How to write a dropshipping product description that converts

The short answer

A product description that converts is not written, it is assembled. The sentences that do the work are already sitting in three places: the ads for this product that have been running longest, the reviews buyers left on the supplier listing, and the spec sheet. Your job is to collect that evidence, put it in the order a buyer asks for it, and cut everything else.

The order that works is almost always the same: what it does for me, proof that it does, the four boring facts that decide the purchase (delivery, fit, contents, returns), then specs for the people who read specs. Write it for a phone screen, not a page. And do not paste the supplier text, for a legal reason, a returns reason and a search reason that are all covered below.

First, the honest part: the description is rarely why nothing sells

Descriptions are the thing beginners rewrite when sales are flat, and they are usually not the problem. Baymard Institute, which manually scores product page UX across 155 or so benchmarked retailers with more than 30,000 scored data points, finds that only 10% of ecommerce sites have product descriptions that are insufficient for users’ needs. The wider picture is much worse than that: on the same research, 48% of desktop sites and 38% of mobile sites reach a “decent” or better product page UX, so the majority of what is broken on product pages is somewhere other than the words.

So read this page as a conversion lever worth an hour, not a rescue. If your store has traffic and no sales at all, the cheaper diagnosis is in why your store gets traffic but no sales, and if the maths of the offer never worked, no wording fixes it: check that first with the profit margin calculator. We cannot guarantee that a better description sells more, and anyone who tells you a template will is selling you the template.

What a description reliably does do is stop the wrong people buying, which is worth real money in a market where returns are the second bill nobody budgets for. That part comes later on this page.

Where the sentences come from (three piles of evidence)

Blank-page paralysis comes from trying to invent claims. You do not need to invent anything. Collect three piles first, and the writing takes twenty minutes.

The first pile is the one most beginners skip, and it is the most valuable, because ad longevity is the closest thing to a public vote on copy. An ad that is still running after a year has survived hundreds of days of somebody deciding whether to keep paying for it. Its angle, its first line and its promise have all been tested with money in a way your first draft has not.

The SpotPeaks radar currently tracks 640 products with live ads across 80 niches, and every one of those records how many days the ad has been live.

As of August 2026 the longest-running ad on our radar sits in pet supplies at 1,304 days. That is not a product recommendation, it is a copy lesson: a message that has held for three and a half years is telling you which promise the market keeps responding to in that niche.

Posture correctors are a useful example because the product is boring and the copy is everything. The long-running ads in that niche almost never describe the strap. They describe sitting at a desk without an aching upper back by the afternoon, which is the outcome the buyer is shopping for. You can read the current set on the posture corrector radar page and the Facebook side of it on Facebook posture corrector ads, then take the angles, not the sentences: copying a competitor’s wording gets you their conversion rate at best and a complaint at worst.

The structure that works, block by block

Buyers do not read top to bottom. They scan for the answer to whatever question is currently blocking them, and they leave when they cannot find it. So the description is a stack of answers, shortest and most decisive first.

Two rules govern all six blocks. Every claim must be checkable, because an unverifiable claim is the thing that turns into a return. And every sentence has to survive the question “so what does that mean for me”: “600D oxford fabric” does not survive it, “600D oxford fabric, so the dog can pull on it without tearing” does.

A worked rewrite

Here is a supplier description of the sort that arrives with a listing, and the same product written for a buyer. The facts are identical. Only the order and the framing changed.

Three things changed and they are the whole method. The title names the buyer and the outcome instead of stacking keywords. The vague claims (“high quality”, “one size fits all”) became numbers, which is also what stops the wrong-size return. And the two questions a buyer always has, when does it arrive and what if it is wrong, are answered on the page rather than discovered at checkout.

Notice what did not change: no cure was promised, no medical claim was made, and no fake urgency was added. Health-adjacent products are exactly where an overclaim turns into a chargeback or an EU advertising complaint.

Write it for a phone

Most of your traffic reads this on a screen the width of a hand. Statista’s global retail device figures, as reported in 2026 roundups, put mobile at roughly 78% of ecommerce site visits but only about 66% of orders, so the phone is both where your readers are and where the drop-off happens. That gap is not caused by wording alone, but a wall of desktop-shaped text does not help it.

What that means in practice: paragraphs of two or three lines, bullets over prose for anything that is a list, the delivery and returns line high enough that nobody has to hunt for it, and no table wider than the screen. Read your own page on your own phone before you publish it. It is free and it is the single most reliable edit on this page.

The four boring facts that actually decide the sale

Beginners spend their effort on the emotional part and skip the logistics, and the logistics are what the buyer is stuck on. Answer these four, in the description, in plain numbers.

• When does it arrive. A date range beats “fast shipping”. If you ship from EU stock, say so: it is a conversion feature. Our guide to finding EU suppliers covers how to get that shipping time down in the first place.
• Will it fit or work with mine. Measurements in centimetres and inches, weight limits, compatibility lists. This is the number one source of avoidable returns in most categories.
• What is in the box. Quantity, accessories, batteries, charger. “1 x cable” when the photo showed two is a refund.
• What happens if they hate it. The return window, who pays the postage, how long the refund takes. In the EU most of this is not optional, which is the next section.

What EU law requires the page to say

This is the part US-written guides on this query skip entirely, and it is not optional if you sell into the EU. Two separate rules land on the product page.

The General Product Safety Regulation requires, for distance selling, that the offer itself carries the manufacturer’s name and postal and electronic address, the details of the EU responsible person, a product identifier such as a type, batch or serial number, and any warnings or safety information in the language of the country you are selling into. Article 19 is the provision that puts those duties on the online offer specifically, which means they belong on the product page and not only in a policy page somewhere. Our compliance pages go into what that means per product type.

