Essay · Growth & Product

The referral mechanic I keep coming back to.

Feb 2025 5 min read

Referral programs that actually work share three properties: they trigger at the right moment, split the benefit altruistically, and frame sharing as identity rather than transaction.


Referral programs have a reputation problem. Most are either invisible — buried in Settings > Rewards, discovered by no one — or so aggressive they feel like a pyramid scheme. The ones that actually work don’t look like either. They look like something you’d want to share even if there were no incentive.

The three properties that matter

Triggered at the right moment. Not after signup — the user hasn’t experienced value yet. Not after ninety days — the enthusiasm has normalised. The sweet spot is right after the first real value moment, when the user thinks “this is actually good.” That’s when sharing feels natural rather than transactional.

The benefit is split, not just recipient-focused. “Get $10 when a friend joins” performs worse than “Give your friend $10, get $10 yourself.” Altruistic framing works better than pure self-interest, especially in markets where social reputation matters. The sender has to feel good about sending it.

The language is about identity, not mechanics. “Invite a friend” is transactional. “Share with someone who’d get it” is identity-based. Users share things that signal something about who they are. If your referral CTA makes sharing feel like selling, it won’t travel far.

What I learned building the Driver Incentive Engine

The ChargePoint referral program was designed for ride-hailing drivers — a community with strong informal networks and shared identity. Drivers already talked to each other. The question was whether we could make our program feel like part of that conversation rather than an interruption to it.

Three things made the difference: timing the prompt to post-successful-charge, framing the benefit as “help a fellow driver save money,” and using driver-to-driver messaging rather than brand-to-driver copy in the notification. The result: 1,247 referrals, 34.8% conversion, without a single paid acquisition channel.

The underrated part

Referral programs fail more often at the receiver end than the sender end. The referred user arrives with context — a specific person recommended this — which creates both an expectation and a responsibility. Most products treat the referred user like any other new user. Handling that handoff well, acknowledging the referral and delivering on the implicit promise the referring user made, is where most referral programs leave conversion on the table.

Léa Thuy Vu is a product and growth professional based in Vietnam. Currently at ChargePoint Vietnam, previously at the Clinton Health Access Initiative. Writes about incentives, product, and behavior change.
All writing Next Essay Building for Southeast Asia →