We turned IKIGAI's "advertise to every woman nearby" into an intent engine that sells out every intake — and grows membership that stays.

Part of the Ikigai series · one client, four lenses

This is the Performance Marketing cut of the IKIGAI engagement — the media engine. The same story is told three other ways across the Funnel & CRO, Performance Creative, and Data & Analytics cuts. Here we focus on one question: once a premium yoga brand decides to grow on paid, who do you actually advertise to, and how do you make the spend compound?

When we started, the answer was crude — advertise to every woman near the studios, on one broad message, with broken tracking underneath. We replaced it with an intent-based retargeting engine where every stage of the buyer gets its own audience, offer, and landing page. The result: a >12× ROAS introduction offer at a $5 CAC, membership that grew +20% and held, and a $6,000 teacher-training program that sells out every single intake.

Meet IKIGAI.

IKIGAI runs three premium yoga studios across Hong Kong with a genuine product and a five-year following. The full story — the brand, the slow season, and how the whole engine fits together — lives on the overview case. This page zooms into the media: targeting, retargeting, attribution, and the subscription economics that make or break a studio.

The challenge — spray-and-pray, on broken tracking.

The media account wasn't failing loudly. It was failing quietly — spending on the widest possible audience and calling whatever came back "performance."

1. Targeting was a blunt instrument. Ads ran to every woman within a radius of each studio — men had been switched off because "they didn't work" — on one broad message. No distinction between someone who'd never heard of IKIGAI and someone who'd already taken a trial.
2. No retargeting worth the name. People visited the page and didn't buy. People took a trial and didn't convert to membership. Members lapsed. None of them were being spoken to differently — the highest-intent audiences in the business were being left on the table.
3. Attribution was broken, so Meta optimised for the wrong people. Purchases weren't reliably reaching the platform, so the algorithm learned from weak signal and chased cheap clicks instead of real members. The full teardown is the Data & Analytics cut — but its symptom showed up here, as wasted spend.
4. Membership spiked, then leaked. A push could lift sign-ups for a month — and then churn would quietly claw it back. In a subscription business that's not growth, it's a treadmill.
5. The high-ticket program launched, then went quiet. The $6,000 teacher training ran as occasional one-off launches — a burst of ads, then silence — so intakes were a gamble instead of a standing, sold-out engine.

Our approach — from one audience to an intent matrix.

We rebuilt the media in a deliberate order: fix the signal, segment the intent, then nurture the high-ticket on a permanent cadence.You can't retarget on broken data, and you can't sustain membership with a one-off push — so the sequence matters.

1. Fix the signal before scaling the spend

Every retargeting audience and every "optimise for purchases" instruction is only as good as the data Meta receives. The account was optimising against a broken signal — purchases happening off-platform and never reported back — so the algorithm was effectively flying blind. Before touching audiences, we rebuilt attribution so a real membership purchase reaches Meta and the AI can learn from it. The full rebuild is the Data & Analytics cut. For the media, the point is simple: clean signal in, real members out.

• Purchases + key events wired back to Meta — so optimisation chases members, not cheap clicks
• Audiences rebuilt on trustworthy signal — retargeting and lookalikes seeded from real buyers, not noise

2. Build the retargeting matrix — one audience per intent

The single biggest unlock wasn't a clever creative — it was recognising that "people near a yoga studio" is five or six completely different audiences with different intent, and each deserves its own offer, message, and destination. Someone who visited but didn't buy needs a reason to come back. Someone mid-trial needs a nudge to commit. A lapsed member needs a win-back, not a "discover us." We stood up a campaign per intent and matched each to a dedicated landing page.

The retargeting matrix — the core of the media rebuild.

A note on what's shown: this is the high-level cut. The working build is roughly 10× more nuanced — each row splits by studio, season, prior engagement, and creative angle, with its own conversion event. The matrix above is the floor of the work, not the ceiling.

3. Make the high-ticket a standing engine, not a launch

A $6,000 teacher-training program doesn't sell on a cold ad and a deadline. It sells on trust built over weeks. So instead of bursty launches, we run it as an always-on nurture: success-story campaigns from past graduates, win-back ads to everyone who enquired but missed the last intake, and early-bird offers that open the next cohort before it's announced. The result is an intake that's effectively pre-sold by the time it opens — sold out, every quarter.

• Always-on social proof — graduate success-story campaigns running continuously, not just at launch
• Win-back for every near-miss — enquired-but-didn't-enrol gets the next cohort first
• Early-bird pre-sell — the cohort fills before the public announcement
• Multichannel nurture — Meta + email + webinar working the same list, not one channel guessing

The shift, before → after.

Why "sustained" is the whole game in subscriptions.

Any agency can buy you a spike — a burst of sign-ups that looks great in a monthly screenshot and quietly leaks away. In a membership business the only number that compounds is the one that holds. The retargeting matrix is what makes sustained growth possible: by continuously winning back lapsing members and converting trials properly, the base grows faster than it leaks — so a good month doesn't quietly unwind into an average quarter.

What IKIGAI's media rebuild taught us about paid for subscriptions.

1. "Everyone nearby" is the most expensive audience you can buy. — The cheapest growth is hiding in the people who already raised their hand — visited, trialled, lapsed. Spend there first. Broad prospecting only earns its place once the retargeting matrix behind it is catching everything it stirs up.
2. Fix attribution before you scale spend. — If real purchases never reach Meta, the algorithm optimises for the wrong people and every extra dollar buys faster confusion. Clean signal isn't a "data" nicety — it's the precondition for media that compounds.
3. In subscriptions, a win that churns isn't a win. — Anyone can buy a spike. The job is a base that out-grows churn — which means win-back and retention campaigns are as important as acquisition. We optimise for the line that holds, not the screenshot that pops.
4. High-ticket is nurtured, not launched. — A $6,000 program sells on trust built continuously — success stories, win-backs, early-bird — so the intake is pre-sold before it opens. Run it as a standing engine and "sold out" stops being luck.

The rest of the IKIGAI engagement.

This is one of four lenses on the same engagement.

1. The Funnel & CRO cut covers the architecture each audience lands in
2. The Performance Creative cut covers the per-audience ads the matrix needs
3. The Data & Analytics cut covers the closed loop that makes the signal trustworthy