Separately, EU consumer law gives the buyer a 14-day withdrawal right, and the practical detail is that the clock and the conditions have to be visible before purchase, not discovered afterwards. The mechanics, including who pays return postage, are in dropshipping returns and refunds, and the wider EU picture including VAT and the flat customs duty on non-EU parcels is in dropshipping in Europe.

Practical version: add a short compliance block at the bottom of every description with the manufacturer, the EU responsible person, the identifier and the warnings, and put a one-line summary of the return right next to the delivery line at the top. It costs you nothing in conversion because the buyers who read it were the nervous ones anyway.

Descriptions are a returns lever, and returns eat margin

The reason to be precise rather than persuasive is arithmetic. Return-reason surveys keep putting description accuracy near the top: a 2026 compilation of shopper return research reports that an inaccurate description or a discrepancy from the product images accounts for around 60% of shoppers’ stated top reasons for returning an online order, and the Narvar Consumer Report that most of these roundups trace back to lists “product not as described” alongside wrong size, colour or fit as the leading causes. Treat the exact percentage with suspicion, because these are self-reported survey aggregations republished by vendors rather than measured order data, and the roundups disagree with each other on the level while agreeing on the ranking.

The direction is what matters, and it costs real money. When a returned dropshipped item is not worth shipping back to a supplier in China, every point of return rate costs you roughly one percent of your sale price straight off the bottom line. On a product with a 30% contribution margin, going from a 5% to a 12% return rate removes nearly a quarter of your profit per order. That is the size of prize for writing “fits chests 71 cm to 122 cm” instead of “one size fits all”. Run your own numbers with the profit margin calculator and the breakeven ROAS calculator if you want to see how fast the returns line moves the target your ads have to hit.

The SEO part: do not paste the supplier text

Copying the manufacturer or AliExpress description onto your store is the default beginner move and it is the one thing search guidance names directly. Google’s spam policies describe thin affiliate pages as ones where “product descriptions and reviews are copied directly from the original merchant without any original content or added value”, and the scaled content abuse policy covers generating many pages, by AI or otherwise, “without adding value for users”. Paraphrasing the same text with a model does not fix it: the wording becomes unique, the information does not.

What adds value on a dropshipping product page is genuinely narrow, and worth doing precisely because it is narrow: your own measurements from the sample you ordered, your own photos, the delivery time from your actual warehouse, the objections you pulled out of reviews and answered, and the compliance detail. None of that is in the supplier feed. All of it is checkable.

Two mechanical points while you are in there. Write the meta description yourself rather than letting the platform truncate the first paragraph: Google measures it in pixels, roughly 920 on desktop and 680 on mobile, which lands at about 150 to 160 characters with the important part inside the first 120. And check that your platform is emitting product structured data with at least name, image and offers including price and currency, which is what Google requires for a merchant listing result. Shopify and WooCommerce both do this out of the box; a custom theme may not.

Using AI without producing sludge

AI is genuinely good at this task and genuinely bad at it unsupervised. It is good at turning your three piles of evidence into ordered, readable blocks in seconds, and at generating five angles when you are stuck on one. It is bad at knowing whether the strap is neoprene, whether the cable is in the box, or whether the product is safe to claim anything medical about, and it will happily write all three with total confidence.

So the division of labour is fixed: you supply the facts, it supplies the arrangement, and you check every number against the spec sheet before publishing. The SpotPeaks launch kit is built around that division. It takes a product off the radar and drafts the listing, the bullets, the ad angles and the video script from the real product data, and it leaves the measurements and the compliance block for you to fill from your supplier, because those are the two things no model can know.

How to tell whether it worked

A description change is testable, cheaply, if you decide the measure before you change anything.

• Add-to-cart rate on the product page. The description sits between the product view and the add to cart, so that is the step it should move. Site-wide conversion rate is too noisy to read at beginner traffic levels.
• Return rate and refund reasons. The slower measure and the more valuable one. Tag every return with a reason from day one, even at ten orders.
• Support questions. Every question a customer emails you is a hole in the description. Answer it on the page and the next hundred buyers do not have to ask.

Do not run a formal A/B test on a new store. At a few hundred visitors the result will not be significant, and the arithmetic behind that is in the traffic and no sales guide. Fix the obvious holes, watch the reasons, and move on to the thing that moves more money, which at your stage is almost always the creative and the price.

The ten-minute checklist

• The first line names the outcome the ad promised, in the buyer’s language.
• Every vague adjective has been replaced by a number or deleted.
• Delivery time and where it ships from are visible without scrolling.
• Measurements, contents and compatibility are stated exactly.
• The 14-day withdrawal right and who pays return postage are stated.
• The GPSR block is present: manufacturer, EU responsible person, identifier, warnings.
• No sentence was copied from the supplier listing.
• No medical, safety or income claim you cannot evidence.
• You read the whole thing on your own phone.
• The meta description is written by hand, about 150 characters.

The verdict

Good product copy is evidence, ordered. Gather the ads that survived, the reviews that complained and the spec sheet that proves; answer the four boring questions in numbers; add what EU law requires; write it for a phone. That is the whole job, and it takes under an hour once the evidence is in front of you.

It will not save a product whose economics never worked, and we will not pretend otherwise. Next step: pick the product first and check that it can carry an ad budget at all. Start with the winning-products radar to see what already has ads running and how long they have survived, run the number through the profit margin calculator, then come back and write the page. If you are earlier than that, the ordered version of the whole path is in how to start dropshipping and the rest of the product page is in how to build a store that converts.

FAQ