Bulk editing workflows - Matrixify, CSV, and the admin bulk editor
Phase 2 taught the safety habits - export first, small batches, know your blast radius. This phase is the operational sequel: running data changes at real scale as routine work. First, the toolbox, because "bulk editing on Shopify" is three different tiers of tool, and matching tier to task is most of the efficiency.
The three tiers
The admin bulk editor (select products → Edit): spreadsheet-style editing of core fields directly in the admin. Right for: quick edits to dozens of products, standard fields, one-off corrections. Its ceilings arrive fast - limited field coverage (metafields support is partial), no import/export, no repeatability. Treat it as a power tool for small jobs, not a workflow.
Native CSV export/import: Shopify's built-in product CSV covers core fields and gets you a proper export-edit-import loop. Its historical blind spot - metafields - has narrowed (newer import flows carry metafield columns), but its structural quirks remain: the multi-row-per-product format (one row per variant, images stacked) that makes spreadsheet formulas fragile, and its overwrite semantics that punish column-mapping carelessness. Right for: full-catalog field updates where its format fits and no tooling budget exists.
Matrixify (the de facto standard for serious Shopify data ops): everything - products, variants, metafields, collections, translations, redirects, customers - in a saner one-row-per-entity format, with the operational features that turn edits into workflows: partial exports (only the fields you're changing - which shrinks blast radius by construction), scheduled and repeating jobs, dry-run validation, background processing for six-figure row counts, and export-of-changes for auditability. If your catalog passed a few thousand products or your team runs recurring data jobs, this tier is where you live.
Workflow patterns that scale
The patterns that separate routine from risk, layered on Phase 2's habits: the surgical export - export only ID/handle plus the columns you're touching; a file that can't overwrite what it doesn't contain is safer than discipline (Matrixify's column selection makes this default behavior; with native CSV, delete columns before editing). The pipeline spreadsheet - for recurring jobs (seasonal price updates, supplier data merges), a maintained sheet with formulas that transform input → import format, so the job is "paste new data, export, import," documented and repeatable, not re-invented quarterly. The staged rollout - Phase 2's small-batch principle, formalized: 10 → validate → 10% → validate → rest; at operations scale, validation means checking rendered pages and downstream surfaces (a collection, the feed preview), not just admin fields. The change log - one shared document: date, what changed, file link, who ran it. Thirty seconds per job; it's the difference between "what happened to the descriptions in March?" being answerable or archaeology. That log is also this phase's connective tissue - the QA, versioning, and workflow lessons ahead all build on changes being legible.