Comparison articles about visual search apps tend to have one thing in common: the author picked the photos, the author picked the winner, and you cannot check either. This is the opposite. It is the protocol we use on ourselves, written so you can use it on us. It takes about ten minutes, needs no account beyond the apps you already have, and ends with a scorecard you can defend.
Step 1: pick three photos you would actually search
Do not pick flattering photos. Pick the three kinds of image that show up in real life, and use the same three in every app you compare.
- A clean product shot. A retailer image or a well-lit photo of the item alone. This is the easy case; an app that struggles here struggles everywhere.
- A person wearing it. An Instagram or TikTok screenshot with several pieces in frame, cropped the way you would crop it in a hurry. This is the case that matters most.
- A hard case. A paused video frame with motion blur, a vintage or second-hand piece, or a photo where the brand is genuinely unknowable. This is where you learn whether the app is honest.
When we ran our own battery of sixteen photos, the hard cases taught us the most: the museum handbag nobody sells, the hockey jersey whose look-alike was from a different team. You can read that run in full in what came back from 16 real photos.
Step 2: search the same piece in every app, the same day
Fashion search results depend on region, on the day's inventory and on which garment in the picture you point at. Keep those constant. Use the same region setting or the same phone, search on the same day, and pick the same garment in each image. If one app lets you choose a piece and another decides for you, note that; it is part of the result.
Step 3: score the top result on five questions
Only the top result counts. A wall of forty look-alikes is not help; it is homework. For each photo and each app, answer these five questions with yes, no or partly, and write the answers down.
| Question | Why it matters |
|---|---|
| 1. Is the top result the same product, not just the same look? | A similar dress at a different price is a different purchase. |
| 2. Does the app tell you how sure it is? | Remode labels every result exact, best visual match or similar. An app that gives no label is asking you to guess. |
| 3. Can you buy it where you live? | A perfect match at a store that does not ship to your country is a photo of a product, not an offer. |
| 4. Is a second-hand option shown when one exists? | For sold-out or expensive pieces, resale is often the only real path. |
| 5. Does the price on the result match the store page when you open it? | Prices are scraped and cached; a mismatch is a warning sign. In our own run, two of 291 listings showed impossible prices, and we published them. |
Step 4: check what happens after you leave
Finding something is half of a shopping decision. If the app can save the item, save it, then see what it does. Does it watch the price, tell you about a restock or a second-hand listing, or does it forget you the moment you close it? Remode watches saved pieces in the background for free; other tools may not offer this at all, which is fine, as long as you know before you rely on it.
Step 5: keep the tool that is honest with you
Add up the yeses if you like, but the real decision is simpler. The app worth keeping is the one whose labels you can trust when you are about to spend money. An app that calls every look-alike a match will occasionally look brilliant and will eventually cost you a return. An app that says "similar" when it is not sure looks less impressive in a screenshot and saves you money in a year.
What a fair result looks like
Expect no app to be perfect on the hard case, and be suspicious of one that claims to be. Expect region to change the winner: the same sneaker search that returned a Hibbett listing for the US returned Decathlon for Switzerland in our run. Expect timings between a few seconds and half a minute; a search that returns instantly is usually returning a cache of look-alikes rather than checking stores. And expect the result to change next month. That is why we date every test we publish.
Run it on us
The browser trial gives you three Remode searches with no app and no account, which is exactly the size of this protocol. Take your three photos, run them, and compare the labels with what your other tools say. If we come out worse on your photos, we want to know: hello@remode.app, with the screenshots attached. Our own published run, method and raw data are on How we test.
Three searches, no app, no account.
Run your own test in the browser, then decide whether the full app on iPhone earns a place on your home screen.
