Acc. TRL-003 · Treatment protocol

Four steps. All of them written down.

Capture, rebuild, diff to under 1%, hand-off. The order never changes, which is rather the point.

  1. Capture

    I document the original: every real page, its words, its images, its SEO fields, and screenshots at desktop 1440 and phone 390. This is the "before" photograph — the reference everything else is measured against.

  2. Rebuild

    The site is rebuilt in native Keel: a custom theme and layout, real Keel parts, clean markup. Old scripts and plugins are quarantined and not carried over.

  3. Diff to under 1%

    New screenshots are compared to the originals, pixel by pixel, at both widths. I keep adjusting until the visual difference is under 1% — and note anything that's different on purpose.

    Target · 1440 & 390< 1% difference
  4. Hand-off

    You review it before anything is published. You get a written record: pages moved, fields filled, items quarantined, and any known differences. Then it goes live.

Why under 1% and not zero? Fonts render a hair differently across platforms. I'd rather tell you that than pretend otherwise.