base studio
base studio / products / solo founder playbook
DATA SET · LIVE claude skill · production editorial series · in publication

the playbook for founders who'd rather not die.

We watched 101 founder interviews. 49.5% openly discussed failure. We coded every failure mode into a structured data set, packaged it as a Claude skill you can install, and wrote the editorial series no one wants to write.

49.5%
of 101 founders openly discussed failure on camera.
the other 50.5% are why you've never seen this data before.
Multiple failures
before success
15
Financial crisis
13
Burnout & isolation
12
Building without validation
11
Refusing to pivot
10
No safety net
8
Shiny object syndrome
7
01 what it is

Three surfaces, one data set.

Most founder content is "things I wish I knew" rewrites. This is the inverse: we transcribed and tagged 101 long-form founder interviews on Starter Story, extracted every failure-related claim, and structured them into a taxonomy. Then we shipped that taxonomy three ways — as a Claude skill, as a public data set, and as an editorial series.

surface 01

The Claude skill

Install solo-failures in Claude Desktop and ask it to diagnose your situation, pre-mortem your plan, or post-mortem a failure. It ranks risks by data, not vibes.

surface 02

The structured data

Failure modes, frequencies, root causes, anti-patterns. Open patterns.json on GitHub. Use it for your own research, dashboards, or content.

surface 03

The editorial series

One essay per failure mode. The first card pack — "Why startups die · 7 hard truths" — is out as a Xiaohongshu image series and long-form post.

02 the seven modes

Ranked by frequency. Not by drama.

Every claim below is grounded in a count from the 101-video corpus. Click through to the essay (where shipped) for the kill mechanism, early warning signs, and the cheapest intervention.

01
Multiple failures before success
failure is the median, not the exception.
15/ 101 videos
02
Financial crisis & near-bankruptcy
runway math kills more startups than any single bad decision.
13/ 101 videos
03
Burnout from overwork & isolation
the founder breaks before the business does.
12/ 101 videos
04
Building without customer validation
months or years of building, zero users at the door.
11/ 101 videos
05
Refusing to pivot after initial failure
sunk cost + identity = late pivots = no pivots.
10/ 101 videos
06
No safety net (quit too early)
financial pressure compresses your time horizon into a vise.
8/ 101 videos
07
Shiny object syndrome
new idea every 4 weeks = compounding reset to zero.
7/ 101 videos
03 use the skill

Diagnose. Pre-mortem. Post-mortem.

The skill takes your current situation (in-flight crisis / pre-launch plan / post-failure recap) and matches it against the failure-mode corpus. It returns the top 1–3 failure modes that fit, the chain of how this one actually kills you, the early warning signs, what the 101 founders did to survive, and one concrete next move for the next 48 hours.

"I've been building a B2B SaaS in stealth for 8 months. $15k savings left, no users, planning Product Hunt next month. 14-hour days, no day off in 6 weeks."

→ matches: building-without-validation (11) · burnout (12) · no-safety-net (8). Concrete next move: get one paying customer this week or stop coding.
04 the data

Method, plainly.

Corpus 101 long-form Starter Story founder interviews · transcript + annotation
Failure rate 49.5% of videos discussed failure openly (50 / 101) — base rate for the taxonomy
Coding scheme Categories: Challenges/FailuresValidationCost/ExpensesAdvice/Lessons
Top 7 modes By video-count: 15 / 13 / 12 / 11 / 10 / 8 / 7 — full counts in patterns.json
Skill format Claude skill (Markdown spec + JSON knowledge base) · installable in Claude Desktop & Claude Code
License Data: CC-BY 4.0 · Skill: MIT · Editorial series: all rights reserved
Source github.com/yayashuxue/solo-failures

Get the next essay before you need it.

One mail per failure mode. No drip campaign.

sign up for email list