Why startups fail: five signals a decision engine scores early

Why do startups fail, and which five signals does a decision engine score?

The EQOS Asia startups study reports that 90% of startups fail within 10 years and that the rate has not changed in 20 years (EQOS Asia startups case page, as of October 2026). It draws on PitchBook, CB Insights, Carta, Y Combinator, and Crunchbase data. A rate that stays flat points to causes that repeat from one cohort to the next.

A decision engine groups the repeating causes into 5 signals: unit economics, runway, founder-market fit, governance, and team health. Each signal maps to one of the 9 dimensions, so the engine can place a number on it. The figure lists the five in the order a reviewer usually checks them, from the cheapest question to the deepest.

How do unit economics show why startups fail before a round?

The study scores clarity at 0.362 and notes that unit economics remain blurry for most startups. It recommends real, verified figures over at least 6 months, and a real LTV to CAC ratio above 3, in place of a ratio projected to maturity (EQOS Asia startups case page, as of October 2026). Clarity could then rise from 0.362 to 0.50.

The check is practical. Ask for cohort data by month, the cost of acquiring each customer, and the margin after delivery. Projections of 5 years rest on growth assumptions, and measured months rest on events. A founder who shares 6 months of clean cohort data gives an investor a signal that no pitch deck can match.

What does the runway signal say about why startups fail?

The study points to cycles of 18 to 24 months between rounds and scores stability at 0.362. Its rule is direct: a runway under 18 months without a credible path to the next raise or to profitability is a red flag (EQOS Asia startups case page, as of October 2026). Burn must be justified by real growth, and speculative hiring does not qualify.

The funding climate raises the stakes. The study reports a loss of 40% of venture volume between the 2021 peak and 2025, with $300 billion of capital waiting to be deployed, and a more selective investor base. A team that knows its runway in months and its burn per growth point can plan the next round from a position of choice.

Why does founder-market fit matter in why startups fail?

Founder-market fit is the alignment between the vision of the founder and what the market confirms. The study scores coherence at 0.425 and reads the growth at all costs narrative as a persistent source of misalignment, even after visible failures such as WeWork and FTX (EQOS Asia startups case page, as of October 2026).

The recommendation is to audit the alignment between founder, team, and market, and the study sets a target of 0.55 for coherence. A simple test helps: list the 3 customer facts that confirm the vision, with dates and names. When the list is short, the founder has a clear next experiment, and when it is long, the team has material for a sharper pitch.

Which governance and metric signals show why startups fail?

The study scores integrity at 0.530 and communication at 0.407. After WeWork and FTX, investors ask for more governance, more transparency, and standard metrics, and integrity improves slowly (EQOS Asia startups case page, as of October 2026). Integrity is the strongest of the 9 dimensions after energy, so it offers a base to build on.

Metrics are better standardised today, with ARR, NRR, and CAC to LTV in common use. The study warns that metric gaming remains frequent, for example biased cohort analysis or non-recurring revenue counted as ARR. The check is to ask how each metric is computed and to request the raw table behind it. A clear definition takes minutes to write down.

How does team health show why startups fail in a decision score?

The study scores care at 0.295, the lowest of the 9 dimensions, and cites a UC Berkeley study in which 72% of founders burned out (EQOS Asia startups case page, as of October 2026). It also notes mass layoffs and a hustle culture that is increasingly recognised as costly. Care is the dimension where the sector has the most room to improve.

Team health appears in data a reviewer can request: turnover by quarter, average tenure, and founder workload. A decision engine treats a falling trend as a signal for a conversation, and a stable trend as support for the plan. The check supports both investors and founders, since a rested team executes the other four signals more reliably.

How do the five signals combine into one startup decision score?

The study aggregates the 9 dimensions with a weighted geometric product and returns φ(S) = 0.333, labelled fragile, with 1 of 9 dimensions in the critical zone (EQOS Asia startups case page, as of October 2026). Energy stands highest at 0.580, driven by innovation, with 75% of 2025 deals in AI. The weakest dimension weighs as much as the strongest.

That is why a startup with strong innovation can still score low. The study advises measuring φ(S) before investing or scaling, and fixing fundamentals first when the score falls below 0.4. The table sets the five signals beside their dimensions and current scores, so a team can see which one to repair first.

How can you score the five startup signals on your own case?

Rate each signal from 1 to 10 with one line of evidence, then enter the ratings in the free simulator to read a first φ(S) estimate in about 10 minutes (EQOS Asia simulator page, as of October 2026). Repeat the exercise each quarter and note which dimension moved. A moving dimension shows where effort pays.

For a deeper read, the pricing page lists the available tiers with their dates, and the method page explains the Audit pillar and its 3 priority risks. The startups study is a prospective analysis built on public sector data. This page is general information, not legal, tax or investment advice.

Frequently asked questions

About the author

Stéphane Jambu

President, EQOS Forecast Asia

Stéphane Jambu leads EQOS Forecast Asia and works as a consultant in e-commerce, search visibility, and AI. LinkedIn profile