TL;DR
You bought something because the listing said it did X. Eight months later it doesn't do X, you go back to the page, and the page now says Y. Or the listing is gone entirely because the seller delisted it.
Marketplace listings are editable documents. Sellers change specifications, shorten warranty descriptions, swap photos, and remove claims — often for perfectly innocent reasons, and always without notifying anyone who already bought. Your order confirmation records what you paid for. It does not record what you were told.
Capture the listing at the moment of purchase. It takes fifteen seconds with Convert: Web to PDF, produces a real PDF with selectable text and working links, and runs entirely on your machine.
Then, if you ever need to claim, Convert: Anything to PDF merges the listing, the receipt, and photos of the fault into one file that a returns team can actually process.
General guidance, not legal advice. Consumer rights vary by country and by contract.
What changes, and when
Understanding the pattern tells you what's worth capturing.
Specifications get corrected. Battery life, capacity, dimensions, compatibility. Sometimes a genuine error being fixed. From your side, indistinguishable from a claim being walked back.
Warranty descriptions shorten. "3-year warranty" becomes "1-year warranty, extendable" — often because the promotional period ended and nobody thought about buyers from the promotional window.
Photos are swapped. A newer revision of the product replaces the images, and the accessory shown in the photo you bought from isn't in the box any more.
"Includes" lists get trimmed. Cables, adapters, mounting kits quietly leave the bundle.
The listing disappears. Delisted, out of stock permanently, or the seller left the marketplace. This is the worst case, because there's nothing left to point at.
Prices change, which matters for price-matching and for establishing replacement value on an insurance claim.
The marketplace wrinkle
On marketplaces with multiple sellers for one product, the page you bought from and the page you return to may not be the same seller's version at all. The product page is shared; the seller-specific terms aren't. Capturing at purchase pins down which one applied to you.
What to capture, and when
At purchase — fifteen seconds
The full product page, including:
- Title and model number
- The complete specification table (expand any collapsed sections first)
- Warranty and returns text
- What's included in the box
- The seller's name and rating
- Price and any promotional terms
- Delivery estimate
The order confirmation, which you'll also get by email, but the on-site version often shows more detail — the exact seller, the specific variant, and the returns window.
The seller's own returns policy page if they have one separate from the marketplace's.
For high-value purchases, that's fifteen seconds each and worth doing without thinking about it. For a £12 cable, don't bother.
When something goes wrong
- Photos of the fault, in sequence: whole item, the problem area, close-up, serial or model plate
- The current listing again — the diff between then and now is the whole point
- Every message with the seller
- Any diagnostic or error screen
- The returns portal page showing your request and its reference
Doing the capture properly
Expand everything first
Product pages hide most of their content behind tabs and accordions: full specifications, warranty details, shipping terms, Q&A. A collapsed section isn't rendered, so it won't be in the PDF. You'd archive the headings and none of the substance.
Click through every tab. Expand every "show more". Then convert.
Stay out of Article Mode here
Article Mode runs the page through Readability and produces a clean single-column read. It's excellent for articles, and wrong for this: it discards the specification tables, the seller information block, and the page structure — exactly the parts that carry the evidence.
Use default mode. Use Remove Elements for the cookie banner and the chat widget if they're covering text, and stop there.
Turn on headers and footers
Chrome's print engine can stamp the source URL and capture date onto every page. For anything you're keeping as a record, enable it. A PDF that shows where it came from and when is worth considerably more than one that doesn't.
Handle the lazy images
Product pages defer image loading. If you convert without scrolling, you'll get grey boxes where the photos were — and the photos are frequently the thing in dispute. Use Load All Images, or scroll the gallery through first.
Name it so you'll find it
seller-product-2026-08-07.pdf. When you're looking for it in fourteen months, the date is what you'll search on.
Why the browser matters here
Most of these pages are personalised or behind a login: your order history, your returns portal, your message thread with the seller, the price you were shown.
URL-based converters can't reach any of that. PDFCrowd and similar services fetch the address from their own servers with no session, so an order page returns a login screen. That's structural, not a setting.
They also can't see what you saw. Marketplace prices and availability are personalised by region, account, and sometimes by browsing history. A converter fetching the URL from a datacentre in another country gets a different page than the one you bought from — which defeats the entire purpose.
Screenshot tools produce an image. That captures the appearance and loses the text: not searchable, not quotable in an email, and the links to the seller's other pages are dead. When you're trying to paste an exact sentence from a warranty description into a support ticket, an image is much less useful than it looks.
Convert: Web to PDF converts what's rendered in your tab, locally, through Chrome's print engine. Logged-in pages work. The text stays selectable. The links stay clickable. Nothing is uploaded.
Making a claim that works
Send one file
Returns and warranty teams process high volumes. A single well-ordered PDF gets read; twelve attachments get partially read.
