First off, thank you so much for believing in me and subscribing to TheAuxoLetter!

If you're new here, I'm Sanjai Kathirvel, and this newsletter is all about solving business revenue and growth problems. Word-of-mouth has always been my favorite topic – I discovered my favorite and best products through WOM and referrals, and I bet you did too.

Last week, we covered the 2,079-year psychology behind why referrals work.

This week, I'll show you exactly how to build one that doesn't suck.

Most referral programs fail because founders obsess over reward amounts but ignore WHERE they ask, WHO they target, and HOW they measure true incrementality. Here's the four-part framework that actually works.

Most founders get referral programs completely wrong. They spend weeks debating whether to offer ₹500 or ₹1,000, but completely ignore WHERE they ask for referrals. They target their most loyal customers but miss the users with fresh networks.

Here's the framework used by companies that actually make referrals work:

Here's what most people get wrong: They think they can optimize their way to success by A/B testing referral screens. But conversion rates on referral screens are relatively fixed at 2-5%. You can't optimize your way out of low volume.

The real strategy: Get more impressions by asking at psychologically perfect moments.

Where to ask (ranked by effectiveness):

The "Holidizing" Secret from Uber: Instead of boring "₹500 off" year-round, they created seasonal campaigns that gave people social justification:

This works because it gives people a REASON to refer beyond just the money. It's social justification for what might otherwise feel like spam.

The counterintuitive truth everyone misses: Target NEW users for referrals, not your experienced ones.

The simple math:

If 5% successfully refer:

But there's a deeper reason: New users have networks that haven't been "tapped out" yet. If Priya's been using your app for 6 months, she's probably already told her interested friends. New users represent completely untapped social graphs.

The segmentation strategy that 3x'd Uber's results: Don't do generic "give ₹500, get ₹500" for everyone. Segment by:

Real example: Uber paid upto ₹1000 for driver referrals in Bangalore but ₹400-500 in smaller cities. Same program, completely different economics based on local CAC and earning potential.

Most people think Dropbox used storage because it was cheap. Wrong.

Storage in 2008 was strategically brilliant because it was:

Compare this to "points" or "credits" that only existing users understand.

The tiered psychology trick that drives action: Instead of "₹500 when you sign up," try this structure: "₹10,000 when you complete 5 actions, ₹2,000 when you complete 3 actions, ₹1,000 when you complete 1 action."

Even if only 20% complete all actions, the ₹10,000 headline grabs attention and partial completions still drive value.

Uber's most successful driver campaign used this exact psychology: "₹10,000 when you complete 100 rides in first month, ₹2000 when you complete 20 rides, ₹800 when you complete 5 rides."

The ₹10,000 headline was pure marketing psychology, but it worked. Most drivers earned ₹2000-4000 which still felt substantial.

Symmetric vs Asymmetric offers (test both):

Test both, but lean toward referrer-centric in consumer contexts.

Standard CAC/LTV analysis completely misses cannibalization. Most "successful" referral programs are just expensive ways to acquire customers who would have come anyway.

Here's how to measure what really matters:

The Twin Cities Test: Run different offers in similar markets:

If Pune gets 2x more referrals but only 1.3x more total signups, you know there's massive cannibalization happening.

The On/Off Test: Turn off referrals completely for a week. If total signups drop by less than 80% of your typical referral volume, most of those users would have come organically anyway.

Your real metric - Cost Per Incremental Customer (CPIC): CPIC = (Total referral spend) ÷ (New signups - Control group signups)

This is your REAL acquisition cost, not the fake number most companies report to investors.

Here's what nobody wants to admit: People who respond to incentives often behave differently than organic users.

Uber's internal data revealed:

Why incentivized users perform worse:

But here's the interesting twist: Uber's driver referrals actually performed BETTER than other channels because money-motivation perfectly aligned with the job requirements.

The lesson: Understand what behavior your incentive selects for, and decide if that's actually what you want for your business.

I want to hear from you, Auxors!

This Week's Question: Have you tried building a referral program? What worked, what didn't, and what surprised you about user behavior?

Or share: A referral program you love as a user. What makes you actually want to share it with friends?

Hit reply and tell me about it. The best insights will be featured in next week's newsletter (with your permission, of course). Let's learn from each other's experiments!

Growth never stops,
Sanjai Kathirvel

P.S. - Remember: Before building any referral program, make sure people already recommend you organically. If they don't love your product enough to share it without incentives, no amount of money will fix that fundamental problem.