The Thread: notes on shopping an outfit from a photo

How it works

Why does an outfit search sometimes return the wrong department?

A person wears an oversized men's shirt and the results show fitted blouses that look nothing like it. The garment is not the problem. The department it got filed under is.

Save to Pinterest

A woman screenshots an outfit: an oversized men's button-down, sleeves rolled, worn loose over jeans. The search comes back with fitted blouses that share a colour and not much else. The shirt itself was described correctly. The results just came from the wrong side of the catalogue.

This is a different failure from Snagfit describing the wrong garment altogether. Why the search picks the wrong garment covers what happens when layering or cropping hides the piece you actually wanted. Here, the garment is right and the fit is what throws the search off course.

What does it mean for a garment to sit in the wrong department?

Every retailer sorts its entire catalogue into sections, most commonly men's, women's, and sometimes a unisex or gender-neutral one, and each product lives in exactly one of those sections. That placement is a decision the retailer made when the product was built and photographed. It has nothing to do with who is wearing it in the picture Snagfit was given.

A men's button-down stays filed under menswear at that retailer no matter who wears it or how loosely. So a search built from a photo where a woman wears that same shirt oversized can end up pointed at a catalogue section the actual product never sat in, and the listings that come back are close in colour and type but built around a different typical shape entirely.

Why do retailers file the same style of garment two different ways?

Because the department is a catalogue decision, not a fact about the garment.

One retailer might sell a boxy button-down as part of its menswear collection. Another might sell what is close to the identical cut as an oversized-fit womenswear piece, styled on a different model, filed under a different section of the site. Both are real products for sale right now. Neither filing is wrong. They are just two separate business decisions about the same kind of shirt.

That is why two screenshots of what looks like the same style of garment, worn by two different people, can produce two unrelated sets of shopping results. The garment did not change. Which catalogue section the search leaned toward did.

What makes a garment read as unisex or borrowed rather than fitted?

Looseness relative to the body is the strongest signal, and it is visible in most photos.

A shirt, hoodie or jacket that extends past the natural shoulder line, hangs away from the torso instead of following it, or has sleeves bunched past the wrist reads as oversized. A straight, boxy body with no visible waist shaping reads the same way. A dropped shoulder seam, where the seam sits partway down the upper arm instead of at the shoulder joint, is a specific detail that almost always signals a deliberately loose cut.

Styling adds its own signal. A garment worn open over another layer, paired with pieces that read as a different department's usual pairings, or clearly several sizes larger than the rest of the outfit, all point toward a borrowed or unisex piece rather than a fitted one bought in the wearer's own size.

What should you do when the results feel off department?

Add a fit word to the search rather than trying a new photo of the same outfit.

Terms like oversized, boyfriend fit, relaxed, and unisex describe the silhouette itself, and shopping listings in both men's and women's sections use the exact same words for it. A retailer selling an oversized shirt in its womenswear section calls it oversized in the listing title, the same word a menswear listing for a similarly cut shirt would use.

Snagfit does not choose a department before it searches. It writes a description of what the photo shows, fit included, and searches shopping listings against that description. When the fit is described plainly, the search reaches listings across departments rather than defaulting to whichever one is more common for that garment type. Uploading a photo a second time will not change this. Rephrasing the follow-up search is what does.

Does a borrowed garment search differently from one bought in the wearer's size?

Not in how Snagfit reads the photo, but the fit itself is worth naming directly.

A borrowed garment, meaning something that belongs to a partner or a friend rather than being bought to fit the wearer, is almost always looser than that person's own size, since it was never chosen for their measurements. That looseness is exactly the kind of detail Snagfit records: how the fabric sits on the body, whether it follows the shape or hangs away from it.

Naming that looseness in a follow-up search rather than assuming the wearer's usual department is the faster route to the actual product, because the fit word is doing the work the department assumption was getting wrong.

Why do a fitted and an oversized version of one garment search so differently?

Because fit changes the words that describe the shape, and those words are what the shopping search is built from.

A fitted shirt reads as tailored, slim or tapered. The same style of shirt cut oversized reads as boxy, relaxed or dropped-shoulder. Those are different search terms even when the garment type, the colour and the pattern are identical, which is the reason two photos of what is functionally the same shirt design can return two sets of results that barely overlap.

This is also why checking a result against what to check before you buy matters more than usual on an oversized search: a listing can match the colour and the garment type while still being a completely different cut, and the product photo on the retailer's own page is where that becomes clear.

Is menswear terminology useful even when the wearer is not searching for menswear?

Often, yes, because a lot of the vocabulary that separates fit and cut by name lives in menswear listings.

Finding menswear from a photo goes through terms like structured versus unstructured shoulder and straight versus tapered leg, and those same words describe fit regardless of which department a garment ends up filed under. A shirt read as having a dropped, unstructured shoulder is describing the same physical detail whether that shirt is filed as menswear, womenswear or unisex at the retailer selling it. Borrowing that vocabulary in a search is often what finds the product a first search, built only around garment type and colour, missed.

