for real.

Brands over-produce: 2.5–5 billion items went unsold in 2023.¹ In the US that surplus built TJX, a $56 billion business.⁶ In India it built an offline chain and no app.

The customer exists: urban, 22–34, brand-aware, sale-trained, living in Zomato, Blinkit, Myntra and Nykaa. But one in five has received a fake online, and fewer than half got their money back.¹⁶ At 70% off, a famous logo reads as a scam.

The reframe: at 70% off, the discount isn't the product. Belief is. A first-time buyer must believe three things without touching the garment: it's real, it will fit, and they'll get their money back if not.

Five bets followed: authenticity as a system, fit before feed, honest scarcity, a checkout that speaks Indian, and recourse you can see.

Nike: $9.7 billion of inventory in Sept 2022, up 44%, with a CFO promising to “more aggressively liquidate”.³ H&M: $4.3 billion unsold in 2018.⁴ Adidas: €1.2 billion of Yeezys.⁵

In the US, that surplus is an industry. TJX sells brand names “20% to 60% below full-price retailers”: $56.4 billion across 5,085 stores.⁶ Online, Otrium does it for 400+ brands in Europe.²

India had the offline version, Brand Factory, 100 stores by 2018 and now Reliance's Fashion Factory,⁷ and no app built for it. Branded off-price lived in sale tabs and grey-market Instagram pages.

Young and urban. India has 270M+ online shoppers; Gen Z is ~40% of them and spreads spend across five or more apps.⁹ Half of Myntra's active base is Gen Z.¹⁰ Gen Z drives 47% of India's fashion spend and 50% of footwear.¹³

Trained by the apps next door. Zomato (20.6M monthly customers) and Blinkit (10.2M) set the bar for clarity and speed.¹¹ District made going out playful.¹² Nykaa delivers beauty in 30–120 minutes; Myntra in 30.^{14, 10} None are competitors. All are the bar.

Brand-aware, not loyal. 1.7× more likely to pick a trend over a brand; 8 in 10 ask a friend first.¹³ The logo is a quality signal, not a religion. That's exactly what an outlet serves.

Sale-trained. Myntra's End of Reason Sale doubles orders.¹⁷ The sticker price is a suggestion; the sale is when you buy. Our pitch is a sale that never ends, which is precisely what makes it hard to believe.

Three weeks, lean: 14 interviews, 9 shop-alongs on Myntra and Ajio, a 240-response survey, 11 app teardowns.

“Is it real?” 12 of 14 raised it before I did. One in five Indian shoppers got a fake last year; only 40% got a refund.¹⁶ And trust fell as the discount rose: fine at 30%, hesitant at 50%, “what's the catch?” at 70%. Our best number was our biggest objection.

“Will it fit?” 11 of 14. India returns 25–40% of fashion bought online, mostly for size.¹⁵ Nobody is “a medium”; they're “a 32 in Levi's, a 30 in Zara”. Forty brands in one feed multiplies that, and with one unit left, you can't order two sizes.

“Will I get my money back?” 9 of 14, and the most emotional. Cash on delivery is trust insurance for a first order.

The quiet one: speed. Overstock ships from brand warehouses, so the honest promise is 7–14 days. We'd say it plainly, early.

At 70% off, the discount isn't the product. Belief is.

Three beliefs, multiplied: it's real × it will fit × I'll get my money back. Any one at zero and the sale is zero. A badge can't carry that; a system can.

Two more rules. Outlet stock never restocks, so urgency is real, and must never be faked. And the things we can't fix, delivery speed and the age of the stock, get said plainly. Honesty was cheaper than surprise.

Four tabs, Home, Categories, Account and Bag, in a floating bar that stays through browse-to-buy. Search and Wishlist are header modes, not destinations. Fewer tabs than Myntra, because a few thousand styles don't need five; the bag badge gets the attention instead.

Gender is a context set once and switched at the top of Home. The product page is the belief engine. The bag is one screen. Orders is where recourse lives.

Sign-in is mobile number and OTP. Two skippable steps follow: name and gender, then sizes. The whole app browses without login. The one thing onboarding pushes is size, because size is what makes the feed worth looking at.

On the name. DealStreet shipped as for real: what a customer says when they see Zara at 70% off, turned into the answer. The account menu links to “How for real works”; the certificate carries the wordmark. The name is the promise.

Every competitor uses a shield icon and the word “Authentic”. Research said that stopped working. So: a chain, each link answering a different version of the question.

Before you buy. A brand tick with a rating count. A trust row under the price. Our Promise: “If product is not original, we will refund you the amount.” A guarantee with money behind it.

