venture is the same, and only the size of each decision shrinks. It depends on assumptions that can be tested cheaply and in order, on honest checkpoints, and on deciders willing to fund in stages. It fails if checkpoints are fudged to keep the money flowing.
The thinking behind it
Rita Gunther McGrath McGrath and MacMillan's Discovery-Driven Planning (1995) proposes that new ventures, whose outcomes cannot be projected with confidence, be planned around assumptions: start from what success must look like, list the assumptions that must hold for it, and test them at checkpoints before larger commitments are made. The small-business lending plan follows that structure, with completion rates, default rates, and customer costs as the assumptions to test.
Ian C. MacMillan In the plan McGrath and MacMillan describe in Discovery-Driven Planning, funding is released as milestones are reached. That is what makes approval cheap for a credit committee: each decision buys one test, and the next tranche waits for the evidence. It also explains the clean line, since staged funding protects the company only while the checkpoints report honestly.
When it fits
Staged funding fits new ventures and uncertain projects whose risk lies in assumptions that can be tested: new products, new markets, new technologies. The plan must break into stages, with deciders willing to fund that way, and a sponsor who will stop if an assumption fails.
How to use it well
List the assumptions the venture rests on, and rank them by how much they matter and how uncertain they are. Plan the stages to test the most important ones first.
- Ask for the first tranche only, with clear tests and a date.
- State in advance what result would mean stopping.
- Report results honestly at each checkpoint, failures included.
- Ask for the next stage with the evidence in hand.
Be ready to stop. A sponsor who keeps every venture going whatever the evidence says loses the trust that makes staged funding possible, and the next first tranche gets refused.
What it costs and where it backfires
Staged funding is slower than a full commitment, and competitors may move faster. The first tranche may be too small to test the assumptions properly. Each checkpoint is a point at which the venture can be killed, and the sponsor carries that risk. Checkpoints have to be honest. Reporting results selectively, or moving the tests to keep the money coming, turns a discovery plan into a slow sunk-cost trap.
When it is used on you
Approve a rival's small first tranche if the tests are clear, and hold them to the checkpoints. Ask in advance what result would mean stopping, write the answer down, and watch for tests that move after the fact. Ask for each checkpoint's results in writing.