You have seen something you like and you are not going to pay what it costs. The problem is turning a vague memory, or a low-resolution screenshot, into a search that returns the thing rather than four hundred unrelated results.
Here is a method.
Start by naming the attributes, not the item
The instinct is to search for what the thing is — "black boots", "green dress". That returns everything.
What actually narrows a search is the list of attributes that make this one distinctive. Work through five, in this order:
- Silhouette. The overall shape. Wide-leg, A-line, boxy, fitted, oversized.
- Length. Midi, cropped, ankle, knee. Be specific, because this is the attribute that most changes the character.
- Colour, precisely. Not "green" — sage, olive, forest, emerald. The difference between them is the difference between finding it and not.
- Material and finish. Ribbed, quilted, patent, satin, boucle, raffia, brushed.
- The one distinctive detail. The thing you would mention describing it to a friend. A square neckline. A chain strap. An exaggerated collar. A contrast sole.
Chain three or four of those together and the search stops being hopeless. "Sage green ribbed knit cardigan" finds things. "Green cardigan" does not.
Work out what is actually driving the appeal
Worth pausing on, because it changes what a successful search looks like.
Sometimes what you like is the shape. Sometimes it is the colour. Sometimes it is the way it was styled in the photograph and the garment itself is ordinary.
If it is the styling, you do not need that garment at all — you need the proportion, and you may already own something that gets you there. This is not a trick to avoid buying; it is genuinely the most common reason a purchase disappoints. The thing arrives and it turns out the appeal was the whole outfit.
Where to search
Retailer site search is usually poor at descriptive queries and good at categories. Use the filters rather than the search box — colour, length, material — and browse.
Search engine image search handles a screenshot well. It is strong on finding the original and weak on finding alternatives, because it optimises for visual similarity rather than for the shape at a different price.
Marketplace and aggregator search covers more shops in one query but returns a great deal of noise. The attribute chain above is what makes it usable.
Visual search in an app is the fastest route from a photograph to results, and the quality depends entirely on whether it is searching for a matching image or for a matching garment. Those give quite different answers.
Judge the results properly
Once you have candidates, three checks before you buy.
Check the composition. A similar shape in a very different fabric hangs differently and photographs the same. Reading a composition label covers this in ten seconds.
Check the fit language. Boxy, relaxed and regular are meaningfully different garments. The fit words retailers use translates them.
Check the colour against your own palette. A close match on shape in a colour that does not suit you is a worse buy than a looser match in one that does. If you do not know your palette, the free quiz takes two questions.
The honest limits
A few things worth saying plainly.
You will not always find one. Some things are distinctive because of a construction or a fabric that costs what it costs, and the alternatives will be visibly not the same.
Very cheap is sometimes exactly what it looks like. A large saving on something structured — outerwear, boots, tailoring — is usually explained by the materials, and the composition label will tell you which case you are in.
And a saving on something you would not otherwise have bought is not a saving.
The short version
Name the attributes, work out what is actually driving the appeal, search on the chain rather than the category, then check composition, fit and colour before buying.
That is the manual method, and it works. DUPI does the same job from a photograph — it reads the item, searches real UK shops for the same shape at a friendlier price, and gives you real prices and links. The dupe library is a browsable set of finds already put together this way, and our pricing study shows what the method actually yields: a median saving of 83% across 44 looks — and the categories where it barely closes at all.