Free 20-point conversion & SEO audit for export stores — Free website audit — claim yours

hello@woo.one Mon–Fri · 9:00–18:00 (UTC+8)

Catalogue Cleanup Before an OpenCart Migration

Zombie SKUs, duplicate products, orphan images and free-text attributes. What to delete before exporting, and why the same work costs weeks instead of days after launch.

The most valuable work in a migration often happens before the import: deleting things. A catalogue that has been running for five years on any platform contains products nobody should migrate, and bringing them along makes every later step harder.

This is the cleanup pass we run on OpenCart stores before export, in order of return on effort.

Zombie SKUs: disabled but still indexed

Every mature store has them — products disabled years ago that still have URLs, still appear in the sitemap, and sometimes still have backlinks.

They are worth real money if handled deliberately and worth real harm if ignored:

  • If the page has traffic or links, migrate it as a live product or 301 it to a close equivalent
  • If it has neither, delete it and return 410 — not 404, not a redirect to home
  • If it has neither but there are fifty variants of the same dead model, delete once and be done

The harm comes from migrating them as “disabled products” and then letting a plugin generate URLs for them anyway. You inherit a hundred thin pages on a brand-new site, and the first thing search engines learn about the new domain is that it has thin pages.

Duplicate products: the same item, five SKUs

Duplicates accumulate through manual entry, supplier feeds, and re-listing items that went out of stock. Tell them apart before deciding:

Type How to detect Action
True duplicates Same SKU, different IDs Keep the one with sales history; 301 the rest
Near duplicates Same title, slightly different SKU Merge; assign the surviving SKU; keep the other as a search alias
Legitimate variants Different options, same base Convert to variations during import, not separate products
Regional editions Same product, different market Keep separate, but fix the titles so they are distinguishable

The “legitimate variants” row is where migrations create the biggest mess. OpenCart stores that sold a t-shirt in five colours as five products often end up with five products and five separate pages in WooCommerce, when they should be one product with five variations. Merging after import is far more painful than deciding before.

Orphan images

OpenCart’s image table accumulates files that no product references — deleted products, replaced images, failed uploads. A five-year-old store can carry tens of thousands of them.

Why it matters beyond disk space: image migrations usually fail quietly on orphans, producing broken references or duplicated files with hashed names. Clean the reference graph before exporting:

  1. List all image paths referenced by active products and categories
  2. List all files in the image directory
  3. The difference is orphans — move them aside, do not delete yet
  4. Export only referenced files

Keep the orphan folder until two weeks after launch. Something always turns out to be referenced from an old invoice template or a supplier document.

Attributes that are really categories

A common pattern: years of “attributes” that describe product type — “Material”, “Style”, “Collection” — with hundreds of free-text values, half of them misspelled.

When this data moves to WooCommerce as attributes, it becomes variation data or filter data, and free-text values destroy both. Two hundred values of “colour” that include “Black”, “black”, “BLK” and “Noir” produce a filter with four black options.

Normalise before import, not after. Build a mapping table: raw value → canonical value. It takes a few hours and saves weeks of filter maintenance.

Categories that are not categories

Stores that never pruned their category tree end up with paths like Home > Sale > Clearance > 2020 > Summer, where the top three levels mean nothing and the last two are dead.

Before migrating, collapse the tree to something a customer can navigate. A useful test: can you describe the tree in one sentence? If not, it is too deep. In WooCommerce, deeply nested categories also produce long URLs and complicate breadcrumbs and filters.

Descriptions that are supplier boilerplate

If 400 products share the same 150-word manufacturer description, they are near-duplicate pages. Migrating them as-is moves the duplication to a new domain.

You do not need to rewrite 400 descriptions before launch. But you should:

  • Identify the products with the most traffic and revenue, and improve those first
  • Remove obvious boilerplate from the rest, even if it means a shorter description
  • Never pad with keyword-stuffed filler — that is what makes thin content worse

Customer records nobody should keep

Under GDPR and similar regimes, you are not obligated to carry every record forward, and carrying junk creates obligations. Reasonable rules:

  • Keep customers with at least one order in the last 24 months
  • Keep anyone with an active subscription or open balance
  • Drop accounts with zero orders older than three years
  • Anonymise rather than delete if you need the order history for reporting

Document the rule you applied. If a regulator or a customer asks why an account is gone, “we deleted it during migration per our retention policy” is a defensible answer; “we are not sure” is not.

What cleanup buys you

The point is not tidiness. It is that every one of these decisions becomes expensive after the import: merging duplicate products that already have URLs, fixing attribute filters that already have traffic, redirecting dead pages that already re-indexed.

Cleanup is a day or two of decisions before export. The same work after launch is weeks, and it is interleaved with bug reports from customers.

The verification checklist in our pre-launch guide assumes this pass happened — it compares post-import counts against a baseline, and a baseline built from a dirty catalogue will hide real problems. If you are planning a migration from OpenCart, the OpenCart side is where we usually start, and the first deliverable is a cleanup report rather than an import file.

S

Infrastructure and security engineer at woo.one, based in Foshan, China. Works on WordPress, WooCommerce and OpenCart systems, self-hosted mail, and the operational side of running stores for exporters.

More about the team

No pressure · No sales script

Ready for a store that sells while you sleep?

Send us your current site or product list. You will receive a fixed-scope proposal, a 20-point audit and a realistic timeline within two business days.

Reply within 1 business day NDA on request 100% code ownership
Chat with us
WhatsApp Get a quote