What to test first, and what each test costs
Each assumption is scored on how far the answer moves, how wide the honest band already is, and whether specific work would narrow it.
- Test
- Reconstruct advisor hours per engagement across every case delivered, and plot the trend
- Sensitivity
- Very high. The entire margin conclusion runs through 80 hours; at 40 the absorbed margin at $40,000 is 74.0% and the argument dissolves
- Uncertainty
- Wide. The figure is derived from the company's own process description, never measured
- Cost
- Days of internal time
- What it proves
- Strong. Internal records, no counterparty, no interpretation
- Test
- State realized prices for every engagement, and whether the last tranche closed
- Sensitivity
- High. Fully absorbed, $30,000 returns 41.5% and $50,000 returns 64.9%
- Uncertainty
- Wide. Only the band is known
- Cost
- Hours
- What it proves
- Strong
- Test
- Establish the true pilot-to-annual conversion rate, and whether the stated pipeline includes the pilots already run
- Sensitivity
- Very high. The second question alone halves or doubles every near-term revenue figure
- Uncertainty
- Wide. One conversion, and an ambiguous denominator
- Cost
- Days of internal time
- What it proves
- Strong
- Test
- State the value of the converted account
- Sensitivity
- High. Every scenario prices conversions, and no contract value is evidenced
- Uncertainty
- Narrow once asked
- Cost
- One conversation
- What it proves
- Strong
- Test
- Price the next three cases at the top of the band and hold
- Sensitivity
- High. It is the difference between 41.5% and 64.9%
- Uncertainty
- Moderate. The ladder suggests room; nothing proves it
- Cost
- No cost, a decision
- What it proves
- Strong. A close or a refusal both inform
- Test
- Offer the Navigate monitor as a separate annual subscription to three existing clients
- Sensitivity
- High. It is the only naturally recurring component and would change what the revenue is
- Uncertainty
- Wide. Never sold separately
- Cost
- Weeks of selling time
- What it proves
- Strong either way
- Test
- Run one engagement with a deliberately reduced advisor role and measure quality and hours
- Sensitivity
- Very high. It is the test that decides the category
- Uncertainty
- Wide
- Cost
- One engagement
- What it proves
- Strong, and the only direct test of the software thesis
- Test
- Document value delivered in one completed engagement using the client's own numbers
- Sensitivity
- Moderate. It substantiates pricing rather than changing the economics
- Uncertainty
- Wide. The $150K figure has no derivation
- Cost
- One client conversation
- What it proves
- Moderate. Client-reported, so it needs care
- Test
- Sell one annual license without a founder in the room
- Sensitivity
- High. It is what an investor funds when they fund go-to-market
- Uncertainty
- Wide
- Cost
- A quarter
- What it proves
- Strong
- Test
- Name the competitive set properly, including the adjacent decision-intelligence category
- Sensitivity
- Moderate
- Uncertainty
- Moderate
- Cost
- Days
- What it proves
- Moderate
Each assumption from the ratings above is scored on three separate things: how much the answer moves if it is wrong, how wide the honest uncertainty already is, and whether a specific piece of work would narrow it and what that work proves. Sequencing follows how much each test tells you per dollar, not what depends on what.
Tests 1 through 4 cost almost nothing and settle the questions everything else depends on. They are internal records and one phone call, and they should be complete before the next investor meeting rather than after it. A company that walks into diligence without them is discovering its own numbers in the room.
A prerequisite rather than a test: a SOC 2 Type I, which takes four to eight weeks and unblocks enterprise procurement. It buys no revenue directly and its elapsed time is what makes it urgent.
The risks no experiment retires, and what to do about each
These are not hypotheses and treating them as testable produces mitigations that neither test nor protect. Each needs a position taken deliberately.
- Treatment
- Make the strategy robust. Concentrate investment on the parts that are not reasoning: the durable record, the gated method, the client relationships. Anything whose defense is prompt structure should be treated as temporary and not capitalized
- Treatment
- Take the bet with eyes open, and shorten the horizon it is taken over. The current 10x price advantage is the thing being sold; plan on it narrowing rather than holding
- Treatment
- Build an escape hatch. Keep the burn low enough that the round can be smaller than hoped without the plan collapsing, which argues against hiring delivery capacity ahead of contracts
- Treatment
- Make the strategy robust. Test 9 is the direct measure of whether the motion transfers, and it is also the mitigation
- Treatment
- Take the bet, with the evidence that it is a reasonable one: transition spending is becoming more selective rather than smaller, and a more selective buyer needs better selection
Section added · The brief asks which actions most reduce execution risk; this section sequences them by what each one tells you per dollar