Zepto: App Notifications Teardown

A product teardown of push and in-app notifications by Shivam Bajaj. Built only from public sources; I have no access to Zepto's internal data.

Overview

Zepto is an Indian quick-commerce company that delivers groceries and other daily needs from a network of dark stores. In its October 2025 round it raised $450M at a reported $7B valuation (TechCrunch).

1,139

dark stores as of 31 Mar 2026, per press coverage of its draft IPO filing

66

cities, same source

~48M

annual transacting users (4.79 crore), same source

Competition is intense: estimated order shares put Blinkit well ahead, and Zepto's monthly-user growth has been reported to slow against Blinkit and Instamart (Fynd/Entrackr; IndMoney filing analysis). When growth slows, getting existing customers to come back becomes a bigger lever, and notifications are one of the cheapest ways to do that. Zepto has also been reported to be shifting its retention approach toward fewer discounts and a paid club (Inc42).

Store, city and user figures are as reported by secondary coverage of Zepto's draft IPO filing. Check them against the filing before quoting.

Personas

Illustrative archetypes, not real people. Pain points are linked to a source where one exists; the rest are my hypotheses.

The daily top-up shopper

Who: Orders milk, bread and vegetables most mornings

Goals: Get the usual basket fast, at a predictable time

Pain points: Hypothesis: generic promotions crowd out the reminder that would actually help

Opportunity: A "your usual basket" nudge at the habitual time

The late-night snacker

Who: Orders late evening, often after work or a party

Goals: Get a craving fixed in minutes

Pain points: Hypothesis: alerts at odd hours feel intrusive and get muted

Opportunity: Quiet hours and relevance-first timing

The lapsed user

Who: Has not opened the app in a month or more

Goals: A reason to come back that is worth their time

Pain points: A generic blast is easy to ignore. Zepto reportedly found recently active users respond better to push (AppsFlyer)

Opportunity: A win-back message tied to what they bought before

The privacy-sensitive user

Who: Shops for personal items and dislikes feeling watched

Goals: Control over what the app says about them

Pain points: A reported personalisation misfire raised trust questions (exchange4media)

Opportunity: Exclude sensitive categories and offer clear notification controls

Why Notifications Matter in Quick Commerce

What they do

  • Push notifications reach users outside the app, so they can be timed and tailored. Typical uses are cart reminders, restock reminders and offers.
  • In-app notifications appear while the user is already browsing, such as banners and order-status nudges.
  • Cart-recovery pushes are widely described as the highest-return use, often sent in a sequence over about a day or more (Venn guide).

Benchmarks (use with care)

  • An India benchmark (MoEngage, 2020, media and entertainment, not grocery) reported a 6.3% push click-through rate, with personalised messages at 6.07% versus 3.67% for non-personalised (MoEngage).
  • Globally, only slightly more than half of iOS users opt in to push in e-commerce (Pushwoosh).
  • Vendors calculate click-through differently, so benchmarks across vendors are not directly comparable.

What Zepto Is Reported to Do

I could only find a few public data points, so this section sticks to those.

  • Segmentation by recency. Zepto separated recently active users, since those active in the last seven days responded better to push, and noted it needed a deeper understanding of preferences to personalise content (AppsFlyer case study).
  • Attribution into CRM. It fed campaign attribution data into its internal BI platform to enrich CRM data (same source).
  • Deep linking. It used deep linking to take users from outside the app to the relevant screen (same source).
  • Result. The vendor reports about a 35% lift in remarketing conversions. This is vendor-published, so treat it as indicative.
  • Where it can go wrong. Trade press covered a reported personalisation misfire, raising questions about how far personalised marketing should go (exchange4media).

User Experience: Notification to Landing Page

Illustrative wireframes of the journey. They are not screenshots of Zepto's app, and the copy is placeholder text.

9:41TuesdayQuick-commerce appYour cart is waitingMilk and 2 more items are still in your cartQuick-commerce appTime to restock?You usually reorder eggs about nowQuick-commerce appOffer in your areaPlaceholder offer text

1. Lock screen: three common notification types (cart, restock, offer)

←CartOpened from your reminderMilk 500 mlIn stockItem 2In stockTotal, all-in₹ xxCheckout

2. Deep link opens the cart directly with items in stock, not the home screen

←NotificationsOrder updatesAlways onCart remindersMax 1 per dayRestock suggestionsWeeklyOffers & promotionsOffQuiet hours10 pm – 8 amSave

3. Proposed: a preference centre with frequency caps and quiet hours

What a good journey needs

  1. Right segment: send to users likely to act, such as recently active ones.
  2. Right content: reference something the user did, such as their own cart, and avoid sensitive categories.
  3. Right timing: respect time-of-day patterns and a daily and weekly cap.
  4. Right destination: deep link to the exact screen, with live stock and price.

Proposed Mockups

Three screens for the opportunities below. Illustrative layouts with placeholder copy, not Zepto's design.

Stay in the loopGet what mattersOrder status, delivery updatesand reminders you choose.✓ Order updates✓ Cart reminders (max 1 a day)Turn on notificationsNot now

Value-first permission prompt: explains what the user will get before the system dialog, aimed at opt-in rate

←InboxAllOrdersOffersOrder on the wayArriving soon2mItems still in your cartTap to continue1hWeekly restock ideaRead1dManage notifications

In-app inbox grouped by category, so missed pushes can still be found

←CartOne item sold out at your storeMilk 500 mlIn stockItem 2Sold outSee similarRemoveCheckout

Deep link with live stock check and a clear fallback when an item is sold out

Try the Prototype

A working prototype of the proposed notification controls. Send test campaigns and watch the rules decide what reaches the phone. It runs in your browser with sample data and a simulated clock. It is not Zepto’s app, and I have not seen its settings beyond public reports.

9:41 AM5G ▮▮▮

What to try

    Switch off to see today’s behaviour as I understand it: everything is delivered, and a sold-out link is a dead end.

    Send a test campaign

    What the rules did

      Opportunities

      My suggestions, not claims about Zepto's roadmap.

      IdeaWhyRisk to watch
      Notification preference centre by category, with quiet hoursGives users control and reduces blanket opt-outsFewer promotional impressions; offset by better opt-in retention
      Daily and weekly frequency caps across all campaignsPrevents several teams stacking messages on one userNeeds a central send-orchestration layer
      Sensitive-category exclusion rulesPersonalisation misfires hurt trust more than they help conversionRules must be reviewed regularly
      Deep link with live stock checkAvoids sending users to an item that is out of stock at their storeExtra latency on tap
      Holdout groups to measure true liftShows whether pushes cause orders or just claim orders that would have happenedShort-term revenue left on the table

      Metrics

      Funnel

      • Delivered rate = delivered / sent
      • View rate = viewed / delivered
      • Click-through rate = clicked / viewed
      • Add-to-cart rate within a short window after a click
      • Conversion to order within 24 hours

      Quality

      • Time on the landing screen after a tap
      • Share of users who tap zero notifications in a day or week
      • Clicks per notification received, per user per day and week
      • Incremental orders versus a holdout group

      Guardrails

      • Opt-out and uninstall rate after a send
      • Notification permission opt-in rate
      • Complaints or negative feedback mentioning notifications
      • Revenue per message, to track send cost against return