Three fatal flaws that killed real business plans

Every plan our engine reviews produces three ranked fatal-flaw candidates: the specific structural problems most likely to end the venture, each with the test that would confirm or refute it. Across the three anonymised sample reports on this site, three distinct flaw patterns emerge. They are worth studying because none of them is exotic. Each one is visible in the plan's own text, each one was flagged by multiple independent review roles, and each one recurs constantly in plans we see. Everything below comes from the sample reports themselves, which you can read in full.

Flaw one: the structural mismatch reframed as an advantage

[Company A], a commercial real-estate operator that filed to go public in 2019, carried USD 47.2 billion of fixed 15-year lease liabilities funded by member revenue its customers could cancel at short notice. The plan did not hide this. It disclosed the mismatch, then described it as an operational advantage. All five review roles independently converged on the same conclusion: a long-duration fixed liability funded by short-duration cancellable revenue converts any demand shock into a solvency event, because the company must keep paying rent on empty buildings while its customers walk away month to month.

This is not a technology company with real estate; it is a leveraged sublease arbitrage wearing a mission statement, and the plan asks investors to fund the widening jaws of its own duration mismatch.

The engine's verdict was KILL, with the confirming test spelled out: model member revenue under a 30 percent occupancy shock held for twelve months against the disclosed lease schedule, and see whether the company survives without emergency capital. The public record supplied the answer. The IPO was withdrawn and the valuation collapsed to under 2 percent of its private peak within 24 months. The lesson is not that mismatches are always fatal. It is that reframing a disclosed structural risk as a strength, instead of pricing it, tells a hostile reader the plan has been narrated around reality rather than stress-tested against it.

Flaw two: a moat that rests on one unauditable number

[Company B], a consumer marketplace that filed to go public in 2020 after a pandemic-year revenue decline, is the survivor of the three, and that is exactly why it matters. The review produced a MODIFY verdict, not a KILL: the roles found genuine moat evidence, including a natural experiment in which the company cut marketing spend and demand partially returned on its own. But the moat narrative was asserted through five parallel claims while being load-bearing on a single figure, an 86 percent direct-traffic statistic whose computation the plan never disclosed. Alongside it sat a broader disclosure vacuum: no take rate, no cohort economics, no city-level revenue concentration, and a USD 3.4 trillion market figure built by bundling three unrelated categories.

The engine's test was simple: publish the auditable definition of the direct-traffic metric. If it survives a strict definition, the moat holds and the plan strengthens materially. If it includes branded paid search and re-engagement pushes, the moat collapses to one contestable figure. This company had the underlying strengths to survive its own opacity; it listed successfully. Most plans that lean on a single undefined hero metric do not, because the first diligence analyst who asks for the definition and does not get a clean answer discounts every other claim in the document. An unauditable number is not evidence. It is a liability with good presentation.

Flaw three: the raise that guarantees a re-raise from weakness

[Company C] is a composite ASX junior-explorer prospectus, assembled from the standard shape of the genre: a flagship project, six drill targets, a 24-month programme, and a net raise of AUD 5.4 million. The engine's top-ranked fatal flaw was arithmetic, not geology. Costed against realistic Australian drilling, assay and listed-entity overhead rates, the raise cannot fund the described programme to a decision-quality result. The vehicle is structurally engineered to run out of cash before its own hypothesis is tested, which means a dilutive re-raise from a position of weakness is not a risk in this plan; it is the plan's most predictable event.

The compounding flaws made it worse: a claimed moat consisting of re-interpreted public geophysics, which is by definition non-proprietary, and a governance posture with no directors-and-officers cover beyond the statutory minimum and a single managing director carrying execution, technical judgement and investor relations at once. The engine's prescribed fix was blunt: raise AUD 8 to 10 million or cut the programme scope by about 30 percent, commission an independent geologist's report so the moat re-forms around evidence, and repair the governance and funding posture together, not sequentially. This flaw pattern is the least dramatic of the three and the most common. It rarely produces a headline collapse. It produces a slow spiral of discounted placements that transfers the company from its founders to its rescuers.

What the three have in common

None of these flaws required inside information to find. Each was derivable from the plan's own disclosures, which means each was findable by the author before the audience saw it. That is the entire argument for adversarial review: the flaws are in the document, the question is only who finds them first, and on whose invoice. A structured interrogation before the real audience reads the plan lets you fix the mismatch, define the metric, or resize the raise while those are still editing decisions rather than public ones.

The three full sample reports, including every role excerpt, section verdict and confirming test quoted above, are free to read: Sample A, Sample B and Sample C. For the framework behind the section verdicts, see the seven sections every business plan gets interrogated on.