Cutting User Acquisition Cost by 30%

We Were Paying ₹340 to Acquire Users Who Cost ₹40 to Serve

At ₹340 blended CAC, the unit economics didn't work. The average first-month revenue per user was ₹280. We were losing money on every new user before they had a chance to become profitable. The growth team celebrated user count milestones. The finance team quietly flagged that we'd hit ₹2.8 crore in acquisition spend for ₹2.1 crore in first-month revenue.

What made it worse: nobody had decomposed the ₹340 number. It was a blended average. Some channels were delivering users at ₹80. Others at ₹780. We had no idea which was which — or why — because we hadn't built the attribution infrastructure to find out.

The acquisition spend wasn't the problem. The absence of channel-level unit economics visibility was the problem. We were averaging our way to the wrong answer — blending highly efficient channels with catastrophically inefficient ones and treating the average as the truth.

Six months later, blended CAC was ₹238. Revenue per acquired user in month 1 was ₹295. The business was cash-flow positive on first-month revenue for the first time. Here's the full breakdown.

Building the Attribution Layer

The Channel-Level Truth

Before doing anything else, I spent 3 weeks building proper attribution. UTM discipline, install-to-event tracking, first-purchase attribution at channel + campaign + ad-set level. The numbers, when they came in, were striking.

Acquisition Channel	Monthly Spend	Installs	CAC	D30 ARPU	Unit Economics
Google App Campaigns (brand)	₹18L	22,500	₹80	₹340	✓ Profitable D30
Google App Campaigns (generic)	₹34L	19,000	₹179	₹210	Break-even at ~D45
Meta – Lookalike (custom)	₹12L	6,800	₹176	₹280	Profitable D30
Meta – Interest targeting	₹28L	6,200	₹452	₹190	✗ Never profitable
YouTube pre-roll	₹14L	3,900	₹359	₹180	✗ Never profitable
Influencer + affiliates	₹8L	1,200	₹667	₹110	✗ Worst performer

Interest-targeted Meta ads (₹28L/month spend) were producing users with ₹190 ARPU who cost ₹452 to acquire. We were spending ₹262 more than we earned per user, at scale, every month. This channel had been running for 9 months. Nobody had computed this because channel-level ARPU hadn't been tracked against channel-level CAC.

The attribution work revealed a clear pattern: channels that brought in high-intent users (brand searches, custom lookalikes from our best-value users) had fundamentally different economics than channels that brought volume from broad targeting. The problem wasn't the channels — it was the absence of economics discipline in how we allocated between them.

The User Quality Problem

Not All Users Are Worth the Same CAC

Beyond channel analysis, I pulled user quality cohorts by acquisition source. The 90-day LTV distribution told a more nuanced story than simple D30 ARPU.

User Segment (by source)	D30 ARPU	D90 ARPU	Churn by D90	True LTV Signal
Brand search (high intent)	₹340	₹680	28%	Strong — deep engagement from day 1
Custom lookalike (best users)	₹280	₹590	31%	Strong — similar profile to power users
Generic app campaigns	₹210	₹340	54%	Medium — converts but shallow
Interest targeting	₹190	₹220	71%	Weak — churns before second transaction
Influencer/affiliate	₹110	₹130	82%	Minimal — one-time curiosity install

The interest-targeting user wasn't just more expensive to acquire — they also churned 2.5× faster than the brand-search user. The CAC disadvantage was compounded by a lower lifetime, meaning we were paying more for users who stayed less. These were two-sided unit economics failures, not one.

This analysis led to a strategic decision: stop optimising for volume and start optimising for value. The target wasn't a lower number of users per rupee. It was a higher revenue per rupee of acquisition spend — which meant being willing to pay more per install on channels that delivered high-LTV users.

The Reframe

CAC Is the Wrong Metric. LTV/CAC Is the Right One.

The growth team had been optimising to minimise CAC. That's the wrong objective function. A user acquired at ₹450 who generates ₹1,800 over 12 months is more valuable than a user acquired at ₹150 who generates ₹180 and churns. One has LTV/CAC of 4.0; the other has 1.2.

The goal isn't a lower CAC number. The goal is a higher LTV/CAC ratio. These look similar but point to completely different optimisation strategies.

The reframe changed every downstream decision: which channels to scale, which to cut, what the landing page needed to communicate, how tight the funnel qualification should be.

Specifically: we stopped bidding to maximise install volume on brand campaigns (which were already efficient) and started bidding on value signals — users whose profile matched our high-LTV cohort. We stopped buying YouTube and interest-targeting Meta at scale, and reinvested that budget into custom lookalikes built from the top-20% LTV user pool.

The Four Levers

How We Moved the Number

Lever 1: Channel Reallocation

Cut interest-targeting Meta spend from ₹28L to ₹6L/month. Cut YouTube from ₹14L to ₹3L. Reinvested ₹33L/month into custom lookalikes and brand search expansion. Blended CAC dropped without reducing install volume.