Order it as an argument:
- Summary page — what you bought, when, from whom, what's wrong, what you want
- Order confirmation
- The listing as it was at purchase — with the relevant claim highlighted if you can
- The listing as it is now, if it changed
- Photos of the fault
- Correspondence to date
Points 3 and 4 next to each other do the work. You're not asserting that the page changed; you're showing it.
Be specific about what you want
"The listing said 3-year warranty at the point of purchase (page 4), the product failed at 14 months, and I'm requesting a repair under that warranty" is actionable. "This is unacceptable" is not.
Keep it proportionate
Most returns are straightforward and need none of this. A capture at purchase costs fifteen seconds and only ever gets used in the small fraction of cases that go wrong — which is exactly why it has to be cheap enough to do without deciding first.
What this is and isn't
It is: contemporaneous documentation of what a page said when you looked at it. Searchable, quotable, dated, and far better than a memory.
It isn't: notarised, cryptographically timestamped, or tamper-evident. You made it, so in a formal dispute it's your document rather than an independent record.
In practice that distinction rarely bites, because most of these disputes are resolved by a returns team reading a clear file, not by a tribunal weighing evidence. A dated PDF of the listing usually ends the conversation, because the seller can check their own version history and see you're right.
If a matter is genuinely heading to legal proceedings, that's the point to involve someone who deals with preservation properly.
Worth knowing: the Wayback Machine sometimes has marketplace listings, and it's an independent third-party record, which is stronger than your own capture. Coverage is patchy and personalised or logged-in pages are never captured — but it costs nothing to check, and when it has the page, it's the better artefact.
Quick reference
| Purchase type | Capture at purchase | Extra |
|---|---|---|
| Electronics over £100 | Listing, specs, warranty, order | Serial number photo on arrival |
| Appliances | Listing, warranty, installation requirements | Installer's paperwork |
| Marketplace / third-party seller | Listing, seller page, seller's returns policy | Message thread |
| Refurbished or open-box | Listing, condition grading, warranty | Photos on unboxing |
| Made-to-order or custom | Full spec, your configuration, lead time | All correspondence |
| Under £20 | Nothing | — |
The refurbished row is worth flagging: condition grading definitions ("Excellent", "Very Good") are defined on a page that changes, and they're the exact thing disputes turn on.
Frequently asked questions
Can a seller change a product listing after I've bought from it?
Yes. Listings are live pages, and sellers routinely update specifications, warranty text, photos, and included accessories without notifying past buyers. Your order confirmation records the transaction, not the description you relied on — which is why capturing the page at purchase is worth the fifteen seconds.
Is a PDF I made of a listing useful for a warranty claim?
For most disputes, yes — it's contemporaneous, dated, quotable documentation, and it usually resolves things because the seller can verify it against their own records. It isn't notarised or tamper-evident, so it carries less weight in formal proceedings than an independent record would. Check the Wayback Machine too; where it has the page, a third-party archive is stronger.
Why does my captured product page have blank boxes instead of photos?
Product pages defer image loading until you scroll to them. Use Load All Images before converting, or scroll the full gallery first. On a page where the photos are the evidence, this step isn't optional.
Should I use Article Mode for product pages?
No. Article Mode rebuilds the page as clean single-column text and discards specification tables, seller details, and page structure — the parts that matter for a claim. Use the default mode, and Remove Elements only for cookie banners or chat widgets covering the text.
Can I capture my order confirmation and returns portal pages?
Yes, because the conversion happens in your browser on the page you're already viewing. Logged-in pages work normally. Hosted URL-based converters can't do this — they fetch the address from their own servers without your session and get a login screen.
How do I get the capture date into the PDF?
Enable Chrome's print headers and footers in the extension settings, which stamps the source URL and date onto each page. Also date-suffix the filename so files sort chronologically and you can find them a year later.
Is it worth capturing every purchase?
No. Do it for anything expensive, anything with a warranty you might use, anything from an unfamiliar third-party seller, and anything refurbished or custom. Skip small everyday purchases. The habit only works if it's cheap enough to do without deliberating.
Bottom line
The listing is the promise, and the listing is editable. Your receipt proves what you paid; it doesn't prove what you were told you were buying.
Fifteen seconds at checkout closes that gap. Expand the tabs, load the images, convert in default mode with headers and footers on, date the filename. You'll never look at most of them.
The ones you do look at will be the ones where the page changed — and having both versions side by side turns an argument into a document.
Convert: Web to PDF captures the listing, including behind a login, entirely on your machine. Convert: Anything to PDF merges it with your receipt and fault photos into one file a returns team can process. Both free, no watermarks.
While you wait for the replacement to arrive: CineMan AI picks something worth watching.