How to bulk-update 100,000 Shopify prices without breaking your store (2026)

Shopify does not have a built-in undo for bulk edits, which makes a bulk reprice the highest-stakes routine job in ecommerce. Done right, it is invisible. Done wrong, it sells your bestseller at cost for a weekend, or prices a whole collection at €0.00 because a spreadsheet column shifted by one. The bigger the catalog, the worse the math: at 100,000 variants, a 1% error rate is a thousand wrong prices, live, at checkout.

The four standard options

• The admin bulk editor. Fine for a handful of products. It is manual, it paginates, and there is no way to express a rule: you type the new prices in, one cell at a time.
• CSV export–edit–import. The workhorse, and the most dangerous. The spreadsheet is the logic, the import is instant, there is no preview, the import can't be undone, and a malformed CSV silently drops rows. Most catalog horror stories start with "the import looked fine".
• Bulk-editing apps. Faster and safer than raw CSVs, and the good ones have scheduling. The rule language is a form, though: when your pricing logic needs cost data, currency rounding, or an exception list, you hit the edge of what the form can say.
• A custom script against the Admin API. Fully expressive, and now you own rate limits, retries, pagination, partial-failure recovery, and a run log. A script that worked in April fails silently in October because someone renamed a tag.

What a safe bulk reprice needs

Whatever tool you use, the checklist is the same. Compute against source data (cost, compare-at, currency), never against yesterday's output. Guard with floors and ceilings: a margin floor turns a data error into a skipped row instead of a loss-making price. Handle the exceptions explicitly: missing costs, gift cards, bundles. Rate-limit like you mean it, because a naive loop at catalog scale gets throttled halfway and leaves the store half-repriced. And above all: see the diff before the store does.

The rehearsal, not the leap

That last requirement (see the diff first) is the one almost no tool provides. It is the difference between a nervous reprice and a boring one. The pattern to look for is a dry run: the job executes against your real catalog, every write is captured and held, and you read a mutation report: 862 price updates, 37 rows skipped for missing costs, here is every single one. Nothing has touched the store yet. If the report is wrong, you fix the logic and rehearse again. If it is right, you approve, and the exact run you read is what goes live.

This is how Collaudo runs every job, including the repricing templates below: describe the rule in plain English ("round every price ending in .00 down to .99, floor at 18% margin, skip anything missing a cost"), read the generated code if you want to, dry-run it on the full catalog, approve the report. The template gallery has the common pricing jobs ready to rehearse on your store today.