Is a high Fragility Index bad?

Higher means more exposed. 75 and above is Critical Fragility, 45 to 74 is Elevated Risk, below 45 is Resilient. It is a map of where a partner will attack, not a verdict on whether the business works.

What is the Fragility Index built from?

Three to eight load-bearing axioms, each scored 1 to 10 for how exposed it is, plus a five-axis radar: defensibility, unit economics, distribution and CAC, customer inertia, and incumbent retaliation. The 0 to 100 index summarises how much of the thesis rests on claims that could be false.

Is the Fragility Index a probability that the startup fails?

No. It measures how much of the argument depends on untested claims, from the text you submitted. A strong business with a weak deck can score high; a weak business with airtight evidence can score low.

How do I lower a Fragility Index?

Run the verification test attached to each axiom and replace assertions with evidence, adopt the steelmanned rewrite where the counter-thesis lands cleanly, and fix the radar axis that is a crater before spending a warm intro.

The score

What the Fragility Index actually measures

Updated 2026-09-16

Every Dialectic dossier opens with one number between 0 and 100. Founders read it the way they read a credit score, which is the wrong mental model. Higher is worse, it is not a prediction, and it is computed from your text, not from your company. This guide explains what it measures, how it is built, and what to do with it.

Higher is worse

The Fragility Index answers one question: how much of this thesis rests on claims that a skeptical partner could falsify in the first meeting? The dossier renders three bands:

75–100 · Critical Fragility. At least one load-bearing axiom is likely false and a partner will find it in the first meeting.
45–74 · Elevated Risk. The thesis holds on paper but has an axis a skeptical partner will drill; evidence is thin where it matters.
0–44 · Resilient. Every axiom has a falsification test and the radar has no crater. Rare on a first pass.

What it is built from

The index is not a vibe. It summarises two structures that the dossier shows in full:

Load-bearing axioms. Dialectic extracts between three and eight claims that, if false, collapse the story. Each is scored 1 to 10 for fragility and tagged by category: market demand, unit economics, user behaviour, defensibility, or execution friction. Each comes with a verification test you can actually run.
The 5-axis radar. Defensibility and moat, unit economics, distribution and CAC, customer inertia, and incumbent retaliation, each 0 to 100. Outward on the chart is worse. One axis at 90 or above is a crater, and it is the first place a partner will drill.

The per-axiom scores and the radar are the evidence; the index is the headline. Read the headline, then argue with the evidence.

Two worked examples

Both of these are the sample dossiers built into the product, unchanged.

ScribeAI, a seed pitch for an AI outbound SDR: 78/100
Critical, but not because the ROI math is wrong. The story rests on a moat claim (incumbent retaliation 90%, defensibility 88%) that HubSpot can invalidate with one product update. The verdict: high execution risk: commoditization trap & spam horizon deflation.
Nexus, a board memo to raise a marketplace take-rate from 15% to 28%: 86/100
Higher, and the crater is different: unit economics at 95%. The first axiom scores 10/10 because a 28% take exceeds the suppliers’ entire net margin. Same engine, Forensic CFO persona, and the fatal claim is about someone else’s P&L rather than your own moat.

An eight-point gap between the two is not a ranking of the businesses. It reflects that the Nexus memo puts more of its weight on a single claim that is arithmetically false, while the ScribeAI deck spreads its risk across a moat story that is merely unproven.

What it is not

Not a probability of failure. A great company with a lazy deck scores high.
Not a comparison to other startups. The benchmark line in the dossier is context, not a league table.
Not stable across rewrites. Change the axioms and the number moves; that is the point of the Audit Vault deltas.
Not a rating of the founder. It rates the argument in the text you submitted.

How to lower it before the meeting

Run the verification test attached to the highest-fragility axiom first. Replace the assertion with the result.
Where the steelmanned counter-thesis lands cleanly, adopt the rewrite rather than arguing with it.
Fix the crater axis. One number at 90 draws the whole conversation; four axes at 60 do not.
Rehearse the lethal question in the Hot Seat until Acknowledge, Pivot, Prove comes out in under thirty seconds.
Re-run the audit and compare in the Audit Vault. A falling index across iterations is the signal that preparation is working.

Do this in Dialectic

Paste a deck narrative or upload the PDF. The free benchmark mode returns an index and radar in seconds; the paid audit adds the full axiom breakdown, steelmanned rewrites, lethal questions, and the Hot Seat.