Ashish Narayan

The research lens on a retail-academy website rebuild

The archive before the redesign

Nothing is invented.

  1. Nothing is invented
    01Nothing is invented — The rule that decided what the first deliverable would be, before any design existed.Full size ↗
  2. The shape of the engagement
    02The shape of the engagement — Relationship to the work, and the dates version control can actually support.Full size ↗
  3. The first deliverable was an archive
    03The first deliverable was an archive — Why a rebuild starts with a transcription of the thing being replaced.Full size ↗
  4. Four rules, set before the work began
    04Four rules, set before the work began — Each one removes a specific way a rebuild lies about its subject.Full size ↗
  5. Fourteen places the source disagreed with itself
    05Fourteen places the source disagreed with itself — Every one needed a decision, a default, and a reason written down.Full size ↗
  6. When the only honest answer is an inference
    06When the only honest answer is an inference — Two listings published a salary with no period. The reasoning is kept next to the result.Full size ↗
  7. Thirty gaps, numbered and owned
    07Thirty gaps, numbered and owned — What the research could not find becomes a register, not a silence.Full size ↗
  8. Twelve recorded, four published
    08Twelve recorded, four published — The archive captured what students said about. It did not capture what they said.Full size ↗
  9. The retouching line, and the test that drew it
    09The retouching line, and the test that drew it — A generative pass was tried on two photographs. It returned people who were not the same people.Full size ↗
  10. Translation as a research problem
    10Translation as a research problem — 1,664 of 1,664 strings, and the 42 that stay English on purpose.Full size ↗
  11. Where the archive itself was wrong
    11Where the archive itself was wrong — The research corrected its own record twice. That is the part most write-ups delete.Full size ↗
  12. Two things nobody asked the research to find
    12Two things nobody asked the research to find — A transcription pass over a whole site turns up what an audit was never scoped for.Full size ↗
  13. What this artifact does not establish
    13What this artifact does not establish — The stated-limits section. Every artifact in this set ends with one.Full size ↗

The whole rebuild — content archaeology, information architecture, the design system, the bilingual architecture and the engineering. Executed. Counts are quoted from the repository's handover document, each of which names the command that regenerates it. No outcome is claimed, because the site has not been deployed.

Back to the written studyAll case studies

B-SIDEOff the record