WHAT WENT WRONG, IN WRITING

The write-ups of your worst engagements are your most valuable documents

Success gets summarised. Failure gets examined. A review of something that went badly traces cause, decision and consequence with a candour that nothing else the firm produces comes close to.

Check your data Six questions, roughly two minutes.
Candour that success never producesCause, decision and consequence linkedAnonymized before anything moves

THE UNCOMFORTABLE QUESTIONS

What firms ask about failure records

These name people and mistakes. That is not going anywhere.

Identifying fields are stripped as part of preparing the data, and anything you will not license simply is not in scope. That boundary is set before anything moves.

Could this be used against us?

The data is anonymized and built into general data sets for AI labs. Nothing is listed anywhere and we do not disclose which labs take what.

Ours are informal, just email threads.

That is common and frequently better. Informal reviews are more candid than formal ones, and candour is the valuable part.

Is failure really worth more than success?

Often. What carries value is visible reasoning, and people reason far more explicitly about things that went wrong.

WHY FAILURE ANALYSIS RATES WELL

Four properties that make a post-mortem unusually rich

  • Causation is traced. Someone worked backwards from an outcome to the decision that produced it, which is exactly the hard reasoning.
  • The counterfactual is stated. Good reviews say what should have been done instead, which is rarer than describing what was.
  • Honesty is unusually high. These are written internally, for people who already know it went wrong.
  • Corrections follow. A post-mortem usually changes a procedure, and the change is traceable to the reasoning.

THE SENSITIVITY QUESTION

This is the material firms guard most, so scope comes first

  1. 01

    Establish what exists

    Six questions about the systems in use. Nothing is accessed at this stage.

  2. 02

    You set the boundary

    Reviews naming individuals or matters you will not license stay outside the licence, agreed before signature.

  3. 03

    Anonymization in preparation

    Identifying fields are stripped as part of preparing the data, ahead of any onward sharing.

  4. 04

    Paid on completion

    Typically $100K to $2M, Net 30 to 60, once the data is approved, shared, and anonymized.

WHERE THESE LIVE

Four places a firm records what it got wrong

Most firms produce these somewhere, even informally, and then never read them again.

Engagement reviews

Looking back at a piece of work that overran, underdelivered, or ended badly, and asking why.

Incident write-ups

A specific failure reconstructed in sequence, with the decision points identified.

Lost-client analysis

Why a relationship ended, examined internally rather than explained outward.

Retrospectives

Team discussion of what to do differently, which surfaces disagreement about the cause.

THE TRACK RECORD

Operators who have had things go wrong

“Built Roofer.com from nothing in a fragmented trade, raised $7.5M across venture and private equity, and ran the P&L throughout.”
Nathan MathewsPartner
“Eight acquisitions closed across the group, $100M+ in revenue across businesses built with partners.”
PolysharesAcross the group

SPEAK WITH A MANAGING PARTNER

The write-ups nobody wanted to produce have a value

Six questions and a short conversation. Nothing you tell us leaves Polyshares.

Check your data