A Growth Loop Driving 30% Repeat Users

We Spent More on Retargeting and Retention Got Worse

It took us an embarrassingly long time to notice the pattern. Every month: acquisition spend increased, retargeting budget increased, D30 retention stayed at 4%. The numbers moved in lockstep — and in the wrong direction for what we were trying to achieve.

Month 8 of this cycle, I finally pulled a scatter plot that nobody had thought to run: retargeting spend vs D30 retention, plotted by month. Correlation: essentially zero. We'd been spending more to get less.

The leaky bucket insight is cliché by now, but we were living it in full fidelity: pour in users, watch them leave. Respond by pouring in more. The problem wasn't the faucet. It was that we'd never asked what the bucket was actually doing to hold water.

The decision wasn't "spend more on retention." It was "design retention into the product" — because notifications and ads can remind someone to return, but they can't give them a reason.

The Retention Reality

What the Cohort Curves Showed

We pulled 12 months of retention cohorts. The curves were consistent to the point of being depressing — same shape, same slope, regardless of acquisition source, acquisition month, or how many times we'd run re-engagement campaigns.

Retention Day	Rate	Sessions/User	Revenue/User
D1	42%	1.0	₹0 (onboarding)
D7	18%	1.4	₹28
D14	8%	1.2	₹41
D30	4%	1.1	₹58
D60	2.1%	0.9	₹72

D7 users had 1.4 sessions in 7 days. Not engaged — drifting. And critically: the D7 users who were still around at D30 showed the same 1.1 session average. Even the retained users weren't really retained — they were just slower to leave.

The re-engagement campaign metrics: 3.8% click rate on weekly push notifications, 2.1% on email blasts. We were interrupting people at a cost and achieving almost nothing from it. The users who came back via notifications left again within 2 days. Reminding someone to return isn't the same as giving them a reason to.

The Exception in the Data

The 4% Who Stayed — and What Was Different About Their D1

Instead of studying why 96% left, we studied why 4% stayed. We pulled their D1 behavior and compared it to the users who churned within 30 days. The difference was striking and specific.

D1 Action	D30 Retention Rate	Context
Set a price alert	41%	Created a product-native reason to return (alert fires = return visit)
Saved a comparison	34%	Product now holds their work; return to use it
Invited a friend (D1–3)	38%	Social commitment; can't leave without explaining to friend
Used 3+ features on D1	28%	Depth of initial engagement predicts return
Read an article/guide	8%	Information consumption — no product hook attached
No core action on D1	3%	Never found the value; no reason to come back

Users who set a price alert on D1 had 41% D30 retention — 10× better than the baseline. The alert wasn't a notification feature. It was an engineering of a future return trigger. The user had effectively made an appointment with the product — and the product was going to call them when it was time.

The Insight

Retention Is Designed on Day 1

The data told us something counterintuitive: D30 retention wasn't determined by what we did at D14 or D21. It was determined by what happened on D1. By the time you hit Day 14 trying to win users back, the window had been closed since Day 7.

Notifications tell users to return. Price alerts create an event that pulls them back. One is a push; one is a pull. In the physics of user retention, pull always wins.

The redesign question became: how do we make every D1 onboarding create at least one product-native pull trigger? Not ask users to opt into notifications — give them something the product will do that requires them to return.

The Three Loops

Engineering Pull-Based Return Reasons

The Price Alert Loop

User sets alert → price changes → specific notification fires ('Your rate for Hero Splendor has dropped ₹340') → user returns to compare and decide → sets next alert. Self-reinforcing. Each loop creates the next one.

The Weekly Savings Report

Every Friday: 'Your saved comparison moved — here's the current best option.' Not a newsletter. Not a blast. A specific update on their specific data. Something changed that's relevant to them. That's what earns the open.

The Referral Return Loop

User invites friend → friend joins → user gets update: 'Ravi just saved ₹2,100 using your link.' Social progress report. Gives the referrer a reason to check in — and to share again. Referral becomes a retention mechanism, not just acquisition.

