A Wishlist for Zara

If you’ve ever dropped a link to a dress from Zara into a saving tool and gotten an empty card back — no photo, no price — that’s not a glitch on your end.

Why a Zara link comes back empty

Zara, like Massimo Dutti, Bershka, Hugo Boss and most of the bigger retail groups, actively defends its product pages against automated traffic. The target is usually price-scraping software run by competitors, checking stock and pricing across the market every day. But the same defenses can’t tell that traffic apart from a harmless wishlist bot pulling a photo and a price.

The practical result is that the link saves, but nothing behind it does. No photo, no name, no price — just a blank card. Most people don’t come back a week later to hunt down what that link was supposed to be.

What actually works: a screenshot

Open the product page in the Zara app or a browser, take a screenshot, and send it to the bot. An AI model reads the image the same way you would — brand, item name, current price, the old crossed-out price if there is one, and whether a size is showing as available.

This isn’t a workaround for the store’s defenses — nothing visits the site at all. It’s just looking at a picture that’s already sitting on your phone.

The three-price trick

During a Zara sale, three numbers often sit in a row — the original price, a mid-point, and the final price — next to a red badge like "−66%". A person reads that instantly. A model just sees a cluster of digits unless it’s been told what to look for.

So the model is trained on a simple rule: the final price is the smallest number that isn’t struck through, the old price is the largest struck-through one, and the percentage badge is never treated as a price at all.

If you also want price tracking

Worth being upfront about a limit here: tracking a price change needs a link, because the link is what gets checked again later. A card built only from a screenshot sits in your wishlist with a photo and a price, but it doesn’t update itself.

The fix is simple — send the link for the same item as a separate message. The screenshot builds a clean card, the link turns on tracking.

Why the store bothers defending itself at all

This isn’t personal, and it isn’t aimed at wishlist tools. Large chains get hit daily by automated requests from competitors pulling their prices and stock to undercut them within hours. The defense against that doesn’t look at who’s asking — it looks at how the request behaves: whether it carries the fingerprints of a real browser, whether the timing between clicks looks human, whether too many identical requests are arriving back to back.

The catch is that this kind of filter can’t distinguish a competitor’s scraper from a harmless saving tool — both look like code, not a person holding a phone. So every automated request gets caught in the same net, and no wishlist can honestly promise a permanent way around it. The defenses shift over time specifically so that yesterday’s workaround stops working.

How to shoot a screenshot that actually reads

The main rule: capture the product page itself, not a search results grid or your bag. A single item’s page shows everything at once — name, brand, photo, price, and the selected size if you’ve picked one. In a list or a cart, part of that information is cut off or missing entirely, and reading it gets noticeably harder.

Second, don’t crop tight around the price. If the old crossed-out number falls outside the frame, the model has no way of knowing it was ever there. An extra inch of screen in the shot beats a cropped-off digit every time.

One more small thing that helps: if you’ve already picked a colour or a size in the app, screenshot that exact view rather than the general product page with every variant showing. That way the card matches the item you actually want, not the closest guess.

Not every store fights this hard

Plenty of stores don’t lock things down this tightly — many mass-market retailers hand over their data to anyone who asks, and for those a plain link works with no screenshot needed. The difference isn’t company size; it’s how aggressively a chain fights scraping of its own prices.

Zara isn’t a worst-case example — it’s standard among large clothing chains selling at similar prices across many countries, none of whom want a competitor learning their pricing in real time.

What the model can and can’t see

Recognition only reads what’s visible in the image — brand, name, price, whether a size shows as in stock. It never visits the store’s site or sees anything beyond the picture you sent, which is exactly why the method works: it isn’t tricking any defenses, just looking where you were already looking.

That same limit sets the ceiling on accuracy: a cropped, overexposed, or tiny-print shot can be misread the way a person glancing at it for a second might misread it too. No one can promise perfect reading every time, so it’s worth checking the card rather than trusting it blindly.