Why Product Description Quality Affects Ecommerce Rankings
The duplicate content problem is the whole subject. The answer is not to rewrite everything. It is to decide which pages deserve original writing and to accept that most do not. A page pretending five thousand bespoke descriptions are achievable is not useful.
Why Manufacturer Copy Is A Problem
Most product descriptions were written by the manufacturer. The same words appear on every shop selling that item, sometimes hundreds of them, all competing for the same searches with identical text.
Where it comes from. The product feed.
Descriptions arrive with the catalogue. Publishing them as they are gets a shop trading quickly, which is a reasonable decision that nobody revisits.
What it is not. A penalty.
Nothing is being punished and no action has been taken against your shop. That framing is common and it sends people after the wrong fix, which block two deals with properly.
What it genuinely costs. The ability to be chosen on the merits of the page.
When several pages carry the same words, the words cannot be the deciding factor. Something else decides. That something else usually favours whoever is larger.
Why it gets worse as a shop grows. The part people miss.
Adding a thousand products adds a thousand pages of somebody else's text. Range makes this problem larger rather than diluting it.
What Duplicate Content Actually Does
The phrase duplicate content carries an implication of punishment that does not apply here. Correcting that first is necessary, because the wrong diagnosis leads to expensive and useless work.
The myth. That duplicate text triggers a penalty.
It does not. Search engines expect retailers to carry manufacturer copy, because that is how the industry works. Nobody is being sanctioned for it.
The real effect. A page with nothing distinctive has no reason to be chosen.
Where several pages say the same thing, one gets shown and the others largely do not. The choice is made on everything except the text: the strength of the shop, its price, availability, delivery and reviews.
Why that is worse than a penalty would be. Because it cannot be appealed or fixed by removal.
A penalty has a cause you can address. This is simply the absence of a reason to prefer you. The factors that decide it in a competitor's favour are mostly things a smaller shop cannot change quickly.
What follows for the work. The goal is distinctiveness rather than originality.
Rewriting supplier text into different words with the same meaning satisfies originality and changes nothing, because the page still offers no reason to be preferred. Block five is about what does.
The Scale Problem
Five thousand products and one person. That is the situation most shops are actually in. It is why the standard advice on this subject is useless.
What the standard advice says. Write unique descriptions for every product.
Correct in principle and impossible in practice. At a realistic rate for something worth publishing, a full catalogue is years of work. The catalogue changes while you are doing it.
Why acknowledging that matters. Because a plan nobody can execute is not a plan.
Shops given this advice either do nothing, having concluded the task is impossible. Or they start at product one and abandon it around product three hundred.
The alternative most shops reach for. Doing it faster and worse.
Generating thousands of descriptions quickly produces text that exists rather than text that helps, which per block two changes nothing about whether the page deserves to be chosen. Block seven deals with that directly.
What actually resolves it. Deciding that most products will not get original writing.
That is a decision rather than a failure. Everything from block four onward assumes it.
Deciding Which Products Get Original Copy
Three factors decide it, used together. Any one of them alone produces a list that wastes the budget.
Demand. Does anybody search for this product by name?
Per our product page guide, most items have no search behind them at all. Writing for those is writing for nobody.
Margin. Is it worth having when it sells?
A description costs the same to write whatever the product earns, so the return varies entirely with what is left after cost of goods.
How contested it is. Could you realistically be preferred?
An item stocked by every national retailer at a keener price is not going to be won by better copy. An item few shops carry may be won by very little.
The test. A product needs all three to qualify.
Real search behind it, margin worth having. A realistic chance of being preferred. Products failing any one of the three go to the treatment in block five's final paragraph instead.
One addition worth making. Check whether you will still stock it.
Writing for an item about to be discontinued is money spent twice. A surprising amount of description work goes into products already being wound down.
What A Good Description Contains
A good description answers what somebody needs to know before committing. That turns out to be the same thing that makes the page distinctive, which is the useful coincidence at the centre of this subject.
Dimensions and materials. Complete, not partial.
Manufacturer copy is frequently vague here. A buyer who cannot confirm something fits will not risk it.
Compatibility. What it works with, plus what it does not.
The second half is rarer and more valuable. Telling somebody what a product is not suitable for prevents a return and builds more trust than any claim about quality.
What is included. The most common cause of complaints on any shop.
The question your customers keep asking. The single most valuable sentence available.
Whatever your staff answer repeatedly about a product is the thing the description fails to say. Nobody else selling that item has it, because nobody else has your customers.
Why this also solves block two. Because none of it came from the manufacturer.
