Are there any universal patterns across the case studies?

Yes. All 5 case studies share: (1) the project was 90%+ complete, (2) the creator was the only one holding it, (3) the creator had a list of 'small fixes' that grew into a death-by-a-thousand-cuts, (4) there was no public ship date. The cure for all 5 patterns is the same: ship anyway, on a fixed date, even if the small fixes aren't done.

Blog · Case studies

The 90% rule in practice

Q: Why is the polish phase so deadly?

Three reasons: (1) the creative work is done, so the original motivation is depleted; (2) the imperfections are all visible at once, creating paralysis; (3) the audience becomes real, creating anxiety. The polish phase has none of the momentum of the building phase and all of the stress of the shipping phase.

Q: What can I do to avoid the 90% trap?

The 'ship the 90% version' rule. Bad-and-shipped beats good-and-unpublished. Pick a ship date, announce it, and ship on the date regardless of readiness. The date is a forcing function.

5 case studies of real projects that died at 90% complete. A SaaS that almost launched. A novel that almost shipped. A renovation that almost finished. The 90% rule is the most common death pattern for any project. Here's what it looks like in practice.

Published July 13, 2026 · ~4 min read

Why case studies

The 90% rule is a pattern, but patterns are easier to see in specific cases. These 5 case studies are composites of real projects I've advised on or known about. Names changed, details generalized, but the patterns are real. If your project matches one of these 5, you have a 90% trap on your hands. The cure is the same in all 5 cases: ship anyway, on a fixed date, even if the small fixes aren't done.

The 5 case studies

Case 1: The SaaS that almost launched

A solo founder spent 18 months building a B2B SaaS. 95% complete. The last 5% was "polish": onboarding flow, empty states, error messages, the marketing site. The polish list grew. The founder picked a launch date, missed it, picked another, missed it. 6 months later, the SaaS was still 95% complete. A competitor launched a worse version with worse UX. The market moved on. The SaaS is now dormant on the founder's machine.

The pattern: the polish list was a procrastination device. The empty states didn't matter. The error messages could have been "Something went wrong."

Case 2: The novel that almost shipped

A novelist spent 4 years on a literary novel. 90% complete. The last 10% was the second half of chapter 22 and a final revision pass. The novelist kept "polishing" the first 21 chapters — adjusting dialogue, refining prose, fixing the timeline. 3 years later, the novel was 90% complete and the first 21 chapters had been revised 14 times. The second half of chapter 22 was still unwritten.

The pattern: backward edits. The novelist kept polishing the beginning instead of writing the end.

Case 3: The renovation that almost finished

A homeowner spent 2 years on a kitchen renovation. 90% complete. The last 10% was the backsplash tile, the cabinet hardware, the final paint touch-up. The homeowner kept "improving" — adding crown molding, swapping the cabinet pulls, repainting a wall. The renovation was never "done" because the polish list kept growing.

The pattern: scope creep disguised as polish. The crown molding wasn't in the original plan.

Case 4: The album that almost released

A musician spent 3 years on an album. 90% complete. The last 10% was mixing, mastering, and the album art. The musician kept "tweaking" — re-recording vocals, adjusting the mix, adding a final overdub. The album was 90% complete for 2 years. The musician's style evolved; the album's style stayed in 2024. The musician couldn't ship the album that was no longer theirs.

The pattern: identity shift. The musician had outgrown the album.

Case 5: The startup that almost raised

A founder had an MVP, 10 customers, and a pitch deck. 90% ready to raise a seed round. The last 10% was a better deck, a better demo, a better metric, a better narrative. The founder kept iterating. 6 months later, the runway was running out and the deck was still 90% ready. A competitor raised on a worse deck with worse metrics. The market moved on. The startup is now dormant.

The pattern: perfectionism in the face of a moving market. The deck was good enough 6 months ago.

The 4 patterns they all share

90%+ complete, but not shipped. All 5 had the bulk of the work done.
Solo creator. No co-founder, no accountability partner, no public ship date.
Growing polish list. The list of "small things" was a procrastination device.
No forcing function. No external deadline, no public commitment, no ship date.

The cure for all 4: a public ship date. The date is the forcing function. The polish list can be addressed after shipping. The 5 case studies are 5 examples of what happens without a date.

Frequently asked questions

What is the 90% rule?

The pattern where 70-90% of projects die in the last 10%. The polish phase is where they die.

Why is the polish phase so deadly?

Motivation is depleted, imperfections are all visible, audience becomes real. None of the building-phase momentum, all of the shipping-phase stress.

What can I do to avoid the 90% trap?

Ship the 90% version. Bad-and-shipped beats good-and-unpublished. Fixed date, ship on the date.