A practical audit of product descriptions across a changing catalogue
A catalogue never sits still. New models arrive, colours vanish, specifications get corrected, and some products stay listed long after they’ve gone. An audit that treats the catalogue as static will miss all that, and you’ll end up polishing pages that no longer match the current offer.
The point of the audit is to find where descriptions fail to help customers, and where published information no longer matches its source. It isn’t to make every description longer or to force one paragraph structure onto very different products. I start with the shape of the catalogue and the decisions a shopper has to make.
Define the audit scope
Choose the product set and the question the review should answer. At Garey I keep each audit to one product group, since a catalogue-wide pass never gets finished. Is the worry missing specifications, duplicate copy, outdated claims, variant confusion, weak descriptions or maintenance? State the date of the catalogue extract and the fields included. A broad audit can help. An undefined one gives you a huge spreadsheet and no action.
Group products by meaningful characteristics: product type, template, supplier, lifecycle status, market, or attribute completeness. Do not assume that every product in a category has the same needs. A simple accessory and a complex device may require different descriptions. Use groups to locate patterns, then inspect examples before generalising.
Include status and availability context. Distinguish active, temporarily unavailable, discontinued, preorder, and seasonal items if the business tracks them. The right content action depends on the product’s role and expected return. An out-of-stock item may retain useful search and customer value; an obsolete item may need a replacement path or an archive decision.
Create a review framework
Set criteria that connect copy to customer needs. Does the description accurately identify the product? Are important attributes, compatibility, care, use, or limitations clear where relevant? Does it distinguish variants? Are claims supported by approved source material? Are the next steps and related policies reachable? Use criteria that can be checked, and allow “not applicable” where a feature does not belong.
Keep factual and editorial findings apart. A missing measurement is a data issue, a confusing sentence is a copy issue and an unsupported benefit is an approval issue. Each needs a different owner. A single severity column hides that, so record issue type, evidence, impact and next action separately.
Use actual customer evidence where available. Support questions and onsite search can point to missing explanations, but they do not automatically apply to every product. Product documentation and supplier data can verify attributes, but they may not be written for customers. Record which source supports each change and where a human reviewer is needed.
Audit data and rendered pages together
Export product identifiers, titles, descriptions, variants, key metafields, status, availability, and other fields that matter to the criteria. Preserve IDs so results can be traced back to Shopify records. Do not use names alone as identifiers; similar products can have similar titles.
Review a sample on the live storefront. The description in an export may not match the rendered page if a theme, app or metafield changes the visible content. Look at mobile and desktop, variant selection, expandable sections and anything that appears only under certain conditions. Check that the facts on the page agree with the source record.
Look for duplicate or near-duplicate language, but do not treat all reuse as automatically wrong. Standard care instructions or policy language may be intentionally shared. The question is whether the shared text prevents a shopper from understanding the specific product or creates an inaccurate claim. Separate reusable factual modules from product-specific explanation.
Prioritise work by consequence and effort
A useful queue distinguishes incorrect information, missing decision-critical details, confusing variant relationships, and low-priority style improvements. Correct inaccuracies and unsupported claims through the appropriate owner first. Then address gaps that block product comparison or use. Cosmetic rewrites can wait when essential facts remain incomplete.
Estimate the effort and dependencies honestly. Some descriptions are fixed by updating a verified data field. Others need a product specialist, translator, legal review or photography. Copywriting is not the answer to a missing fact, so mark the products that need a source decision instead of asking editors to improvise.
For large groups, pilot the review process on a representative subset. Check whether the criteria identify real issues and whether the owners can complete the resulting tasks. Adjust the workflow before expanding. This avoids producing a large list that cannot be actioned.
Maintain the audit after publication
Record the change, reviewer, evidence source, affected products, and any exception. Recheck the live pages after updates and confirm that the underlying product data will remain correct in the next import. Define triggers for another review, such as a supplier change, new variant, updated instructions, or repeated customer question.
Turn findings into maintainable tickets
A description audit becomes actionable when each issue has a product identifier, exact field or page location, source needed, responsible owner, and acceptance check. Avoid sending writers a spreadsheet row that says “improve copy” without explaining the customer question or evidence. If the issue is that a warranty detail is missing, first ask who can confirm the current terms.
Group tasks when a common template or data field causes the same problem, but keep product-specific exceptions visible. A shared correction can save effort; an overly broad replacement can spread an error. Review a small set after any bulk edit, including records that were not expected to change, to detect unintended side effects.
Set a maintenance trigger instead of relying on a calendar alone. Supplier specification changes, new model introductions, repeated support enquiries, and changes to return conditions can all prompt review. The trigger should notify the data owner and content owner, not only the SEO team. This keeps descriptions tied to the products they describe rather than to the date of the last audit.
Related reading
- Preparing a Shopify migration when the catalogue keeps changing
- Organising product care instructions across a catalogue
- A Shopify collection audit for products with changing stock
An audit is only useful if it leads to an owned maintenance process. Which product group has changed most since its descriptions were last reviewed, and what source can verify the current facts?