What is the honest limit here?

Snagfit describes the garment it can see. It does not know, and cannot know, which department a specific retailer chose to file that exact product under.

What a fit-aware search can do is narrow toward listings that share the actual silhouette in the photo rather than the silhouette that is typical for that garment type. That gets you closer to the real product across more of the catalogue than a search that only names the garment and the colour, even though it still cannot guarantee which department the listing that matches will turn up in.

Common questions

Why did my search show clothes that look nothing like the photo?

This usually happens when the garment in the photo is loose, oversized, or styled in a way that does not match how that type of garment is normally photographed and sold. Snagfit describes what it sees, a type of garment, its colour, its material, its pattern and its fit, then searches shopping listings for that description. If the fit reads as unusual for the garment type, the search can pull listings from a catalogue built around a different typical fit.

What does it mean for a garment to be filed under the wrong department?

Most retailers organise their entire catalogue into men's, women's, and sometimes a unisex or gender-neutral section, and every product sits in exactly one of those sections regardless of how it is worn or styled in a photo. A garment sold as a men's button-down stays in the men's section even when a woman is wearing it in the picture Snagfit was given, so a search built from that picture can end up pointed at a section the actual product is not filed under.

Why do retailers file the same style of garment differently?

Because the department a product sits in is a decision the retailer made when they built the catalogue, not a description of the garment itself. A boxy button-down might be filed as menswear at one retailer and as a unisex or oversized-fit womenswear piece at another, depending on which collection it was designed for and which model wore it in the retailer's own photos. Two garments that look identical in a screenshot can be filed two different ways across two different shops.

What visual cues suggest a garment might be unisex or cross-department?

Looseness relative to the body is the biggest one. A shirt, hoodie or jacket that visibly extends past the shoulders, hangs away from the torso, or has sleeves bunched past the wrist reads as oversized rather than fitted, which is the fit description Snagfit records. Boxy proportions through the body, a dropped shoulder seam, and a straight hem that does not follow a waist are the other signs. Styling context matters too: a garment worn open over another layer, or paired with pieces from a different department, often signals a borrowed or unisex piece.

What should I do when the results feel off department?

Try a follow-up search that names a specific fit adjective rather than repeating the same photo. Terms like oversized, boyfriend fit, relaxed, or unisex describe the fit itself rather than assuming a department, and shopping listings across both men's and women's sections use those same words. A second search built around fit rather than around the garment type alone often reaches listings the first search missed entirely.

Can Snagfit search both departments at once?

Snagfit does not choose a department before searching. It writes a description of the garment from what the photo shows and turns that into a shopping search, so results can come from either department depending on which listings match the description most closely. When a garment is genuinely unisex, checking results from both a menswear-style search and a womenswear-style search, using different fit words in each, covers more of what is actually for sale.

Does a photo of someone wearing a partner's or friend's clothing search differently?

Not differently in how Snagfit reads it, but the fit itself is a strong signal worth using. A borrowed garment is almost always looser than the wearer's own size, since it was not bought to fit them, and that looseness is exactly the fit detail Snagfit records and writes into the search. Naming that looseness explicitly in a follow-up search, rather than assuming the department the wearer's usual clothes come from, tends to find the actual product faster.

Why do fitted and oversized versions of the same garment search so differently?

Because fit changes the words used to describe the silhouette, and those words are what the shopping search runs on. A fitted version of a shirt reads as tailored, slim or tapered, while an oversized version of the same shirt reads as boxy, relaxed or dropped-shoulder. Those are different searches even when the garment type, colour and pattern are identical, which is why two photos of what is functionally the same style of shirt can return listings that barely overlap.

Is this the same problem as the search picking the wrong garment entirely?

No, and the fix is different for each. Picking the wrong garment means Snagfit described a different piece of clothing than the one you wanted, usually because of layering or cropping, and the fix is a tighter crop around the piece you care about. Returning the wrong department means Snagfit correctly described the right garment, but the fit and styling pointed the search at a catalogue section the product is not actually filed under, and the fix is adding a fit word rather than recropping the photo.

Keep reading

How it works

How does a filter change what a search finds?

A filter is a colour edit someone chose on purpose, added on top of whatever the room's own light already did. Snagfit reads the finished picture, filter and all.

6 min read
How it works

Why does sheer and mesh fabric confuse a search?

Sheer and mesh fabric let the camera see through the garment, into whatever is underneath. Snagfit then has to guess where one layer ends and the next one starts.

6 min read
How it works

What does an AI see in an outfit photo?

Five fields per garment, eleven possible categories, and a search query four to eight words long. Here is the exact shape of what comes back, with a real example.

7 min read

← Back to Snagfit