AI HealthTech case study: 2k → 12k+ monthly downloads in six months by rebuilding acquisition, lifecycle messaging, and the ICP loop instead of buying more traffic.
Challenge
The app had real adoption — ~2,000 monthly downloads, decent retention, and a clinical advisory board. But the growth team was reactive: paid budgets shifted week to week, lifecycle messaging was inconsistent, and there was no shared definition of who the best users were. The CEO wanted to fundraise on a growth story she couldn't yet tell with numbers.
Approach
Two weeks of diagnosis surfaced three constraints — an unclear ICP (the team was acquiring two distinct user types and serving neither well), a paid acquisition setup that mixed brand and performance in the same campaigns, and a lifecycle programme that triggered on installs rather than first-action. We rebuilt around a sharper ICP, separated brand and performance budgets with explicit decision rules, and wrote new onboarding and re-engagement sequences keyed to the activation event that actually mattered.
Outcome
Monthly downloads moved from ~2,000 to 12,000+ over six months. The number wasn't the win — the win was that the team could explain *why* it moved, what to scale, and what to drop. The growth story became fundable, and the founder spent less time approving experiments.
Monthly downloads
6×
2k → 12k+ over six months
Activation rate
+38%
After lifecycle rebuild keyed to first-action
Cost per qualified install
-41%
Clean separation of brand vs performance budgets
Founder hours per week in growth
-50%
Estimated, vs the engagement baseline
A consumer-facing AI HealthTech app had cleared the messiest part of product-market fit. About 2,000 people downloaded the app each month, retention curves were stabilising, and a clinical advisory board lent the brand credibility. The CEO had raised a small angel round on the strength of those signals and was preparing for a Series A conversation.
The growth team was three people and an external agency. They were busy. Paid budgets moved week to week, lifecycle emails went out, and an in-house creative could ship a new ad set in under a day. What was missing was a system — a shared answer to “who is the best user we want more of, and what’s the cheapest way to reach more of them?”
The brief from the CEO: “I need a growth story I can tell with numbers. Right now I can show charts that go up and to the right, but I cannot defend any of them on a board call.”
Three constraints showed up quickly.
The ICP was actually two ICPs. The app was being used by clinicians and patients, with very different motivations and lifetime values. Acquisition campaigns lumped them together; messaging spoke to neither cleanly. The first decision was to pick one ICP for the next quarter and de-prioritise the other in paid acquisition.
Brand and performance budgets were mixed. A single Meta account ran awareness videos alongside install campaigns, with shared learning. CAC numbers were hard to read because the algorithm was optimising across goals that conflicted. The fix: split the account, set explicit decision rules for each, and rebuild creative briefs by intent.
Lifecycle triggered on the wrong event. Onboarding emails fired when the user installed the app. The activation event that actually predicted retention was the first completed in-app session — a noticeably different cohort. Re-keying lifecycle to the activation event opened a much wider design space.
We ran the rebuild in two parallel streams.
The acquisition stream was the agency’s. We rewrote campaign briefs around the chosen ICP, re-architected the Meta account with separate prospecting and lower-funnel campaigns, and added explicit “kill” rules at the ad-set level so dead creatives stopped absorbing budget. The agency owned execution; we owned the decision rules.
The lifecycle stream was internal. We mapped the user journey from install through first session, day-three, day-seven, and day-twenty-one, and wrote a sequence of triggered messages keyed to behaviour rather than time. Re-engagement campaigns targeted the cohort that installed but never activated — a bigger and more recoverable group than the team had assumed.
By month six, monthly downloads had moved from roughly 2,000 to 12,000+. Activation rates were up around 38%. Cost per qualified install was down 41% on the previous quarter, with most of that improvement coming from the cleaner account structure rather than new creative.
The number that mattered to the CEO wasn’t the download count — it was that the team could open a board update and explain, line by line, what had moved and why. The growth meeting moved from a status report to a decision meeting. The agency relationship became contractually narrower and operationally faster. The founder dropped from approving each campaign to approving the rules.
A senior operator embedded part-time, a willing internal team, an agency comfortable being managed against decision rules instead of activity reports, and a founder who was prepared to pick one ICP and stop hedging. The rebuild was not technically complex. It was a discipline problem dressed up as a marketing problem.
"Daniel walked in, saw the whole system at once, and named the two levers that mattered. Inside 60 days we had a weekly rhythm and pipeline actually built to our ICP."
Start with a 20-minute growth diagnosis. You will know whether the right next step is a sprint, advisory rhythm, fractional support, or no engagement at all.
Last updated: 11 May 2026