Context
Elomia is an AI that listens and gives psychological support around the clock. A controlled study at NTU KhPI put it 4% below a human psychologist and ahead of every other bot tested, at $9 a month against roughly $100 an hour.
Half a billion people live with depression and 96% never see a psychologist. The company gives three reasons: a session costs money most people don’t have, seven in ten patients lie to a therapist out of fear of judgement, and a session has to be booked in advance by somebody who can’t predict when they’ll need it. During COVID, calls to support hotlines rose 891%.
Diagnosis
The team wasn’t short of ideas or effort. It was short of evidence. Tracking didn’t support a reliable funnel view, so every test result was arguable.
Targeting was the second problem, and customer development settled it. Five personas, 26 attributes each, and two of those fields decided everything. Every trigger was an emotional state: work stress, loneliness, self-esteem, anxiety, chronic fatigue. And the answers on where and when were consistent. Nobody used Elomia in a therapy hour. They used it at work, or in the evening while doing something else.
The clinical side had the same finding from the other direction. Up to 45% of suicides happen between six in the evening and one in the morning, when almost no psychologist is working. The demand window and the supply gap are the same hours.
Everyone in this category buys media on age, gender, and interests. But nobody opens a mental health app because they’re 34 and live in Chicago. They open it because of how they feel right then, and demographics can’t see that moment.
The product could. Elomia classifies twelve emotions in conversation and follows how a state moves across a day or a week. That signal sat inside the app and nowhere near the media buying. Connecting the two is the engagement in one line: stop describing who to reach, start describing when someone needs reaching.
Decisions
- Fixed analytics first, with the developers, before anything else
- Built the targeting around emotional state rather than socio-demographics, which is the mechanism behind the acquisition cost figure
- Kept it away from the individual: names and locations are stripped before anything is stored, and the buying works on population signals
- Made Twitter the primary channel at $0.03 a click, Google Ads secondary at $0.60 for a better-qualified audience
- Defined the brand as Hero and Caregiver, positioned as "your mental health companion", and voiced every channel as Elomia herself
- Reworked onboarding: fewer steps, explanations earlier, a clearer way in
Results
The targeting system cut acquisition cost by 70%. Lead-to-install conversion rose 32% within a month, and 47.3% of users recommended Elomia to a friend.
The product carried it. 230,000 messages across 7,900 sessions, 91% of users reporting a better mood. On the UK launch the bot ran 1,848 consultations in fourteen days. One psychologist doing that volume would need about 300.
The strategy set February 2021 targets on the back of it: 100,000 leads, 4,000 paying users, $10,000 in monthly recurring revenue, and $1.50 to acquire a paying customer against a $162 lifetime value. That last ratio decides whether any of this becomes a business, and it was written down rather than assumed.
Two things stayed hard. Ad platform moderation collides with mental health constantly, and account blocks forced repeated changes of creative. And the network misread context in about 10% of dialogues, which the strategy recorded rather than hid. Sixteen psychologists covered that gap, handling the 30% of messages the model couldn’t and training it as they went.