Lever 2: Landing Page Qualification

Rewrote ad creative and landing pages for intent-qualification: 'CarInfo is for people who own vehicles and need [specific task]'. Generic curiosity traffic filtered out. Conversion rate on landing → install dropped from 22% to 16%, but quality jumped: D30 ARPU of acquired users rose ₹60.

Lever 3: Organic + Referral as CAC Offset

Built referral mechanic tied to insurance renewals (₹200 wallet credit). Referral installs have near-zero CAC and LTV matching the referring user profile. At peak: 22% of monthly installs came from referral. Blended CAC calculation absorbs this volume.

Lever 4: Funnel Conversion on High-Intent Channels

Brand search campaigns had ₹80 CAC but the install → first session → first transaction funnel still had a 34% drop-off at onboarding. Fixed onboarding flow for this segment: deep-link to specific lookup type based on ad creative. Drop-off reduced from 34% to 18%.

Rollout

The 6-Month Execution

Month 1 · Foundation

Attribution Infrastructure Built

→ Full attribution visibility. Channel economics mapped.

✓UTM discipline enforced across all campaigns
✓Install-to-event tracking implemented via Firebase + analytics layer
✓Channel-level ARPU computed for first time; 6 channels profiled
✓Unit economics document shared with growth team and leadership — reset expectations

→ Full attribution visibility. Channel economics mapped.

Month 2–3 · Reallocation

Budget Shifted to Efficient Channels

→ Blended CAC: ₹340 → ₹278. Install volume held steady.

✓Interest-targeting Meta: ₹28L → ₹6L (gradual, 2-week ramp-down)
✓YouTube: ₹14L → ₹3L
✓Custom lookalike (top-20% LTV users): ₹6L → ₹26L
✓Brand search: expanded match types + bid strategy shift to value-based

→ Blended CAC: ₹340 → ₹278. Install volume held steady.

Month 4 · Quality Optimisation

Landing Page + Onboarding Redesign

→ CAC: ₹278 → ₹254. Per-user revenue rising.

✓Ad creative rewritten for intent-qualification (not volume maximisation)
✓Deep-link from ad → specific app feature based on creative context
✓Brand search onboarding drop-off: 34% → 18%
✓D30 ARPU of acquired users: +₹60 vs. prior 30 days

→ CAC: ₹278 → ₹254. Per-user revenue rising.

Month 5–6 · Compounding

Referral Scale + Optimisation Loop

→ Blended CAC: ₹238. D30 ARPU: ₹295. Unit economics positive.

✓Referral mechanic scaled: 22% of installs referral-sourced at peak
✓LTV/CAC tracking dashboard built: live visibility per channel, per cohort
✓Ongoing optimisation: weekly channel review against LTV/CAC threshold (minimum 2.0)
✓Affiliate program shut down entirely: LTV/CAC was 0.28

→ Blended CAC: ₹238. D30 ARPU: ₹295. Unit economics positive.

Results

The Outcome

−30%

Blended CAC

₹340 → ₹238

₹295

D30 ARPU

up from ₹280 baseline

2.8×

LTV/CAC Ratio

from 1.1× baseline

22%

Referral Share

of monthly installs at peak

Metric	Month 0	Month 6
Blended CAC	₹340	₹238
D30 ARPU (acquired users)	₹280	₹295
LTV/CAC Ratio (blended)	1.1×	2.8×
Interest-targeting spend	₹28L/month	₹6L/month
Referral share of installs	4%	22%
Unit economics status	Negative at D30	Positive at D30

Reducing CAC by 30% while simultaneously improving per-user revenue isn't a coincidence. Both moved in the right direction because we stopped optimising for cheap volume and started optimising for quality-efficient volume. The channels that cost more per install delivered users who were worth more. That's the LTV/CAC bet — and it pays.

Insights

The Mental Models I Took Away

Attribution infrastructure is growth infrastructure — not an analytics nice-to-have

Every month we ran without channel-level LTV tracking was a month of capital being misallocated at scale. The 3 weeks to build proper attribution paid back within the first month of reallocation decisions. Companies consistently under-invest in attribution because it produces invisible ROI. The ROI is in every budget decision you make more accurately afterward.

Intent-qualifying ad creative lowers volume and raises quality simultaneously

Counterintuitive but reliable: writing ad copy that describes the specific user the product is for reduces CTR (because fewer people who don't fit click) and dramatically increases post-install conversion. 'CarInfo: Instant RC lookup for vehicle owners' gets fewer clicks than 'Free Car History Report!' — but the people who click it install, use it, and pay. Qualification in the funnel entry is cheaper than filtering post-install.

The referral mechanic is the most underrated component of paid CAC reduction

Referral users have three cost advantages: near-zero acquisition cost, better-than-average LTV (they come with context from the referrer), and they self-select into the product because someone they trust used it. At 22% referral share of installs, the blended CAC calculation absorbed a large volume of effectively free high-quality installs. Every paid CAC optimization should be happening alongside an organic/referral growth initiative.