Loop	Return Trigger Type	D14 Retention	D30 Retention
Price Alert	Product event (price changed)	52%	38%
Weekly Report	Data update (comparison moved)	41%	29%
Referral Activity	Social event (friend action)	48%	34%
No Loop (control)	Push notification / retargeting	12%	4%

What Changed in the Product

The D1 Onboarding Redesign

The onboarding flow needed to do one thing above all else: get the user to create at least one loop trigger before they left. Previously, the hero D1 action was "browse plans" — passive, informational, creating no future pull.

Primary CTA Redesign

→Old hero action: 'Browse Plans' (passive, no hook)
→New hero action: 'Set Your Alert' — price alert setup as primary D1 objective
→'Save this comparison' made default upon comparison completion
→Referral nudge moved to within first comparison result view (not buried in settings)

Notification Quality Overhaul

→All calendar-based notifications disabled
→Only event-triggered notifications allowed: specific plan name, exact change amount
→Weekly report: uses user's actual saved comparison data — not generic market averages
→Price alert threshold: initially any change >0.1% → changed to >2% change (reduced fatigue)

Retargeting Cut

→Retargeting spend reduced 40% — freed budget reinvested into loop infrastructure engineering
→Remaining retargeting: only for users who set an alert but haven't converted (high intent)
→General re-engagement campaigns eliminated entirely

Rollout

The 10-Week Journey

Weeks 1–3 · Diagnosis

Mapping D1 Action to D30 Retention

→ Loop hypothesis validated. Engineering brief written.

✓Pulled 12-month cohort data; tagged D1 action for every user
✓Identified: alert-setters 10× better D30 than non-alert-setters
✓Defined 3 loops; mapped to existing product capabilities

→ Loop hypothesis validated. Engineering brief written.

Weeks 4–6 · Onboarding Redesign

Making Loop Activation the D1 Goal

→ D1 loop activation: 18% → 47% of new users

✓D1 flow redesigned: 'Set Your Alert' as primary CTA
✓Comparison save made default on completion
✓Referral visible within comparison result (not in settings)

→ D1 loop activation: 18% → 47% of new users

Weeks 7–10 · Infrastructure

Loops Running

→ D14 retention rising: 8% → 19%. D30 moving.

✓Price change notification system: 4-hour refresh cadence
✓Weekly savings report email: personalized per user's actual data
✓Referral activity feed in dashboard

→ D14 retention rising: 8% → 19%. D30 moving.

Week 10+ · Compounding

The Curves Start Changing

→ D30 retention: 4% → 18%. Retargeting spend down 40%. Revenue per retained user up 119%.

✓Loop 1 users upgrading to Loop 2 at 41%
✓Retargeting spend cut 40% — organic return rate increased
✓New D30 retention: 18% across full user base

→ D30 retention: 4% → 18%. Retargeting spend down 40%. Revenue per retained user up 119%.

Results

What Changed in 10 Weeks

4.5×

D30 Retention

4% → 18%

47%

D1 Loop Activation

up from 18%

−40%

Retargeting Spend

replaced by product-native pull

+119%

Revenue per Retained User

at D30

Metric	Before	After
D30 Retention	4%	18%
D14 Retention	8%	19%
Revenue per User at D30	₹58	₹127
Retargeting Spend	₹2.1L/month	₹1.3L/month
Organic Return Rate	18%	43%

Insights

What I Know Now That I Wish I'd Known Earlier

D30 retention is designed on D1, not D14

D1 action type predicted D30 retention with a 0.74 correlation in our data. By Day 7 without a loop trigger, users were effectively gone — the decision had already been made. Every retention investment at D14+ was catching users who'd already moved on.

Event-triggered notifications have 15× the impact of calendar-based

Calendar-based notifications (3/week): 3.8% click rate. Price alert fired (specific, event-driven): 61% open rate, 38% return visit. One is spam. One is a service. The production cost is the same.

Social loops create retention obligations, not just incentives

Users who referred a friend had 2.8× D30 retention. The social dimension isn't just about acquisition — it's about accountability. When someone you know joined because of you, leaving feels like letting them down.

Cutting retargeting spend improved retention metrics

We cut 40% of retargeting spend and redirected the budget to product engineering for loops. Retention improved. More retargeting spend doesn't fix a pull problem — only pull fixes a pull problem.

What Hurt Us First

The Mistakes We Made Before Getting It Right

✕ Price alert notifications fired too often

Why it broke

Initial threshold: any price change >0.1%. Users received 3–5 notifications per day during volatile market periods. Unsubscribe rate hit 34% in week 2. We'd built a spam machine, not a useful alert.

