Retros

3 retrospectives — 17 commitments — shipped
Jump to
All Commitments (17) — the rules we made Retro 1 — Mar 21 — Site v1 launch Retro 2 — Mar 21 — Print cards, domain pivot Retro 3 — Mar 22 — Evidence, dedication, ship

All Commitments

Every rule we made across all 3 retros. These are active — they govern how we build.

Discovery

  • Nail print specs up front — size, color mode, content checklist
  • Unit convention confirmed in discovery — dual metric+standard default
  • Theme colors confirmed in discovery — never assume from source
  • Ask for structure before building lists — tabs, groupings, count
  • Validate content additions with the user before implementing

Design

  • No emoji as UI — text elements only
  • Warnings placed near the action, not at top of pages
  • Center-align navigation elements by default
  • Scale SVG strokes with size — adjust proportionally
  • When icons fail once, pivot to text immediately

Process

  • Formula verified against actual daily usage before publishing
  • Always present deliverables as HTML — browser, not terminal
  • All structured data stored as triples
  • Cross-reference existing pages before building new content
  • Keep both print cards in sync from the start
  • Design print cards content-first — content, height, then fonts
  • Single find-replace pass for name/domain refactors
Retro 1
March 21, 2026 — Water of Life Site v1
6
Pages Built
2
Print Cards
11
Issues Found
8
Commitments

What Went Well

Demo-driven iteration produced a site that matches Joey's voice and vision — build, show, fix, show again.
Dark theme + tron blue landed clean — distinctive, fits "water," pure CSS variables, no framework.
Print cards are strong — B&W, scannable QR, bold type, 4×6 photo print, hand-to-someone simple.
Writing is Joey's voice — direct, personal, no-bullshit. Truth section hits hard. Nobel callout gives weight.
Caught 1L vs 1.5L error mid-demo by comparing to actual LMNT usage — site now reflects reality.

What Could Be Better

Too many rounds on print cards — wrong size, wrong colors, wrong layout, bad QR — 5+ iterations.
Specs not nailed in one conversation before building.
Used emoji icons for results and timeline, then had to rip them all out.
Should have gone text-only from day one on dark themes.
Metric vs imperial inconsistent until flagged mid-demo.
Unit convention not established during discovery.
Built gold, Joey wanted tron blue — wasted a pass.
Color assumed from source material, not confirmed.
Warning at top of page was too aggressive — took 2 tries to place before formula.
Warnings should be near the action, not blocking content.
11 issues logged during demo — most avoidable with better discovery questions.

Commitments

Nail print specs up front — size, color mode, content checklist before building.
No emoji as UI — text elements only. Codified in a-d-d spec.
Unit convention confirmed in discovery — dual metric + standard is default.
Theme colors confirmed in discovery — never assume from source material.
Warnings placed near the action, not at top of pages.
Formula verified against actual daily usage before publishing.
Always present deliverables to user as HTML — open in browser, not terminal markdown.
All structured data stored as triples — the board format is the universal data model.
Retro 2
March 21, 2026 — Print Cards, Domain Pivot, Content Updates
3
Icon Approaches
1
Domain Pivot
5
Commitments

What Went Well

Text-based labels (Hydration / Heart Function / Muscle Recovery) replaced failed SVG/PNG icons — cleaner, more readable.
Vertical stacked mineral rows work better than horizontal layout on both card formats.
Domain pivot from waterlife.supply to wateroflife.supply was clean — all files, DNS, QR, deploy config in one pass.
Tagalog integration — "Buhay na tubig" (Water of life) matches the domain name perfectly.
Print card design converged: benefit word big, mineral name small with underline, left-aligned centered ingredient block.
Eggs added to food sections — quick content update, no friction.
Voice updated to match Joey's authentic tone — rawer, more honest.

What Could Be Better

Too many iteration cycles on print card icons — SVG freehand, PNG via Pillow, back to SVG, then text.
Should have started with text. When icons fail once, pivot immediately.
Portrait card diverged from landscape approach during bullets phase.
Both cards should stay in sync from the start.
Spacing/overflow issues on both cards required multiple rounds of tweaking.
Design content-first — list all content, measure total height, then set font sizes.
Old waterlife.supply directory couldn't be renamed (device busy) — had to copy, leaving orphan.
Windows file locking. Copy-then-clean is the safer pattern.
Sari-sari cost estimate added then removed — wasted a cycle.
Content additions should be validated with the user before implementing.

Commitments

When icons fail once, pivot to text immediately — don't try 3 different rendering approaches.
Keep both print cards in sync from the start — changes to one reflected in the other immediately.
Design print cards content-first: list all content, measure total height, then set font sizes.
Validate content additions (like pricing) with the user before implementing.
Single find-replace pass for name/domain refactors — not file-by-file.
Retro 3
March 22, 2026 — Evidence Page, Dedication, Ship
80+
Evidence Entries
5
Nobel Prizes
6
Pages Updated
4
Commitments

What Went Well

Evidence page landed big — 80+ entries, 20+ traditions, 5 Nobel Prizes, tabbed UI with sub-nav. Zero to comprehensive in one session.
Dedication section has real weight — medal SVG reuse, narrative about Our Lady, Wikipedia link. Feels earned.
Back-to-top button — simple, site-wide, zero friction. Should have been there from day one.
Email routing done — [email protected] working via Cloudflare free tier. Two-minute setup.
CTA fix was clean — every page links to every other page. Caught via screenshot, fixed all 6 in one pass.
a-d-d-guidance v0.4.0 shipped — 50 lines of retro-driven improvements version-bumped and pushed.

What Could Be Better

Evidence page took multiple rebuilds — 25 entries, then 200, then tabs, then separation, then missing Nobel.
Should ask for scope and structure (tabs, groupings, count) before building.
Ohsumi's 2016 Nobel was missing — the most important one for the site's thesis. Already in story.html.
Did not cross-reference existing pages when building new content.
Medal SVG stroke too thin at scale — 0.5 at 80px looked ghostly. Fixed to 1.0.
SVG stroke-width must be adjusted proportionally when scaling.
Tabs left-aligned by default — user had to flag from screenshot.
Center-align should be default for nav elements on this site.

Commitments

Cross-reference existing pages before building new content — don't miss things already on the site.
Ask for structure before building lists — tabs, groupings, approximate count. Get the shape first.
Scale SVG strokes with size — stroke that works at 12pt doesn't work at 80px.
Center-align navigation elements by default — tabs, sub-navs, button groups.