Sign in Start free

Product Descriptions At Scale

Manufacturer copy is identical across every retailer selling the item, so you are competing on duplicate text against sites with more authority. Rewriting all of it by hand is not affordable, and running every SKU through a generator produces text that is unique but says nothing, which does not help either. The fix is deciding which SKUs deserve human effort and making the generated tier structurally different rather than just reworded.

CATEGORY
Ecommerce SEO
FORMAT
product-descriptions-at-scale.md
STEPS
6
PRICE
Free - no account
WHEN TO REACH FOR THIS

Use when you have thousands of SKUs on manufacturer-supplied copy and product pages get impressions but almost no clicks.

The skill file

product-descriptions-at-scale.md 
---
name: product-descriptions-at-scale
description: Use when you have thousands of SKUs on manufacturer-supplied copy and product pages get impressions but almost no clicks.
---

# Product Descriptions At Scale

Manufacturer copy is identical across every retailer selling the item, so you are competing on duplicate text against sites with more authority. Rewriting all of it by hand is not affordable, and running every SKU through a generator produces text that is unique but says nothing, which does not help either. The fix is deciding which SKUs deserve human effort and making the generated tier structurally different rather than just reworded.

## What you need first

- a SKU list ranked by revenue and by search impressions
- your structured attribute data - the fields that vary between SKUs
- a sample of manufacturer copy to diff against

## Method

1. Rank SKUs by impressions times margin. The top 5 percent typically carry most of the revenue and get hand-written copy. Everything else gets a structured template.
2. For the templated tier, build copy from your own attribute fields rather than paraphrasing the manufacturer - dimensions, compatibility, what it fits, what is in the box. Attribute-derived text is genuinely unique because your attribute combinations are.
3. Add one variable human sentence per SKU family, not per SKU: a single line about who this range suits, reused across the family, then attribute detail below it.
4. Keep the manufacturer specification table but mark it up as a table and place it below your own text, so the unique content is what leads.
5. Run a shingle-based similarity check across your own catalog before publishing. Anything above roughly 80 percent overlap with another SKU on your site needs more attribute variance in the template.
6. For variants that differ only by color or size, consolidate to one canonical product URL with variant selection rather than publishing near-identical pages.

## What this produces

A two-tier description system: hand-written copy for revenue SKUs, attribute-driven templates for the tail, with an internal duplication check.

## Where this goes wrong

- running manufacturer copy through a paraphraser, which produces text that is unique to a checker but adds no information and still ranks nowhere
- publishing every size and color variant as its own indexable URL and splitting the ranking signal across a dozen pages
- leaving the manufacturer specification block above your own copy, so the duplicate text is what gets weighted first

---

From the QuQi skill library - https://www.quqi.io/skills/product-descriptions-at-scale
Free to download · no account, no email

What you need first

  • a SKU list ranked by revenue and by search impressions
  • your structured attribute data - the fields that vary between SKUs
  • a sample of manufacturer copy to diff against

Method

  1. 01 Rank SKUs by impressions times margin. The top 5 percent typically carry most of the revenue and get hand-written copy. Everything else gets a structured template.
  2. 02 For the templated tier, build copy from your own attribute fields rather than paraphrasing the manufacturer - dimensions, compatibility, what it fits, what is in the box. Attribute-derived text is genuinely unique because your attribute combinations are.
  3. 03 Add one variable human sentence per SKU family, not per SKU: a single line about who this range suits, reused across the family, then attribute detail below it.
  4. 04 Keep the manufacturer specification table but mark it up as a table and place it below your own text, so the unique content is what leads.
  5. 05 Run a shingle-based similarity check across your own catalog before publishing. Anything above roughly 80 percent overlap with another SKU on your site needs more attribute variance in the template.
  6. 06 For variants that differ only by color or size, consolidate to one canonical product URL with variant selection rather than publishing near-identical pages.

What this produces

A two-tier description system: hand-written copy for revenue SKUs, attribute-driven templates for the tail, with an internal duplication check.

Where this goes wrong

  • running manufacturer copy through a paraphraser, which produces text that is unique to a checker but adds no information and still ranks nowhere
  • publishing every size and color variant as its own indexable URL and splitting the ranking signal across a dozen pages
  • leaving the manufacturer specification block above your own copy, so the duplicate text is what gets weighted first

Use this skill in your own AI

The download is a plain markdown file with the name and trigger in its frontmatter. Where an assistant supports skills it can load itself, that frontmatter is what it reads to decide this one applies.

Claude Code Save it as ~/.claude/skills/product-descriptions-at-scale/SKILL.md and Claude loads it on its own when what you are doing matches the trigger line. Put it in .claude/skills inside a project instead if the whole team should have it.
Claude Upload the file in the skills section of your settings. Once it is there it applies itself in any conversation where the trigger fits, so you do not have to remember it exists.
ChatGPT There is no skills format to install into, so paste the file contents into a Project instruction or a Custom GPT instead. It then applies to every chat in that project rather than only the one you paste it into.
Anything else Paste the markdown into the chat before your question. It works in any assistant, it just has to be pasted again each time.

Questions about this skill

When does this beat simply rewriting the descriptions?

When thousands of SKUs sit on manufacturer copy and collect impressions without clicks. Both obvious routes fail at that size: hand rewriting the catalogue is not affordable, and pushing every SKU through a generator produces text that is unique to a checker and informative to nobody. This method decides which SKUs deserve a human and makes the rest structurally different.

What do I need before splitting the catalogue into tiers?

A SKU list ranked by impressions times margin, your structured attribute fields, and a sample of manufacturer copy to diff against. The attribute data is load bearing, because text built from your own fields is genuinely unique while paraphrased text is not. Without the ranking, hand written effort lands on whatever a merchandiser mentioned last rather than on revenue.

What does the two tier system produce, and which part gets used repeatedly?

A two tier system: hand written copy for roughly the top five per cent by impressions times margin, attribute driven templates for the tail, and a shingle based similarity check across your own catalogue. The check is what gets used repeatedly, since anything above about eighty per cent internal overlap means the template needs more attribute variance before publishing.

Which mistake costs the most on a large catalogue?

Publishing every size and colour variant as its own indexable URL, which splits one ranking signal across a dozen near identical pages and multiplies the duplication you set out to fix. Consolidate to one canonical product URL with variant selection. Leaving the manufacturer specification block above your own text is the cheaper error and still weights the duplicate text first.

More in Ecommerce SEO

Faceted Navigation Crawl Control

Use when a category with a few hundred products has generated tens of thousands of crawlable fil...

Category Page Copy That Ranks

Use when category pages sit in positions 8 to 20 for their head term and the page is a bare grid...

Out Of Stock And Discontinued Products

Use when products go out of stock or get discontinued and you need to decide what happens to the...

Product Schema That Earns Listings

Use when product pages are valid in the rich results test but show no price, rating, or availabi...