While you look. The certificate is photo 5 of 5 in the carousel. The first version was a tab; nobody opened it. Moved into the swipe, five of six testers found it unprompted.

After delivery. Rating asks one binary question, was the product original?, so it can be counted, and the count shows on every product from that brand: “‘Original’ · 105”. Not our word. Other buyers'.

If it goes wrong. “Product is fake” is a first-class return reason. Listing it is a promise we expect to be held to.

Ask once, by garment. Onboarding takes topwear, bottomwear and footwear sizes, which is how people actually know their bodies. Skippable, editable from Account.

The exploration that failed. Version one was a three-column wheel picker. Compact, elegant; three of five testers set the wrong shoe size. The chip grid that replaced it is less clever and unambiguous. It taught the team early: this customer won't pay a learning-curve tax.

One tap, everywhere. An “in my size” toggle in every header filters the whole feed. Nobody guessed the icon, so a first-run tooltip says what it does.

Brand size, translated. The size chart maps the brand's label to a chest or waist measurement. It's the only way to compare a Zara M to a Benetton M. Stock per size sits under the chips, so it never sends you to a size that's gone.

Indian e-commerce runs on fake urgency, and customers have learned to ignore it. Our stock is genuinely finite. The rule the founders signed: every urgency signal comes from live data. Stock from inventory, countdowns from real deal windows, nothing generated.

Show what you missed. Sold-out items stay in the grid, dimmed, labelled “You missed this”. In testing that did more for urgency than any timer: “okay, these actually go.” The wishlist has a “Deals you missed” section with one line: “Don't wait for a restock. There won't be one.” The most screenshotted sentence in testing.

Where the line is. Removing from bag offers “Move to Wishlist” first. The countdown only appears on real deal windows. When a product sells out, the bar says “You Missed This Deal”. No “notify me”, because nothing is coming.

The bag tells you what you can return. Outlet stock has mixed policies, so each item carries its own badge: “14 days exchange”, “7 days return”, “Not returnable”. Four of six testers had asked “can I return this?” in the bag. Afterwards, none.

The window before the button. “Standard Shipping | 7–14 Business Days” sits under the address, above the bill. It costs some orders. Discovering it on the confirmation screen costs trust, which is worse.

A coupon moment. Ticket-shaped coupons, a small celebration on apply, and “₹6,201 saved on this order” in green under the total. For this customer the saving is the headline.

Cash on delivery stays. The founders would have dropped it. Research said it's first-order insurance. It sits beside online payment, no penalty.

“Will I get my money back?” is answered after the order, so that's where the design time went. Every item on the order shows its window as a date: “Return/Exchange open till 28 Nov 2025”.

One pattern, three flows. Returns, exchanges and cancellations share a stepper, Items → Reason → Review, with days remaining pinned on top. Reasons are in the customer's words. Review shows pickup slot and refund amount, method and timing before submit.

Exchange inline. Pick a replacement size with the same stock-aware chips. If your size is gone, the flow offers the return path instead of a dead end.

Rating that does a job. Each brand in an order is rated separately, and the answers become the proof counts on every product page.

Colour. Coral marks the primary action, the active tab, the discount chip and the promise. Nothing else. Emerald is money coming back to you. Red is errors and “You missed this”. When the accent has one meaning, three coral elements on a screen means three things worth doing.

Type. One sans, five sizes, two weights. Product names larger than brand names: the garment is what you buy, the brand is why you trust it.

Components. Chips carry sizes, with stock underneath wherever a size appears. Bottom sheets carry every decision that doesn't deserve a page. One stepper for returns, exchanges and cancellations. Coupons are tickets. Status is a colour and a word, never colour alone.

States. Every error names the problem. Every list has an empty state; the wishlist's is a cat. Success screens count down and return you to where you were.

These figures are directional and rounded: reconstructed from what the team tracked at the time, not audited analytics. Benchmarks are cited where they exist.

The two that matter: return rate, which the whole fit strategy was aimed at; and the “not original” flag rate, the only honest measure of whether the authenticity system worked.

Instrument the belief loop from day one. “Was it original?” became the company's early-warning system for suppliers. It should have been dashboarded before launch, not found in a spreadsheet in week six.

Earn a faster tier. Honesty about 7–14 days worked, but the customer lives in a thirty-minute world. Forward-stock the top 200 styles in one metro and test a two-day tier.

Seed proof before launch. A count that starts at zero is a claim with no evidence. A closed pilot would have given every brand a real number on day one.

Do again. Write the reframe on the wall before drawing a screen. Test the elegant idea first so it can die early. Put the uncomfortable truths in plain sight. Treat what happens after payment as the product.