MealsOn · Product case study · Growth

Sell meals,
not days.

Meal subscriptions are not an untried idea — they have been launched, shelved and relaunched by the largest players in the market. So the useful question is not what features to add to Scheduled Meals. It is why this category keeps failing, and what would make it structurally work.

Role

Product lead · end to end

Scope

Next 2–3 quarters, India metros

Output

Strategy, RICE, roadmap, prototype

01The reframe — this category has a graveyard

    2019

    Swiggy launches Daily, a standalone home-style meal subscription app

    2020

    Shelved. Demand dwindles through the Covid period

    2023

    Zomato enters the affordable home-style segment with Everyday

    2024

    Swiggy relaunches Daily in Bengaluru — inside the main app this time

Sources: public reporting on Swiggy Daily and Zomato Everyday

02Diagnosis — four reasons subscriptions break

01

Blocks activation

Commitment barrier

Paying for 30 days of food from a kitchen you have never tried is a large, unhedged bet. Most users never start.

02

Blocks renewal

Menu fatigue

The biggest churn driver in every subscription food business. By week three the user is bored of the rotation.

03

Blocks renewal

Life is not a calendar

Travelling Tuesday, at my parents' Friday. Rigid schedules break against real life, and every broken meal feels like a wasted payment.

04

The core tension

Flexibility is expensive

Kitchens need a forecast to prep against. Every skip and swap degrades that forecast and raises cost to serve. Not a user problem — an operating constraint.

Note the fourth one — it is not a user problem, it is an operating constraint. Any solution that ignores it will look good in a demo and lose money in production.

03Why it deserves investment — a margin play

Demand becomes predictable

Kitchens batch-prep against a known forecast instead of reacting to spikes. Higher utilisation, less waste.

Delivery routes get dense

Known drop points at known times cluster naturally. More drops per rider hour.

You acquire the order once

On-demand re-competes for every order. A subscription converts once and delivers many times.

The strategic argument

Every order shifted from on-demand to scheduled should carry structurally better contribution margin. That reframes the case: this is not a nice feature for busy people, it is a lever on unit economics — which is what earns roadmap space against competing bets. Stated as a hypothesis, to validate against real cost-to-serve data before committing spend.

04Who this is for — four users, one shared job
  • The time-poor professional

    “Stop making me choose every day.”

    Long hours, no energy to decide. Wants lunch solved without thinking.

  • The migrant or student

    “Food that tastes like home, at a price I can hold.”

    Away from home, no kitchen, price-sensitive.

  • The health-goal user

    “Keep me on my plan without the effort.”

    Tracking calories or protein, currently served by macro-tracked players.

  • The remote carer

    “I cannot cook for them, but I can make sure they eat.”

    Buying for elderly parents or new parents in another city.

The shared job: remove the daily decision, without removing control.

05Where the value leaks
DiscoverConsiderCommitConsumeRenew

Leak 1 · Consider → Commit

The user is interested but will not place a large, irreversible bet on an unfamiliar kitchen. They browse and leave.

Leak 2 · Consume → Renew

The user started, then hit menu fatigue or found the schedule did not fit their week. They finish the cycle and never come back.

Sequencing principle: fix the leaky bucket before pouring more in. Driving acquisition into a product that churns wastes the spend.

06The core idea

Buy 20 meals. Use them across 60 days.
On whatever days your life allows.

Kills the commitment barrier

Credits do not expire on a rigid calendar, so an unused day is not a wasted payment.

Removes the schedule fight

Travelling Tuesday? Don't spend a credit. No pause flow, no cancellation, no resentment.

Protects the forecast

Credits are still committed revenue and users plan ahead within the window, so the kitchen keeps a usable signal.

The trade-off I am accepting

Credits weaken the demand forecast compared with a fixed calendar plan. Mitigations: same-day cut-offs for scheduling and skipping; nudge users to plan a week ahead and reward it with a small credit bonus; cap unscheduled credits per window so demand cannot bunch; kitchen capacity buffers on high-variance days.

07Built, not just described

Skip a day and the credit returns to the wallet. On every competitor, that day's money is simply gone.

Live, clickable — six connected screens with real state. QA'd in a headless browser; five defects found and fixed.

08Prioritisation — RICE on stated assumptions
InitiativeReachImpactConf.EffortScore
Per-meal skip / reschedule82.090%114.4
Menu published ahead + swap82.085%1.59.1
Meal credit wallet93.070%36.3
3-meal trial pack72.075%25.3
Multi-address delivery51.090%14.5
AI preference engine72.560%42.6
Nutrition / macro goals41.565%2.51.6
Gift a plan31.570%21.6
Corporate plans23.050%50.6

Reach = relative share of the scheduled-meals base reached per quarter (1–10). Impact = 0.5 minimal to 3 massive. Effort = person-months. My estimates on stated assumptions — with real funnel data I would rescore Reach and Confidence first.

09Sequencing — fix the bucket, fill it, compound it

Phase 1 · Q1

Stop the churn

  • Per-meal skip and reschedule
  • Menu published a week ahead, with swap
  • Multi-address delivery
  • Cut-off times made explicit

Renewal rate · meal completion rate

Phase 2 · Q2

Open the funnel

  • Meal credit wallet
  • 3-meal trial pack
  • Onboarding preference quiz

Trial → paid conversion · plan starts per MAU

Phase 3 · Q3+

Compound the moat

  • AI preference engine live
  • Nutrition and macro goals
  • Gifting and corporate plans

LTV · contribution margin per order

10How we know it worked

North star

Weekly scheduled meals delivered

Volume, not subscriptions sold. A plan nobody eats is churn that has not surfaced yet.

Activation

  • Trial → paid conversion
  • Plan starts per MAU
  • Time to first scheduled meal

Retention

  • Cycle renewal rate
  • Meal completion rate
  • 30 / 60 / 90-day retention

Product health

  • Swap rate ↑ = personalisation failing
  • Skip rate ↑ = schedule fit failing
  • Credits expiring unused

Business

  • Contribution margin: scheduled vs on-demand
  • LTV : CAC
  • Revenue per subscriber

The two product-health metrics are the interesting ones — they are early-warning signals, not lagging outcomes.

11The trade-offs I am consciously making
  • Flexibility vs kitchen forecast

    More user freedom degrades demand prediction and raises cost to serve.

    Cut-off times, skip allowances per cycle, capacity buffers on high-variance days.

  • Upfront cash vs conversion

    A month paid upfront is good for working capital and bad for activation.

    Accept slower cash conversion for a materially larger subscriber base.

  • Personalisation vs choice

    Auto-curation removes decision fatigue but can feel like loss of control.

    Always editable, never locked. An onboarding quiz seeds the profile.

  • Delight vs focus

    Gifting, corporate and nutrition are good ideas that would dilute Phase 1.

    Deliberately deferred to Phase 3. Sequencing is the decision.

12How I worked — AI did the legwork

Research

Used AI to pull the category history — Swiggy Daily's launch and shelving, the 2024 relaunch, Zomato Everyday — then verified every claim before it went on a slide.

Pressure-testing

Argued the prioritisation against AI as a sparring partner: what breaks if flexibility is unlimited, where credits hurt operations. That is where the forecast trade-off surfaced.

Prototyping

Built the clickable prototype with AI assistance, then QA'd it in a headless browser across six flows with state assertions and screenshots.

What I did not delegate

The reframe, the prioritisation calls, the trade-offs I chose to accept, and what I left out. AI compresses the work; it does not make the decisions.

Stop selling a calendar.
Start selling meals.

↑ Top