Shopping

What shopping apps are worth building?

Shopping apps earn their place when they cut a decision short. Three questions recur in shopping ideas: what is this item and where can I get it, have I already got one, and is this price really a saving? Each one needs a different input and a different source of truth, and that choice decides what you can build in a weekend.

• Identify: an item from a photo, barcode or screenshot, returned with a name and a search term the shopper can use.
• Own: a household or wardrobe list the user fills in once, then checks against a photo taken in a shop.
• Verify: a price history view that compares a sale price with the prices the user saved in the last few months, built from their own entries when no public feed exists.

Who shops with these apps, and how do you reach them?

A shopper with a one-off question is unlikely to create an account first. That points the first version at a browser page, a bookmarklet or a mobile web page rather than a store listing. Groups with a repeated question make the best first audience: people furnishing a flat, parents buying for children, and people who buy secondhand clothes. Reach them where they already ask the question, in the comment threads under a product post and in hobby groups that trade items.

Test interest by posting a photo of an item you are considering to two relevant groups, with the question below. Count the replies that name an item the poster already owns, since that shows the question is real.

I’m thinking of buying this. Before I do, would it help to know whether you already own one? Tell me what you have at home.

What makes a shopping idea good?

A shopping idea is good when its answer arrives fast enough to use in the shop and can be checked against the item in hand. Check each idea against this list:

• It returns an answer on one screen while the shopper still holds the item.
• It shows its source, so the shopper can see why a price counts as a sale.
• It starts from a photo or barcode, not a long form.
• It does not depend on a retailer’s private data or on a method the retailer forbids. Read the terms of any source before you rely on it.
• Any money it makes is visible to the shopper, such as a paid tier or a labelled link, and never a ranking the shopper cannot see.

What does a small first version look like?

Hypothetical example: a do-I-already-own-this check for a home kitchen. Once, the user lists about twenty kitchen items and adds a photo of each. In a shop, they photograph a product and get a yes, a no or a close match from the list, with the saved photo beside it so they can confirm the match. Version one has no price data and no retailer connection.

Test it by giving the list to five people who shop for a household and checking whether they open it in the shop, not at home. If they stop before the shop, the matching step is the first thing to fix. A shopper who uses it twice in a month has given the first real signal.

What goes wrong with shopping apps?

• Retailer terms: scraping prices or product pages can break a site’s terms of use. Read them first, and prefer a feed the retailer offers or the user’s own input.
• Stale prices: a price shown without a date is a guess by the time the shopper reads it. Show when each price was seen.
• Undisclosed commission: if a link earns the app money, say so beside the link. A shopper who finds a hidden commission has little reason to trust the next answer.
• Authenticity claims: an app that says an item is genuine or safe makes a claim it usually cannot check from a photo. Say what the match is based on, and nothing more.
• Matching errors: similar-looking items are common on shelves. Show the match and let the shopper confirm it.

Naming matters too. An app with a name close to a large store or marketplace will be confused with it, so pick a name that describes the question the app answers.

Where should you start with a shopping idea?

Three ideas in this category were on the board when this page was written. Was this actually on sale? is the price-check question; a first version could rest on prices the user saved themselves. Point at a product and see if you already own it is closest to the kitchen example above. Take a photo of someone’s outfit and instantly find where to buy it depends on a matching step that is hard to get right, so test the matching before anything else.

Your question not among them? Submit it, or skim the whole board for patterns worth borrowing. Before building, run the validation questions, then cut the build to one output with the MVP guide.