Zepto: adding to an order that's already on its way
A teardown of the moment you remember one more thing after checkout.
Designs a narrow, time-boxed way to add to an existing order, and is explicit about the operational limit that makes it work at all.
What it is
The order left the app before it left the store.
You check out on Zepto, then remember the milk. At ten-minute delivery the order may already be packed — so this design exercise asks whether an 'add to my order' window can exist at all, and what makes it survivable for operations.
The feature
A window, honestly drawn
- 1Checkout
Order placed. The picker starts almost immediately.
- 2"I forgot the milk"
The most common post-checkout regret.
- 3Live countdown
The app shows exactly how long the add-on window lasts. A clear yes or no, not a maybe.
- 4Add from the same store
Only items already in the picker's path qualify.
- 5One bag, one rider
No second fee, no second delivery.
The real constraint
An ops problem wearing a UI problem's clothes
At ten-minute delivery, un-picked time is a minute or two — not the five-to-ten other apps can afford.
Re-touching a packed bag slows every order behind it in that store.
Dark-store economics run on picker speed. A feature that quietly taxes it has to earn its keep.
Tradeoffs
What I'd ship, and what I'd refuse
A short, honest add-on window shown as a live countdown
Telling the user exactly how long they have turns a maybe into a clear yes or no, and it sets the right expectation instead of a false sense of control after the window has already closed.
Cap the add-on to items already in the same store's inventory
Adding something not yet picked in that run means a second trip inside the same store, which erases the whole point. Keeping it to what's already in the picker's path is what makes the window possible at all.
No extra delivery fee inside the window
The value to the user is exactly that this isn't a second order. Charging anything reframes it as one, and the goodwill of solving the 'I forgot milk' moment for free is worth more than the marginal fee.
“The metric is add-ons completed inside the window, not attempts. A high attempt rate with low success just means the window is lying to people.”