Positioning and proof. Lungy’s messaging house gave us a solid backbone. Tagline, value pillars for Anxiety, Stress and Sleep, proof points such as “developed by doctors,” privacy by default, and Apple Health integration. We turned that into precise ads, App Store assets and creative test routes that stayed faithful to the product.

How we work. We plugged Lungy into our standard operating model, the same one we use to scale subscription apps across markets. It combines a measurement spine, a creative engine that can ship variants at high frequency, a staged channel strategy that learns fast before chasing scale, and clear weekly to quarterly cadences.

Our playbook. Internally we codify the above so every engagement benefits from the same standards. The playbook defines the case‑study structure, how we tie creative, media and engineering together, and how we frame tests, benchmarks and AI use.

Lungy entered one of the loudest categories in mobile. Calm and Headspace dominate mindshare, and newer breathwork tools are busy in paid auctions. On iOS, post‑ATT privacy means weak or delayed signals unless SKAdNetwork and subscriptions are wired correctly. Q4 seasonality would distort early tests, particularly in the UK. And because Lungy’s experience is genuinely new, our creative had to show that “magic moment” instantly or risk being mistaken for generic mindfulness content. These realities shaped the pilot plan.

Users needed to feel a change quickly, then want to repeat it tomorrow. The business needed a growth system where efficient installs, strong product page conversion, trial starts and early retention add up to a payback window that justifies scale. Miss any link in that chain and paid acquisition becomes a treadmill. Our job was to build the chain end to end.

Strategy, then scale

We refused to “buy scale” before the system was ready.

Weeks 1–2, wire the truth and brief the work

SDKs went in. Adjust for SKAN, RevenueCat for subscriptions, dashboard live. In parallel we briefed creative, set guardrails and built the first test cells. The six‑week start‑up cadence and roles were set out on a single page so everyone could see the path from first spend to scale.

The creative system: show the magic, say the outcome

From Lungy’s messaging house we pulled clear, outcome‑led copy. No cloudy wellness language. The hero moment was always the same, the world reacts to your breath. Short portrait videos led, supported by square video and clean statics that gave us learning speed across placements. This discipline paid off immediately in early cycles.

Alongside performance assets, our studio produced three broadcast‑quality brand films in a two‑week turnaround, each one mapped to a core intent, Anxiety, Stress and Sleep. We delivered 10‑second, 15‑second and 30‑second masters in multiple aspect ratios for Meta, YouTube and future CTV, with captions, safe‑area framing and platform‑ready slates. We then versioned the films for different audience segments and tested them in multiple variants, using the same concept‑route‑variant hierarchy to isolate what moved attention and intent. The films pulled double duty, upper‑funnel awareness and evergreen anchor assets for new acquisition waves and ASA Creative Sets. Our proposal details the live‑action and motion capabilities, and the pod model that lets us move at this speed without sacrificing craft.

We opened in the United States with a curated two‑tier interest structure around breathwork, anxiety and modern wellness. We limited placements to high‑attention surfaces on iOS 14.5 plus to protect signal quality, and we fixed the headline while testing only two primary text lines to isolate the copy variable. The winner, “This app reacts to your breathing in real time, and trains you to beat anxiety. Try Lungy now,” cut cost per click by about 35 percent versus the alternative. Among four portrait videos, the best unit drove five‑and‑a‑half times more installs than the joint lowest CPI ad. That told us what to run and what to kill.

We then ran a broad Advantage+ App Campaign to compare machine‑learning delivery against our handcrafted audience. Despite a solid CTR on one variant, CPM rose as the model chased a smaller, more expensive sub‑audience, and the campaign delivered a higher CPI than Round 1. The pilot explains why a higher CTR can still lose when CPM inflates, a common surprise in early ML tests. Conclusion, park Advantage+ until the creative bench is deeper.

Rather than argue broad versus narrow, we recut by user intent. Two campaigns, Stress and Sleep, used the same creative to isolate copy. For Stress, “App Hears Your Breaths and Reacts To Beat Anxiety” reduced CPC by roughly 24 percent versus the curiosity headline. In Sleep, a clean static unexpectedly won CPI against video in one ad set, a reminder that clarity can beat motion in discovery.

A quick UK machine‑learning probe during the festive spike showed CPCs surging in a smaller market under heavy competition. We paused and rescheduled the UK scale‑up for a calmer window. This saved spend and reinforced a US‑first plan before returning to the UK.

Armed with winners, we returned to the best‑performing interest set and pitted four headline variants against each other. The combination that paired our proven primary text with “Beat anxiety for good, by breathing with Lungy” and the top creative produced our lowest CPI of the program. Placement‑level analysis showed Instagram Feed reaching about £1.50 on the winning unit, Facebook Feed about £4.49. Best‑ad CPI moved from roughly £7.56 in early cycles to about £4.17, a 44.84 percent reduction achieved by tightening message and placements.

Those wins coincided with Apple surfacing Lungy editorially multiple times. The paid engine and App Store conversion work provided the momentum, the editorial features added organic lift, and together they helped Lungy climb Health and Fitness rankings.

While Meta gave us learning speed, Apple Search Ads was built to harvest intent.