What fixed it

Threshold raised to >2% change AND a relevance filter (user's specific comparison category). Notifications dropped from 4/day to 1.4/week. Unsubscribe rate fell to 3%. Alert open rate improved 4×.

✕ Savings report email used market averages, not user data

Why it broke

'You could save ₹1,200 vs market average' felt abstract and untrustworthy. Users didn't know what 'market average' meant for their situation.

What fixed it

Changed to: 'Your saved plan for Hero Splendor has changed ₹340 since you compared. Here's the current best option.' Specific to their plan. Click rate improved 4× vs generic version.

✕ Pushing price alert setup too early killed initial browsing intent

Why it broke

Making price alert setup the hero CTA immediately confused users who came to browse first. 'What am I setting an alert for if I haven't seen the plans yet?' Exit rate increased during first test.

What fixed it

Browse-first flow: user sees plans, does initial comparison, then 'Save this comparison' CTA appears at natural completion. Alert setup as second action, not first. Same loop activation, less friction.

Future

Where the Loops Go Next

Predictive Price Alerts

Instead of waiting for prices to change, predict when they will. Seasonal patterns, provider renewal cycles, regulatory changes. Alert users proactively before the change — not reactively after.

Cross-Product Loops

Extend loop logic across categories. An insurance alert user showing car loan comparison behavior should get car loan alerts automatically. Cross-sell path as a retention mechanism.

Loop Health Monitoring

Real-time view of loop activation rates, loop return rates, and which loops are degrading over time. Enable the product team to intervene before retention metrics start moving — not after.

We didn't fix retention by spending more on notifications. We fixed it by building a product that gave users a genuine, specific, product-native reason to return — one that existed whether or not we sent them a reminder.

How We Increased Insurance CTA Clicks by 2X Using Behavioral Triggers

Growth

Optimizing Funnel Drop-Off Using Micro UX Interventions

Growth

From 3.8M to 45M+ Monthly Active Users: The CarInfo Growth Architecture

14 min read

Growth

How We Cut User Acquisition Cost by 30% Without Cutting Volume

11 min read

Growth

From 1K to 100K Monthly Users: Scaling MoneyRateFinder with SEO + Product Loops

9 min read

All Case Studies

Browse all 30 breakdowns

Work With Me

Let's build something together

Connect on LinkedIn

Follow for more insights

Where to go next

Consumer App Growth Consultant
The consulting work this sits inside, with the rest of the evidence behind it.
Day 30 retention benchmark, fintech & banking
Published figures, traced to their sources — and where the sources disagree.
Day 7 retention benchmark, e-commerce & retail
Published figures, traced to their sources — and where the sources disagree.
Product Growth Score
Thirteen questions that locate which part of the funnel is actually constraining growth.

Designing a Growth Loop That Brought 30% Repeat Users in 14 Days

Results at a glance

We Spent More on Retargeting and Retention Got Worse

What the Cohort Curves Showed

The 4% Who Stayed — and What Was Different About Their D1

Retention Is Designed on Day 1

Engineering Pull-Based Return Reasons

The D1 Onboarding Redesign

The 10-Week Journey

What Changed in 10 Weeks

What I Know Now That I Wish I'd Known Earlier

D30 retention is designed on D1, not D14

Event-triggered notifications have 15× the impact of calendar-based

Social loops create retention obligations, not just incentives

Cutting retargeting spend improved retention metrics

The Mistakes We Made Before Getting It Right

Where the Loops Go Next

You might also like

Where to go next

We Spent More on Retargeting and Retention Got Worse

What the Cohort Curves Showed

The 4% Who Stayed — and What Was Different About Their D1

Retention Is Designed on Day 1

Engineering Pull-Based Return Reasons

The D1 Onboarding Redesign

The 10-Week Journey

What Changed in 10 Weeks

What I Know Now That I Wish I'd Known Earlier

D30 retention is designed on D1, not D14

Event-triggered notifications have 15× the impact of calendar-based

Social loops create retention obligations, not just incentives

Cutting retargeting spend improved retention metrics

The Mistakes We Made Before Getting It Right

Where the Loops Go Next