Referral programs look simple enough from the outside. You give customers a referral link, they share it, their friends sign up, everyone gets something.
Most Braze teams try to run referrals with what they already have: a shared promo code, a basic Canvas, and a prayer that Braze alone can hold it together. Well, it can't. Braze is built to deliver messages, not manage referral logic.
The brands that have figured this out pair Braze with a dedicated referral engine like Voucherify. Here are some lessons from teams who've been where you are.
.webp)
A referral program has three moving parts: the invite (getting your customers to share), the conversion (the friend takes a meaningful action), and the reward (both sides get something). Braze is very good at one of those three.
The invite is where Braze shines. You can build a Canvas that triggers at the right moment in the customer journey, personalizes the message, tests different channels, and reactivates advocates who've gone quiet.
The conversion and reward are where teams run into trouble. Braze doesn't natively know whether a referred user is genuinely new, whether the conversion event is valid, or whether a customer has already hit their reward limit.
That's the gap Voucherify fills. The cleanest way to think about how the two work together: Braze owns what the customer experiences, Voucherify owns the referral logic underneath.
The instinct when a referral program underperforms is to increase the reward. Taxfix's experience suggests the real lever is usually somewhere else.
Taxfix ran a two-sided program for their German tax filing platform. The reward structure was asymmetric by design: referrers got Amazon gift cards, referred friends got 50% off their first tax return. Not the same reward on both sides, because the motivations aren't the same. The referrer wants portable cash value they can use anywhere. The new customer wants a low-stakes reason to try something unfamiliar. Matching the reward type to the motivation, rather than defaulting to the same thing for everyone, is already further than most programs get.
But the real difference was cohort selection. Instead of inviting all active users, Taxfix targeted by four signals: submission status (had they actually filed?), seniority in the product, channel enablement, and promotional consent. That filtered out users who were technically "active" but unlikely to share meaningfully and it's what drove the 95% lift in participation.
What to take from this: Before testing a bigger reward, test a tighter cohort. And if your referrer and referee have different motivations (they almost always do), your reward structure should reflect that.
Most teams don't build their referral rules properly because nothing forces them to. Northmill didn't have that luxury.
Swedish tax law caps annual referral rewards per customer. That meant every reward decision had to be validated against a running total, automatically, on every redemption. There was no way to approximate this in Braze segments or track it in a spreadsheet, it had to be enforced at the program level. So Northmill built it into Voucherify as a hard cap that fires before any reward issues. Braze handled the CRM and messaging layer. The program went live in weeks.
The interesting part isn't the compliance requirement, it's what the constraint produced. Because they were forced to put all reward logic in one place and enforce it automatically, they ended up with a program that was actually debuggable, auditable, and portable. When they expanded from Sweden to Finland, the same stack handled both markets without a rebuild.
What to take from this: Whatever your compliance environment, treat your reward rules as infrastructure. Define them once, enforce them automatically, and build for the market you'll be in next year.
Referral programs are often optimized on the message. Better subject line, clearer CTA, higher reward amount. Tourlane's results suggest the timing of the invite is a much bigger variable.
Tourlane is a European travel platform. In travel, a personal recommendation carries unusual weight, it's a high-consideration, high-trust purchase. But that trust only converts when the recommendation comes from someone who is genuinely enthusiastic, at a moment when that enthusiasm is fresh. Tourlane built their program around identifying those moments in the customer journey rather than sending the invite on a fixed post-signup schedule.
The results reflect it: 80% conversion rate for referred users, which is unusually high for any acquisition channel. Referrals eventually accounted for over 15% of CRM bookings, making the channel genuinely meaningful rather than a rounding error.
What to take from this: Identify the moments in your customer journey where enthusiasm is highest. That's when to ask for a referral, not when it's convenient for the calendar.
Some categories have a built-in referral advantage. If you're in one of them, you're probably underinvesting in the channel.
Unobravo is an Italian online therapy platform. You don't find a therapist because you saw an ad. You find one because someone you trust told you it helped them and that kind of recommendation carries a level of credibility no paid channel can replicate. The self-selection is also important: the only people who recommend a therapy platform are people who are genuinely committed to it and confident in it. That means the referrer pool is high-quality by definition, and the referred users arrive with real intent.
Beyond therapy, similar structural advantages exist in fintech (you share a bank app when you genuinely trust it), insurance (same logic), SaaS tools with strong team dynamics, and any category where the decision is high-stakes enough that people research through their network rather than through ads. If your product sits in one of these categories, the question is why referrals aren't already one of your top acquisition channels.
What to take from this: Map your referral channel against your category's trust dynamics before you decide how much to invest in it. In some categories, referrals will outperform paid on CAC and retention. In others, they're a nice supplement. Knowing which one you're in changes how you prioritize.
There are three integration patterns to run strong referrals through Braze and Voucherify and most mature programs use all three.
Connected Content is the right starting point. When Braze renders the invite message, it calls Voucherify in real time to generate and assign a unique referral code. The code gets pulled into the message via Liquid.
What kind of information Braze can pull from Voucherify API?
Here is a comprehensive GitHub library with useful Liquid and Connected Code snippets for managing promotions, referrals, and loyalty programs using Voucherify and Braze.
.jpeg)
Custom attributes are better when you need referral data available across channels without a live API call on every render. Voucherify pushes the referral code, link, referral count, and reward status into Braze as custom attributes. From there they're available as Liquid variables in any template and you can build Braze segments on top of them.
With Custom Attributes you will be able to share important properties between Braze and Voucherify, such as:
To learn more about data synchronization and to get more detailed information about other tools that might help you in creating a perfect referral program, head on to Mike’s article: How to Fast-track a Referral Program with Voucherify, Segment, Braze, and Branch?
Webhooks give Braze real-time signals from Voucherify. When a referral converts, Voucherify fires an event to Braze that triggers the reward notification Canvas. This is the part that actually makes the advocate feel something: immediate confirmation that their referral worked, their reward is on the way, and an invitation to share again.
Most teams start with Connected Content, add Custom Attributes once they want to build reactivation flows, and layer in webhooks once they're ready to close the loop with real-time reward notifications.
.svg){kind=icon}
