- Client
- Australian telehealth provider, doctor-led weight management, more than one consumer brand
- Channels
- Google Search, Performance Max, Demand Gen, Meta
- Budget
- Weekly spend held within 5% of month 1
- Role
- Audit, rebuild, weekly optimisation
The challenge
High spend. Few subscribers.
A doctor-led weight-management subscription sold through more than one consumer brand. Customers start an online health quiz, finish it, then pay for a first consultation and subscribe. When I took the account over in January 2026, fewer than one in five completed quizzes became a paying subscriber.
- FINDING 1Fragmented structure.Overlapping Google campaigns, duplicated across brands, competed for the same searchers and split the conversion data too thin to learn from.
- FINDING 2Bidding to the wrong signal.Some campaigns bid straight to purchases with too little volume to train on; others chased cheap clicks on Display and broad audiences.
- FINDING 3No follow-up by funnel stage.Meta ran broad traffic campaigns. People who started or finished the quiz but didn’t pay got nothing aimed at where they dropped off.
What I did: one funnel stage per campaign, judged on paid deposits.
- 01Reset measurement (January). Made billing-system deposits the single source of truth for cost per subscriber, and mapped each conversion action to a stage: quiz start, quiz finish, paid consult.
- 02Consolidated Google Ads (late Jan–Feb). Halved the active campaigns. Paused Display, broad generic and duplicate purchase-bidding campaigns, and moved budget into Performance Max bidding to quiz completions.
- 03Rebuilt Meta as stage-by-stage remarketing (Feb). Replaced broad traffic campaigns with purchase-optimised remarketing to quiz starters, quiz finishers and engaged social audiences, each with its own message.
- 04Added prospecting once the base was efficient (Mar). Lookalikes built from paying customers, short testimonial video and Google Demand Gen to widen the top of the funnel before the spring peak.
- 05Optimised weekly. Spend, quiz starts, finishes and deposits tracked per campaign every week. Bid-strategy changes held for 7–10 days to clear learning before the next one.
- 06Checked every claim for compliance. All ad copy and landing-page claims reviewed against TGA advertising requirements and platform healthcare policies before launch.
Results
Subscribers tripled in the first week after the rebuild, and held for eight weeks.
Weekly spend stayed within 5% of January, so the gain came from efficiency, not budget. The biggest move was at the bottom of the funnel: completed quizzes turning into paying subscribers went from 19.2% to 33.7%.
Month 3 cost per subscriber rose slightly because I deliberately funded new prospecting and Demand Gen tests. Subscriber volume held at about three times the month 1 rate while building future demand.
Months 2–3 averages vs month 1. Month 1 = weeks 1–5, month 2 = weeks 6–9, month 3 = weeks 10–13. Results are shown only as indexes, ratios and percentages; no spend, cost or volume figures are disclosed. Meta is described structurally only, and platform-reported Meta purchases are excluded.
The insight
Fewer campaigns, more signal. Optimise one step above the purchase.
Halving the Google account gave Smart Bidding enough conversions per campaign to learn, and that moved results more than any single bid or creative change. Bidding to quiz completions supplied several times the volume of purchases while still predicting revenue.
In a regulated category, first-party CRM audiences and plain, evidence-led copy do the heavy lifting, and the billing system, not the ad platform, decides where budget goes. By mid-April the challenger brand’s weekly cost per subscriber was running close to 90% below month 1.
Keep reading
See all 6 case studies →