Onboarding quality matters more for high-CAC channels than low-CAC ones

We had a 34% drop-off from brand search install → first session. Brand search users cost ₹80 to acquire — the most efficient channel. But 1 in 3 of them never opened the app. Fixing that drop-off was worth ₹26 CAC equivalent per install. Onboarding optimisation isn't a retention investment — it's a completion investment on already-spent acquisition cost.

What Hurt Us First

What Slowed Us Down

✕ Turned off interest-targeting spend too fast — install volume dipped sharply

Why it broke

Week 1: cut interest-targeting from ₹28L to ₹6L in one move. Custom lookalike campaigns hadn't scaled up yet to absorb the volume. Total installs dropped 38% in 2 weeks. Leadership concern. We had demonstrated the unit economics case but moved the lever too fast.

What fixed it

Rebuilt the transition plan: ramp down 20%/week on low-LTV channels; ramp up replacement channels in parallel before reducing. Always ensure the volume replacement is available before cutting the source. CAC optimisation needs supply-side management, not just demand-side.

✕ Custom lookalike audiences built from total user base, not high-LTV users

Why it broke

First lookalike campaign: sourced from all-time app install list (3.4M users). Performance was only marginally better than interest targeting — because we'd included all the low-LTV users in the seed audience, diluting the signal.

What fixed it

Rebuilt lookalike audiences from top-20% LTV users only (users with D90 ARPU >₹400). Lookalike CAC dropped from ₹245 to ₹176. The quality of the seed is everything in lookalike modelling.

✕ Landing page A/B test ran simultaneously with ad creative change

Why it broke

Changed ad creative (intent-qualifying copy) and landing page (clearer value prop) in the same week. Metrics improved significantly — but we couldn't tell which change drove it. Wasted 3 weeks re-running isolated tests to get clean attribution of effect.

What fixed it

Strict test isolation: only one variable changes per test cycle. Built a testing calendar with 10-day minimum per test. Takes longer but produces actionable, attributable data. A 3-week clean result is more valuable than a 1-week ambiguous result.

Future

Where the Acquisition Model Goes Next

Predictive LTV Bidding

Currently optimising toward D30 ARPU as the value signal. Building a model to predict D90 LTV from D3 behaviour signals — enabling ad platforms to bid directly against predicted long-term value, not early proxy metrics.

SEO as CAC Deflation Engine

The programmatic SEO architecture drives organic installs at near-zero CAC. Each new ranking page adds to a permanent, compounding organic base. The long-term strategy is making paid acquisition a diminishing share of total installs as organic scales.

B2B Acquisition Channel

Fleet operators, insurance agents, and RTO brokers represent high-LTV users who come in batches. Building a structured B2B acquisition motion: dedicated landing pages, custom onboarding, account management. LTV/CAC for B2B channel likely 6–8× vs. consumer.

We didn't spend less money acquiring users. We spent the same money acquiring better ones — and let the compound effect of quality do the rest.

Building a CRM That Handles 180K+ Monthly Transactions at 99.6% Uptime

Product

How We Used ML to Predict Insurance Buyers (50,000+ Data Points)

Machine Learning

Growth

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

14 min read

Growth

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

9 min read

Growth

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

6 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.
App store conversion rate benchmark, subscription apps
Published figures, traced to their sources — and where the sources disagree.
Install-to-purchase conversion benchmark, travel
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.

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

Results at a glance

We Were Paying ₹340 to Acquire Users Who Cost ₹40 to Serve

The Channel-Level Truth

Not All Users Are Worth the Same CAC

CAC Is the Wrong Metric. LTV/CAC Is the Right One.

How We Moved the Number

The 6-Month Execution

The Outcome

The Mental Models I Took Away

Attribution infrastructure is growth infrastructure — not an analytics nice-to-have

Intent-qualifying ad creative lowers volume and raises quality simultaneously

The referral mechanic is the most underrated component of paid CAC reduction

Onboarding quality matters more for high-CAC channels than low-CAC ones

What Slowed Us Down

Where the Acquisition Model Goes Next

You might also like

Where to go next

We Were Paying ₹340 to Acquire Users Who Cost ₹40 to Serve

The Channel-Level Truth

Not All Users Are Worth the Same CAC

CAC Is the Wrong Metric. LTV/CAC Is the Right One.

How We Moved the Number

The 6-Month Execution

The Outcome

The Mental Models I Took Away

Attribution infrastructure is growth infrastructure — not an analytics nice-to-have

Intent-qualifying ad creative lowers volume and raises quality simultaneously

The referral mechanic is the most underrated component of paid CAC reduction

Onboarding quality matters more for high-CAC channels than low-CAC ones

What Slowed Us Down

Where the Acquisition Model Goes Next