Lessons from building & scaling referral programs to millions of users 

A deep-dive into virality, cycle time, repeat cycles and everything you need to build the perfect referral program.

Over the years I’ve worked on a number of referral programs with different apps – music apps, personal finance, payment apps and more recently on a Credit card product.

Why should you listen to me?

I optimised the sharing experience & launched a feature at Walnut app(a personal finance app based on sms) that took us from 1m to 5m users(In a pre-Jio world) and built a gamified waitlist program that we grew to 330k users(with no paid marketing). Both were programs without any direct incentive to the referred user and hence the hardest to build. I also built incentivized referral programs at Gaana & Smartapp(100k+ users).

Taking inspiration from Lenny’s Newsletter, i’ll share my perspective on when referral or invite programs work best –

• Tight knight Networks or products that require trust – High credit score users are more likely to know other high score users.
• Social or products that want to grow quickly – Social products grow quickly by focusing on increasing ‘N’ – invites per user to an exceedingly large number. Google Pay is a great example of a product that did this.
• Apps/Services where friends being present add net value – Again having your friends on Gpay converts a upi handle into a Gpay friend that you can transfer money without worrying.
• New services pivoting a new concept – Airbnb is a great example of this. Staying at a strangers house is a very strange concept unless your friend vouches it for being ok.
• Reduces barrier to entry to try out a new product users want to use/will find useful – Uber is a great example of this. Getting an invite and free credit reduces the barrier to trying out a new cab-sharing service.

Typically referred users have better retention compared to other channels and hence are the preferred mode of user acquisition.

A referral program isn’t necessarily about offering an incentive to users. In fact it’s mostly effective when we create use-cases in the app which are a natural point for inviting new users. Or we simplify the way existing users invite other users to the app.

I’ll start with a few quick lessons I’ve learned and then we will deep-dive into more advanced concepts like cycle time & increasing cycles via gamification.

To build a good referral program, you need to get a few things right.

1. Keep the value prop simple
2. Give control of status to the end user
3. Use a currency in-house(like Dropbox) with marginal cost that increases ‘net value’
4. Ensure it works end-to-end
5. Tackle fraud

Keep the value prop simple

When it’s simple, it’s repeatable. It’s shareable. So a friend can tell another. And another.

One of the reasons for the success of the double referral with similar value is it’s so simple. I get $5 and my friend gets the same. It’s easy to remember and share.

Increase net value

The perfect loop – a viral loop that adds value & increases retention

The Drop-box example is a classic. You got 500mb of additional space for every friend that you invited to Dropbox. And getting additional space helped you store more documents & helped increase your usage of Dropbox. A referral program that increases retention!

Give control of status to user

When i built the first referral program at Smartapp, one of the things we did thanks to my then boss was give complete visibility on the status to the referrer. You would know if your friend registered, completed the goal(register/play a song/buy a subscription) and it made a big difference. I used this trick again to create the OneCard referral program, giving visibility to the end user of their status. And the best part – your friend joining was the perfect occasion to re-engage with the app. Just like the perfect loop described above.

Make sure it works

Referral programs are tricky. Especially when they’re incentivized. Hence it’s important to ensure that the experience is smooth end to end. If due to tech or other limitations it breaks in a few instances, soon users will blast your reviews with 1 stars.

Tackle fraud

If your referral program is successful, you’ll be attacked by fraudulent users very quickly. Hence it’s necessary to build a system even if it’s retroactive to tackle fraud. Also it’s important to design the program in such a way that the incentive is designed to get your target user.

These were the basics. However to really make a mark, one needs to think outside the box and do something different. Gone are the days when a simple double side incentived referral would give you a great boost. Especially in a market flooded with cashbacks, discounts and bonuses.

One has to do something creative. Gamify it a little. It’s exactly what we did with the waitlist 1.0 at OneCard. When we were designing the first waitlist of OneCard, we spent almost a month thinking about the ‘How do we make it standout from the crowd?’ ‘How do we do something different?’ and after iterating on a few different approaches landed on the eventual one.

Finding ‘Metal’ was the most challenging & fulfilling aspect of creating your Metal OneCard.

Here we used the core drive 2 – development and accomplishment. Users were motivated to unlock metal because it was ‘challenging’.

Cycle time – the secret to referral programs

If you’re starting out, you only focus on the viral factor in a referral program. But the secret sauce lies in reducing the cycle time and increasing the number of cycles for existing users.

Cycle time definition – it’s the time for the viral cycle to repeat. Join, use-app, invite friends.

Viral cycle

How quickly does an invited user start referring new users?

Reducing cycle time. A – cycle time of 3 days vs cycle time of 1 day

He also shares a link to understand the impact of a shorter cycle time. You can refer it here. As we can see, reducing the cycle time from 3 days to 1 day has led to a crazy amount of user growth in the same time period.

One of the referral programs i built out had a pretty long registration cycle. And yet we were able to get amazing numbers thanks to our focus on the core drive and motivation.

Increasing invites per user

While reducing the cycle times focuses on getting invited users to invite new users again as soon as possible, i’m also interested in exploring ‘how can we get the user to repeat the invite cycle’? to increase invites per user?

Repeating cycles – If it’s difficult to reduce the cycle time, it’s important to explore repeating the cycle.

Consider a user invites 10 users in the first cycle and then again invites 10 users in the 2nd cycle. And repeats it. That’s thrice the virality! I’ll try to model the combined effect of reduction of cycle time and increase in the number of base cycles.

If you’re designing your first referral program, here are some metrics you need to measure to understand and measure your success –

Viral factor = referred users/all us

You can download the sheet to calculate viral factor and change the conversion rate and invites per user to play with the different rates. This link is re-shared from David Skok’s blog post on viral marketing – you can read it here. https://www.forentrepreneurs.com/lessons-learnt-viral-marketing/

Some of the other tricks i’ve learned over time –

• Adding share at the win moment – It’s important to nudge the user to complete an action at the right moment.
• Using images wisely – Consider the share image. Is it optimized for the network? Does it bring out the value prop when it’s viewed in preview mode?
• The landing page – This is the most important page of a referral program and it’s important to get the copy right. Write something, get user feedback. Understand if your users understand the value proposition. Revise it and continue.
• Personalize it – It’s simpler to share your code ‘DINESH’ rather than sharing your link. Especially if you’re a celeb who’s doing a shout-out.
• Make an awesome share message and #hashtag.

One of the best share messages i’ve ever written

I’m curious to learn more from you’ll. What are some good estimates for the different variables? How does it vary in different industries? I’ll write a follow-up post on this to help you design the best referral program and also model your CAC.

If you found the article useful, please feel free to hit the share with your friends..

Header image courtesy: https://staticsteven.com/i-am-a-strange-loop/ Also a nice blog post on this.

Appendix –

1. You can refer to David Skok’s blog to get a deeper understanding – https://www.forentrepreneurs.com/lessons-learnt-viral-marketing/
2. Lenny’s weekly newsletter

I’m a product builder currently re-defining credit in India. If you’re interested to work with us, please DM me here or on Twitter. I love all things consumer — fintech, social, video and music. You can follow me on Twitter or Linkedin.

This post was re-published from my Substack first written in Sep 2020- https://productnotes.substack.com/p/lessons-from-building-and-scaling