Remitly: 15× weekly acquisitions, CAC from $1,500 to $70

The challenge

Remitly's WhatsApp product line targeted Spanish-speaking US consumers — a high-potential audience inside a regulated category. The program was acquiring 40–50 customers per month, roughly 10 per week, at a CAC near $1,500. That number sat well above what the product economics could sustain.

Two structural constraints made it impossible to scale by throwing more budget at the existing playbook.

Two walls

Constraint 1 — Regulatory restrictions blocked audience targeting. Compliance rules for the category disallowed the standard targeting toolkit on paid platforms. No custom audiences. No lookalikes. No interest targeting. No behavioral segments. The program had to acquire customers through broad targeting alone.

Constraint 2 — The in-house creative queue couldn't keep up. A shared creative team stretched across multiple product lines meant weeks-long waits for each asset. Velocity was the difference between a program that scaled and one that stalled.

Neither could be negotiated away. They had to be designed around.

The approach: one program, three channels

Three channels — Google, Meta, and TikTok — run as a single integrated portfolio, not siloed campaigns. Budget and learnings moved across channels as signal emerged.

Google Search captured intent: broad-match on Spanish-language queries, catching high-consideration demand already searching for remittance solutions. Meta delivered the volume engine: static and UGC creative, broad demographic framing, steady volume at sustainable cost. TikTok was the unlock: native, human-first UGC that drove the step-change in program scale.

Targeting strategy: language as the frame

With the standard targeting toolkit off the table, we flipped the strategy. Spanish-speaking broad match became the primary lever. Platform algorithms found intent signals inside a broad demographic frame.

The bet: with enough creative signal and clean conversion feedback, platform ML could surface the right audience inside a very loose targeting container — no custom segments required. It worked.

Creative strategy: always-on, native, human

Three principles drove the creative engine. One: a minimum always-on asset set — at least one UGC video and three static images live at any given time, with continuous testing and refresh. No dark windows. Two: platform-native voice — creative built to feel like it belonged on TikTok, conversational, human, low production value by design. Not brand film shrunk to 9:16. Three: continuous refresh — tested into winners on a weekly cadence. Losers retired fast; winners extended through variants rather than replaced.

Production model: own the stack, sidestep the queue

Velocity came from owning the production stack end-to-end — not from waiting in a backlog. UGC video editing in hours, not days. Static creative iterated in-house to feed continuous test cycles. Copywriting sized to each channel and audience hypothesis. 1:1 work with in-house designers for the specific assets the program needed, on the timeline it needed them.

The numbers

Net result: roughly a 15× lift in acquisition volume and up to a 95% reduction in CAC, on a regulated product with no audience targeting available. CAC settled near $150 at sustained scale.

What this unlocked

One — A proven paid acquisition motion for the WhatsApp product. From an unproven channel program to one the business could commit budget to with confidence. Commercial unlock.

Two — A TikTok-native creative playbook the in-house team could scale from. Codified what worked — voice, structure, hooks, formats — so the internal team could extend the model without the original operator. Capability transfer.

Three — A cost structure that made continued investment viable. CAC pulled inside the product economics — a CFO-level unlock that opened the door to sustained spend and expansion. Unit economics.