ASIN and Listing URL Import

You paste an ASIN or an amazon.* product URL and press one button. Graflio fetches the live listing page, parses the title, brand, price, star rating, rating count, bullets, the specification table, up to twelve gallery photographs, the colour variations and the reviews Amazon embeds on the page itself, then finds the product's deepest best-seller category, reads up to ten competing listings the same way, and writes a design brief from all of it. That read costs 0.1 credit.

What one paste actually retrieves

The import is a parser, not a guess. Here is the exact list of what comes back from your own listing, with the caps the code applies:

• Title, brand and price as displayed on the marketplace you linked to, in that marketplace's language and currency.
• Star rating and rating count, parsed out of the localised formats — "4,6 de 5 estrellas", "1.234 valoraciones" and the rest all resolve to numbers.
• Bullets from the feature-bullets block, up to twenty stored and the first eight sent into the brief.
• The product overview table — dimensions, material, capacity, model number — up to thirty rows stored, fifteen used.
• Up to twelve gallery images, at Amazon's hi-res variant where one exists, deduplicated.
• Colour variations: up to twenty-four, each with its name, its own child ASIN where the page exposes one, and its swatch re-requested at 320 px so it works as a colour reference rather than a 36 px thumbnail.
• Reviews: the ones Amazon renders into the product page, with star rating, title, body, date and the verified-purchase badge.
• Category signals: the links inside the Best Sellers Rank block first, because they name the exact deep subcategory, and breadcrumb browse-node ids as a fallback.

What it does not read, and why

This is the part most feature pages skip. Amazon's dedicated /product-reviews pages sit behind a login wall, so the review sample is whatever Amazon chooses to embed on the product page — roughly eight to ten, not the full thousand. That is a real ceiling. It is also less limiting than it sounds, because the brief also reads up to ten competitors' review samples, so a category's complaints show up repeatedly even when no single listing gives you many.

Also not read: customer questions and answers, A+ content modules, video, sales or BSR history, keyword or search-term data, and anything that requires a Seller Central login. Variation parsing covers the colour theme; size and style themes are not extracted as families. If your product has no reviews yet, the import still works and the competitor pass still runs, but the brief is drawing on the category rather than on you — which is exactly the honest answer for a pre-launch product.

From scraped rows to a design brief

The raw data is not the point. The import chains into a research pass with a fixed shape: the deepest best-seller category is resolved, that list's first page is fetched (thirty ASINs render server-side), your own ASIN is removed and its rank noted, and the top ten remaining listings are fetched individually with a pause between each. All of it goes into one call that returns a structured brief:

• Four to seven pain points, each tagged own, competitor or both so you can see whether a complaint is yours or the category's, each with the evidence behind it and competitors named by brand.
• Four to six USPs, each grounded in a review, a spec gap, a price gap or a rating gap.
• One positioning angle, in a sentence.
• A visual style: background, accent and text hex values plus a headline and a body typeface, derived from how the product itself looks.
• Six or seven image scripts, each with its own headline copy, two to four callout texts and a composition.

One rule is enforced in the instructions above all of that: every statistic, percentage, review count, price and guarantee in the brief has to appear verbatim in the scraped data. If the listing does not contain the number, the brief does not print it.

How this differs from starting with a photo

Do not read that table as exclusivity. Pixii's own site advertises listing-URL input, and it composites a real pack shot too. Taking a URL is no longer a moat. What the URL is used for is where the difference sits: Graflio spends the import on a competitor and review pass and shows you the resulting brief before any pixels exist. If a tool takes your URL only to pull the hero photograph, you have bought a download, not research.

Marketplaces, failures and limits

Any amazon.<tld> product URL is accepted, and the public API maps twenty-three country codes onto their marketplaces, from us, uk, de, fr, es, it, nl, se, pl and be through jp, au, in, sg, ae, sa and eg. Requests carry the marketplace's own Accept-Language header, so a German listing is parsed in German and the brief is written in the listing's language.

Scraping Amazon is adversarial and the page is honest about it. Long tracking URLs are stripped back to the canonical /dp/ASIN form, browser fingerprints rotate between attempts, and each fetch gets four tries. Three distinct failures come back with three distinct messages: the product does not exist (404), the page parsed but the markup was unexpected, or Amazon is rate-limiting. Data-centre IP ranges sit permanently behind Amazon's captcha wall, so production traffic goes out through a residential proxy. A research job that fails refunds its credit automatically.

Volume limits: twenty ASINs per submission, two hundred saved products per account. Submitting an ASIN you already imported re-runs the research on the same row rather than creating a duplicate, which is how you refresh a brief after a listing changes.

Who should use the import, and who should not

Use it when the listing is live, has reviews and sits in a category with a best-seller list. That is where the import earns its keep, because the brief tells you something you did not already know. Expect less pre-launch: you get the category's argument, not yours. Skip it if you already know what your six images must say and have a designer to execute them — you need a canvas, not a research step, and Canva is better at being a canvas than we will ever be.

Common questions

Does Graflio need my Seller Central login?

No. The import reads the public product page the same way a shopper's browser does. That is also its ceiling: anything only visible inside Seller Central, such as search-term reports or session data, is not available to it.

Can I import a competitor's ASIN instead of my own?

Technically yes, and sellers do it to study a category. The output is a brief and an image set built around that listing's data, so treat it as research rather than as artwork you can ship.

How many reviews does it actually read?

Whatever Amazon embeds on the product page, which is around eight to ten, for your listing and for each of the ten competitors it fetches. The dedicated review pages require a login, so no tool reading the public page gets more than that without an authenticated scrape.

What happens if Amazon blocks the request?

The job retries four times with rotating fingerprints, then returns a rate-limit error and refunds the credit. Re-running the same ASIN a few minutes later normally succeeds.

Is URL import unique to Graflio?

No. Pixii advertises listing-URL input as well. The honest claim is narrower: what Graflio does with the URL is a competitor and review pass that produces a brief you can read before you spend anything on images.