← Shopify services

02 / Self-initiated workflow demonstration

Shopify Catalog &
Management

A repeatable preparation and review workflow for the spreadsheet, SKU and image work behind a collection launch.

Actual local validation / 29 October 2026

500 / 2,000

Synthetic products / variants checked.
Zero store writes. No client data.

The demonstration measures validation coverage, not sales or hours saved.

The brief

Find the gaps
while they are files.

A spreadsheet may look tidy while duplicate SKUs, missing prices or a wrong image mapping remain hidden. The objective was to make those exceptions visible before any store operation.

The implementation

One source.
A reviewable batch.

A local Python CLI normalizes supported CSV headers, checks product and variant relationships, and saves the original input alongside JSON, HTML and CSV findings. A separate image plan maps filenames, SKU, alt text and rights records.

Executed inputs and outputs

The report is part
of the deliverable.

Recorded CLI runs · Synthetic fixtures · 29 October 2026
InputCheckObserved result
500 products / 2,000 variantsDefault dry run0 errors; report and input backup; no import file
The same valid fixtureExplicit review exportDraft / unpublished CSV created locally
Three deliberate data defectsMissing price, duplicate SKU, wrong image mapping3 errors; export blocked
One generated test swatchImage plan, dimensions and source hashWebP copy exported; original unchanged

From manual checking to a repeatable review

Keep the decision.
Automate the checks.

  1. Receive

    Keep the original files and confirm scope and rights.
  2. Validate

    Run format, SKU, option, price and image checks.
  3. Resolve

    Review exceptions and approve a representative sample.
  4. Prepare

    Export a candidate batch; agree any store write separately.

The CLI has no Shopify API writer. Image URLs are checked for syntax and mapping, not fetched or certified as available. The sample uses reserved .test URLs and must not be imported as a real catalog.

What this proves.

  • Consistent checks across a 500-product fixture
  • Exception reports with row and field context
  • Separate image preparation and review
  • Source preservation and documented recovery steps

What stays a decision.

  • Product accuracy and image rights
  • Collection logic and merchandising choices
  • Live import permissions and store backup
  • Whether an existing product should change

Existing handles matched against a baseline block an import candidate. This demonstration does not establish production import or rollback acceptance.

Scope / Evidence / Limits

Inside the project

Project Overview

An offline catalog preparation system demonstrated with synthetic data. No customer information or real store writes are used.

Creative Direction

A clear exception report and an editorial case narrative make the work understandable to an operator and a buyer.

Design System

Current Shopify CSV field names, normalized aliases, consistent report columns and SKU-based image naming. Unknown fields are reported rather than silently dropped.

Desktop & Mobile Experience

The web case presents key results in readable sections and a horizontally scrollable report table. CLI files remain available for spreadsheet review.

Shopify Product / Catalog Setup

500 products and 2,000 variants in the executed fixture; price, SKU, option relationships and gallery mapping checked. Explicit export forces draft and unpublished status.

Workflow & Technical Implementation

Python CLI; default dry run; immutable input copy and SHA-256; optional baseline diff; image processing reuses the existing offline asset engine.

Quality Assurance

Three injected data defects blocked export. An actual image copy was produced without changing its original. URL availability, visual accuracy, rights and production import remain separate human checks.

Deliverables

CLI, README, deterministic sample generator, automated tests, HTML/JSON findings, review CSV, image plan and recovery guidance. No measured labor saving or conversion result is claimed.

Have a catalog to prepare?

Start with a representative sample and an agreed list of checks.

Discuss the batch ↗