Compatibility notes, exclusions and the answers to real questions are yours. That is distinctiveness arriving as a by-product of being useful, rather than being pursued for its own sake.
For everything else. Accurate, complete and well presented specifications.
Presented as consistent fields rather than prose, which can be applied across thousands of products because it is data rather than writing.
Writing For The Buyer Rather Than The Crawler
Copy written to rank reads badly and converts worse. Copy written to help somebody decide tends to rank well anyway, which is not a coincidence.
What search written copy looks like. Recognisable immediately.
The product name repeated at intervals, adjectives doing the work facts should do, sentences arranged around terms rather than around information. A reader senses that it is not addressed to them.
What it costs commercially. More than the ranking is worth.
A visitor who reaches a product page is close to buying. Copy that fails to answer their question loses a sale that was nearly made, which is more expensive than not ranking for the term at all.
Why the two goals converge. Because both reward the same thing.
A search engine is trying to identify the page that best answers somebody's question. A buyer is trying to find that page. Writing to answer the question properly satisfies both. The terms that need to appear turn up naturally when you explain something fully.
Where they genuinely diverge. One place, worth naming.
The exact phrase people search may not be the phrase you would naturally use, particularly where an industry term differs from a customer term. Using the customer's word is right for both. Using your own because it is more precise is the one case where writing for the buyer needs a deliberate check.
Using AI Assistance Properly
Generative tools can support somebody writing at catalogue scale. They cannot replace that person. The difference decides whether the output solves this problem or becomes it.
What they genuinely help with. Three things.
Turning structured product data into readable prose. Producing a consistent first draft across a large set for somebody to work from. Rephrasing something a writer has already decided the substance of.
What they cannot do. Supply what is not in the input.
The valuable parts of block five are compatibility, exclusions and the question your customers keep asking. None of that is in the product data, so no tool can generate it. It comes from the business.
What unattended generation produces. Exactly the problem this page is about solving.
Thousands of descriptions written from the same limited input, in the same register, saying the same things in slightly different words. That is undifferentiated text at scale, which is where the shop started, with more of it.
The rule we apply. Everything published gets checked by a person.
Because generated text is confidently wrong in the ways that matter most on a shop: a measurement, a compatibility claim, a material. A description asserting something untrue about a product is a returns problem and a consumer protection problem before it is a search one.
Where that leaves it. A tool for the person doing the work.
Useful, though not a way to avoid the decision in block four. The same position we take across our guides on content.
Structured Data On Product Pages
Markup lets a search engine read a product page directly: what it costs, whether it is available, what customers rated it. Each of those is a statement about the world.
The governing rule. Structured data is an assertion about reality.
Describing something in markup states that it is the case. Where it is not, that is a false statement published in a machine readable form with your name on it.
Three things we will never mark up. A rating nobody gave. A saving that is not real, meaning a was price the item never sold at. Availability that is not accurate.
Why it matters more on a shop. These are consumer protection questions first.
A false price or an invented rating is a considerably larger problem than anything search would do about it. Our guides on structured data hold the same position throughout.
What markup will not do. Compensate for the page.
Accurate markup on a page carrying identical supplier copy describes an undifferentiated page accurately. It is worth doing and it does not address block two.
Measuring Whether It Worked
Description work is expensive enough to be worth testing rather than assuming. A sample tested properly beats rewriting everything and hoping. Almost nobody does it.
How to run the test. Rewrite a defined group and leave a comparable group alone.
Fifty products of similar type, demand and margin. Change one set, change nothing else on the shop, then wait long enough for the pages to be reassessed.
What to measure. Two things, neither of them position.
Impressions for the rewritten pages, which show whether they are being considered for more searches. And conversion rate on those pages, which shows whether the writing helped somebody decide.
Why conversion matters as much as visibility. Per block six.
A description that lifts conversion has paid for itself regardless of what happened to rankings. That effect appears sooner.
What a result tells you. Whether to continue. Where.
A clear lift justifies extending the programme. No movement is worth knowing before spending on four thousand more. It usually means the constraint is elsewhere, most often in the structure or category work our complete guide puts earlier in the sequence.
Nobody is writing
five thousand of them.
The advice to write unique descriptions for every product is correct in principle and impossible in practice. The work is deciding which products deserve original writing and accepting that most do not.
What is included every month:
£350 per month, one target area. No setup fee, nothing billed separately.
Twenty-two guides.
One subject.
This guide covers product descriptions. The rest of the series covers the sequence, structure, category and product pages, technical health, measurement and everything a store owner has to decide.
