Most business and tax advice is a claim. We show the decision instead.

What this is

An advisory practice for business building and tax strategy

Proof

Real decisions, real tradeoffs

Four worked examples. Each one is a decision that could have gone the easier way and didn't, or a mistake that got corrected in the open instead of quietly.

• 01

  An audit whose job was to disprove itself

  Built a tax-calculation system checked line by line against real statute, then ran a second, adversarial pass whose only job was to try to disprove the first pass's own findings. Several were walked back — not just fixed and forgotten, but explicitly reclassified once closer scrutiny showed the code path was actually unreachable.

  Reclassified findings Counted & moved on

  The tradeoff A second audit pass costs real time and finds fewer new defects than the first — most of its value is catching your own overconfidence, which is a harder thing to budget for than finding bugs.

  Read the full write-up →
• 02

  The audience that drifted to 150 million people

  An ad campaign was found reaching a far wider audience than intended, at a cost per click the underlying economics didn't support, because nothing enforced the targeting spec it was supposed to run against. The fix wasn't a one-time manual correction — it was a machine-checked spec plus an automated guard so the same drift can't happen silently again.

  Enforced spec + guard One-time manual fix

  The tradeoff Enforcing a spec in code is slower to set up than eyeballing a settings screen — it only pays for itself the second time someone would have made the same mistake.

  Read the full write-up →
• 03

  Real books instead of a slicker dashboard

  Faced with a build-vs-buy call for a client-facing product, the decision was to build actual accounting records rather than a nicer-looking dashboard on top of someone else's numbers — on the thesis that the real moat is switching cost, not feature count. The hardest integration in that plan was deliberately sequenced last, not first.

  Real accounting records A dashboard on top

  The tradeoff Real books are slower to ship and harder to demo than a dashboard, and they're the one thing you can't fake with a mockup — wrong books are worse than no books, so the risk sits at the center of the plan instead of the edge of it.

  Read the full write-up →
• 04

  A 9x cost model, corrected and republished

  An internal resourcing decision was built on a cost model that turned out to be wrong by roughly nine times. Rather than quietly patch the number and move on, the correction — and the fact that the original estimate was wrong — was written down and republished alongside the revised conclusion.

  Published the correction Quietly patched it

  The tradeoff Publishing your own miss costs more short-term credibility than staying quiet about it. It's worth it because the alternative is a track record nobody can actually check.

Writing

Where the reasoning is visible at length

The proof points above, worked through in full — the alternative considered, the tradeoff accepted, and what would have to change for the answer to flip.