Merge Shopify CSV Exports

Operations team has 8 suppliers each generating their own audit CSV via the Vendor filter. All exports share the standard Shopify Products schema — strict mode catches any accidental schema drift.

Inputs (all from Catalog → Products → Vendor filter → Export):
  acme-products-2026-05-15.csv     (1,200 rows)
  bestbrand-products-2026-05-15.csv (980 rows)
  ... 6 more files

Config: mode: strict

Output: all-suppliers-merged.csv (~9,500 rows)
  Headers identical across all files; variant rows preserved;
  same Handle within a file = variant; same Handle across
  files = genuine cross-Vendor conflict (run csv-duplicate-
  finder on Handle column to surface for review).

Year-end accounting handoff: 12 monthly Orders exports → 1 annual file. Same Shopify store + same export type = schema is stable. Strict mode.

Inputs:
  orders-2025-01.csv
  orders-2025-02.csv
  ...
  orders-2025-12.csv

Config: mode: strict

Output: orders-2025-annual.csv
  Concatenated in file order (Jan first, Dec last).
  Multi-line orders preserved — orders with N items appear as
  N rows with subsequent rows leaving non-item fields blank,
  same as in the source files.

Shopify added a new metafield column to Products mid-year. Q1 export has 32 columns; Q4 export has 33. Strict mode rejects the merge. Union mode produces a 33-column output where Q1 rows have empty cells in the new column.

Inputs:
  products-q1-2026.csv  (32 columns)
  products-q4-2026.csv  (33 columns — one new metafield)

Strict mode: error — header mismatch
Union mode: succeeds → 33 columns; Q1 rows have empty cell
  in the new column.

Recommendation: add a 'source_export_date' column manually
to each file before merging so empty cells in the new column
are interpretable as 'field didn't exist yet' vs 'no value'.

A single product with 150 variants triggers Shopify's emailed multi-file ZIP. Extract the ZIP, drop all 4 CSVs onto the merger.

Inputs (extracted from Shopify's emailed .zip):
  products_export_1.csv  (~50 rows — the high-variant product
                          plus other products in the batch)
  products_export_2.csv
  products_export_3.csv
  products_export_4.csv

Config: mode: strict (Shopify guarantees identical schemas)

Output: single merged CSV ready for re-import / analysis.

This is the anti-pattern strict mode catches. Products export has columns like Handle, Title, Vendor, Variant Price. Customers export has First Name, Last Name, Email, Total Spent. The two should NEVER merge — they're different entity types.

Attempt:
  products.csv  (Handle, Title, Vendor, ...)
  customers.csv (First Name, Email, Total Spent, ...)

Strict mode: error — header mismatch (the merger refuses
to silently merge incompatible schemas)

Union mode: succeeds but produces nonsense — a 50-column
mess with each row populated for either Products columns
or Customers columns but not both. Useless downstream.

Fix: keep Products and Customers as separate merged files,
one per entity type.

Shopify's documented error wording for product exports that contain too many variants to generate in the standard timeout window. Fix per Shopify's docs: 'split your export into multiple smaller files by filtering your products in the Shopify admin before exporting'. The merger then combines the batches back into one file.

Shopify documents a 15 MB per-file limit for product, inventory, and customer CSV exports. Larger catalogs must be batched at export time. Same applies on the import side — if your merged file exceeds 15 MB and you intend to re-import, split with csv-row-splitter before upload.

Quoted from Shopify's docs: 'If any single product has more than 100 variants, then the CSV file is emailed to you.' The emailed file is a multi-file ZIP containing 2–4 CSVs that share the same Products schema. Extract the ZIP, drop all the CSVs onto the merger in strict mode.

Shopify products with multiple variants produce multiple rows in the export, all sharing the same Handle. The first row carries Title, Body (HTML), Vendor, etc.; subsequent rows leave those blank and only populate variant-specific columns. The merger does NOT alter this — variant rows are preserved verbatim. Critical: don't run a Handle-keyed dedupe after merging or you'll destroy variants. The clean-shopify-csv-workflow handles this correctly by using csv-duplicate-finder (which marks but doesn't delete) instead of csv-deduplicator.

Products and Customers have completely different schemas. Strict mode catches the mistake immediately. Union mode would succeed but produce a 50-column file with each row populated for only one entity type — useless. Keep export types in separate merged files.

Shopify's Orders page has two export buttons: Export orders (the order-line-item view) and Export transaction histories (per-transaction financial view). These produce different CSVs with different schemas. The merger's strict mode catches the difference; don't try to merge them together — keep one merged Orders file and one merged Transaction file.

Shopify occasionally adds new columns to standard exports as new features ship (metafields, new lifecycle stages, etc.). A Q1 export and a Q4 export from the same store may have a 1-column difference. Strict mode rejects; union mode produces a unified file with empty cells in the new column for older-batch rows. Add a source_export_date column to each input file before merging so the empty cells can be interpreted as 'field didn't exist yet'.

If you accidentally merge exports from two different Shopify stores (e.g. a UK store and a US store with overlapping product catalogues), the same Handle from each store will merge as if it were one product with extra variant rows — silently corrupting both products' data. The merger doesn't detect cross-store mixing. Run csv-duplicate-finder on the Handle column after merging to surface duplicates for manual review.