Structure. Four campaigns, Brand, Category, Competitor and Discovery, with exact and broad ad groups, and negatives at the correct levels to control bleed. This mirrors how we run ASA at scale on other iOS subscriptions.
Custom Product Pages. One CPP per intent pillar. Anxiety searchers saw anxiety screenshots and preview first. Stress and Sleep got their own narratives. We paired CPPs with Creative Sets so relevance carried from query to product page to install. 
Bidding and measurement. Start on cost‑per‑tap to win impressions for a newer listing, then move to CPA optimisation once SKAN stabilises. Reconcile ASA tap‑through and product page conversion with Adjust installs, and tie value to trials and subscribers in RevenueCat.

Paid CPI is only half the equation. Product Page conversion determines whether a tap becomes an install, which can halve or double effective CPI. We ran a Product Page Optimisation stream in parallel, testing icon, first screenshot, headline and preview, plus CPPs aligned to ASA themes. The pilot spells out the math simply, move conversion from 20 percent to 40 percent and your effective CPI is cut in half without changing bids. Treat App Store assets like ads and rotate them with the same discipline.

Paid only scales if cohorts convert and stay. Early cohorts showed a healthy uptick in trial starts from paid traffic, but day‑one retention near 10 percent and seven‑day conversion to paying around 2.4 percent would cap LTV if left alone. We prioritised:

• Onboarding re‑sequence. Lead with the most viscerally responsive exercise, then ask for email and permissions once value is felt.
• 7‑day starter plan and streaks. A simple plan that rewards daily practice and celebrates small wins.
• Paywall timing and plan mix. Emphasise annual where justified and present at moments of earned value, not at the door.
• Gentle, contextual nudges. Push and in‑app prompts that nudge, not nag, and that speak to Anxiety, Stress or Sleep intent.

The pilot’s cohort readouts and recommendations informed this roadmap.

AI quietly accelerated everything.

• Ideation and scoring. Generate hook lines in the tone set by the messaging house, score against Octalysis motivations, shortlist for human polish.
• Variant production. Create safe aspect‑ratio and layout permutations so Dynamic Creative Optimisation has breadth without overwhelming the studio.
• Fatigue alerts and anomaly detection. Watch CTR and CPM velocity, alert the pod in Slack when CPI drifts outside control limits, and auto‑compile weekly readouts.
• Audience clustering. Form lookalike seeds from high‑engagement cohorts once signals accumulate.

These workflows are part of our standard toolkit and appear throughout our internal playbooks and proposal materials.

• Human curation first. On opening cycles and limited creative benches, handcrafted symptom‑led audiences beat broad ML on CPI. We reintroduced ML only when creative diversity and scale smoothed delivery.
• Outcome beats curiosity. Clear value headlines, “Beat anxiety for good, by breathing with Lungy,” outperformed “This App Reacts To Your Breathing In Real Time,” especially when paired with our strongest creative.
• Static can win discovery. In Sleep, a clean static outperformed video on CPI, so statics stayed in the mix for coverage and learning speed.
• Respect seasonality. UK testing during the festive spike confirmed auction pressure. We focused on the US, then returned to the UK when costs normalised.
• Post‑install is decisive. Trial and D1 retention were the binding constraint. We stood up a product‑led activation stream so LTV could rise as paid scaled.

• CPI improvement that compounds. Best‑ad CPI moved from about £7.56 to about £4.17, a 44.84 percent reduction driven by message clarity, segment focus and placement selection. Instagram Feed hit roughly £1.50 on the winning ad, which guided future weighting.
• Performance in context. Our regional CPI benchmarks show North America is usually the most expensive territory on a CPI basis, so achieving a sub‑£5 CPI in early discovery for a Health and Fitness subscription app is a strong platform for scale.
• Editorial compounding. Apple’s features added organic momentum and helped Lungy rise in Health and Fitness rankings during the same window, amplifying paid performance.

We intentionally omit specific budget figures in this public version. What matters is the system’s shape, the measurable efficiency gains and the playbook that repeats.

Lungy’s launch confirms a principle we hold across categories. If your product creates a visceral moment of value, remove every ounce of friction between the user and that moment. That is why the ads show the breath‑reaction in the first second, why headlines promise a specific outcome, and why Product Page and Custom Product Pages echo the same promise. The algorithm learns faster, users convert more often, and cohorts monetise better.

• Instrument truth before you spend. Wire an MMP like Adjust for SKAN, pipe trials and subscriptions through RevenueCat, and agree one KPI ledger. Look at CPI by creative and audience next to trial and subscriber cohorts every week. Kill myths fast.
• Segment by symptom, not persona decks. Map everything to Anxiety, Stress and Sleep. Ads, screenshots, preview, and ASA Custom Product Pages should tell the same story the user searched for.
• Staircase channels. Learn fast on Meta with controlled variables. Add Apple Search Ads with CPPs to harvest the highest intent. Expand to Google UAC, YouTube, creators and programmatic only when CPI and product page conversion are predictable.
• Treat App Store conversion as a profit lever. A disciplined Product Page and CPP testing program can cut effective CPI in half without changing bids. Run it in lockstep with paid.
• Win after the tap. Re‑sequence onboarding to a visceral first session, ship a 7‑day starter plan, time the paywall to earned value, and use gentle intent‑specific nudges. Cohort in RevenueCat, iterate weekly.
• Use brand films to seed demand and teach algorithms. Build a small set of broadcast‑quality films around your core intents, deliver 10s, 15s and 30s masters, and version them for segments. Test them like performance assets and let them pull duty as upper‑funnel and evergreen anchors. Our studio’s live‑action and motion capabilities are set up for this exact pattern.