Product Descriptions
Date: 2026-08-17
Writing at catalogue scale without writing nothing. The tension is that good descriptions are specific and specificity doesn’t scale — which makes the real question what to structure and what to write, rather than how to write more.
A product description provides the information needed to decide on one item. Its job is answering “will this work for me”, which is rarely what a supplier’s marketing copy answers — Product Page Design.
Structure most of it
The scale problem has a structural answer:
STRUCTURED DATA WRITTEN COPY
size, weight, volume what it's for
ingredients / materials who it suits
compatibility how it differs
certifications when to choose it
usage instructions
dosage / directions
→ generated from a table → written per
→ filterable product, or per
→ comparable small group
→ never prose → the expensive
part
Specifications must never be prose. A table is scannable, comparable and machine-readable; the same content in a paragraph is none of those — and converting it is usually an afternoon’s work with a large payoff.
Structured attributes also drive filtering, so incomplete data doesn’t just harm the description — it removes the product from every filtered view — Faceted Filtering, Taxonomy and Labelling.
What the written part must cover
WHAT IT IS in plain terms,
immediately
WHO IT'S FOR and, importantly,
who it ISN'T for
→ "not suitable for
broken skin" prevents
a return and builds
trust
WHAT MAKES IT against the alternatives
DIFFERENT you also stock
HOW TO USE IT especially where misuse
causes a bad outcome
and a return
WHAT TO EXPECT timeframe, sensation,
result
→ sets expectations
that prevent returns
Stating who it isn’t for is the most under-used technique and the most trust-building. It costs a small number of sales and prevents a larger number of returns and complaints — Return Rate and Reverse Logistics.
Never use the supplier’s copy verbatim
DUPLICATE CONTENT the same text on
every retailer's
site
→ no reason to rank
→ Technical SEO
WRITTEN TO SELL IN to buyers, not to
shoppers
MAY CARRY CLAIMS that you become
responsible for
publishing
See: Technical SEO
The third point is the one with teeth. Health, efficacy and comparative claims in supplier copy become your claims once published.
[CHECK: ASA/CAP rules and any sector-specific regulation for your category before republishing supplier claims.]
Scaling it
TIER BY VALUE
top 20% of revenue → bespoke copy
middle → template + specific detail
long tail → structured data + a
strong template
TEMPLATE THE FRAME, NOT THE WORDS
a consistent structure filled with
genuinely specific content
REUSE AT RANGE LEVEL
brand and range descriptions written
once, shown across products
USE CUSTOMER LANGUAGE
reviews and Q&A contain the words
and the objections
— Voice of Customer Data
Tiering by revenue is the pragmatic answer. Bespoke copy for everything is unaffordable; a strong template plus complete structured data serves the tail adequately.
On generated copy
Worth addressing directly, since it’s the obvious scale answer:
REASONABLE
a first draft from structured
attributes, reviewed by a person
summarising review themes
generating specification tables
RISKY
publishing unreviewed
→ invented attributes
→ regulated claims
→ duplicate-feeling text at scale
The review step is what makes it usable, and in a regulated category it isn’t optional — a hallucinated ingredient or claim is a compliance problem, not a quality one.
Measuring it
ADD-TO-BASKET RATE by product, against
category average
RETURN REASONS "not as described",
"wrong size" point
directly at the
description
← the sharpest signal
PRODUCT Q&A VOLUME questions asked are
the description's
gaps
SEARCH ENTRY does it rank at all
Return reasons are the most under-used metric here. They name the specific failure, they cost margin so they’re arguable commercially, and they’re already being collected — Contribution Margin.