A stronger case.
Built on evidence.

A structured family profile connects the child’s needs, parenting style and everyday challenges. Expert rules rank the goals, then a ten-day plan helps the family work on one priority.

The current protocol uses sixteen defined topics and structured answers. A family works on one prioritized goal. Daily feedback supports small changes; the ten-day review determines whether to continue, change activity or refer to a professional.

Quick help, a parent journal and a shareable specialist summary support this core loop. Environmental context and wider age coverage remain later product scope.

Internal simulated cases test the expert decision rules before family access. Independent blind review is planned during the first cohort. The pilot measures usefulness, repeated use and agreement with expert guidance; it does not establish clinical effectiveness.

All three cohorts contribute from the first family under predefined measures. Changes to rules, content and product versions must be logged so results can be interpreted across cohorts. Conversion in the paying cohort is directional evidence on a small sample.

These are proposed decision thresholds, not measured outcomes. Freeze definitions and denominators before recruitment. Paid conversion informs the next test rather than becoming a precise population estimate.

M18 target: at least one in four trial families reaches two payments. Renewal, useful activity and acquisition cost will show whether that translates into a sustainable subscription.

Management validation targets, not achieved results. M1–M18: January 2027–June 2028. Targets apply to cohorts with enough follow-up; feasibility must be tested.

Activation supports the pilot

The pilot’s main usage measure remains ≥40% active on day 20. The existing ≥60% activation threshold means reaching the first daily check-in, not proving benefit. Track profile, plan, check-in, attempted application and reported usefulness separately, including timing in the first seven days. Test their link to D20 and renewal; there is no new 48-hour target.

How to read opt-out subscription evidence

Evidence builds in three stages

Keep targets and the financial base distinct

The financial base keeps its conservative 10% conversion assumption. Operational targets of 35–40% do not automatically increase the revenue forecast. The existing 1,201-payer ambition still needs monthly cohort validation.

Reference context

Targets are management choices informed by subscription research. The reference populations differ; none is a specific benchmark for Croatian AI parenting. Feasibility depends on Harmonia’s observed cohorts.

a16z · Activation & retention · CRV · Seed metrics · RevenueCat · 2026 · RevenueCat · Renewals · RevenueCat · Trial length

We begin with Croatian families raising children aged 3–7. The plan connects the number of children to the households they live in, with one subscription per family.

The initial bridge is 175,000 children × 0.80 = 140,000 households, equivalent to 1.25 target-age children per household. These are the plan’s estimates, not a new census cross-tabulation. The factors account for siblings; children and households must not be added together.

The workbook references 2019–2023 births for the launch estimate. Cohort dates must be matched to the actual launch year and adjusted for migration and resident population. DZS reports 170,475 live births in 2020–2024; that is context for cohorts turning 3–7 in 2027, not a measured household total. DZS source ↗

Harmonia combines child development, family dynamics and parenting style in one structured profile. Expert rules and daily feedback shape a continuing plan, with AI adapting everyday guidance at scale.

Competitor descriptions follow official product sources checked on 9 October 2026. Trade-offs are our assessment of the published service models. Harmonia features and advantages are planned and will be tested in the pilot.

Good Inside · Joy · Joy AI · MissPoppins

The trial is a planned billing design. Store-specific implementation and renewal terms need to be settled before launch. The first two pilot cohorts remain free; the third tests actual payment.

The €16,742 other-services forecast covers packages, institutional services and consultation commissions. These forecasts are separate from subscription ARR and have not been presented as contracted revenue.

The monthly plan totals 17,644.68 registrations and €47,875.06 of scheduled channel spend, including €1,500 for the early DACH waitlist test. The headline rounds registrations to 17,645. The €107,350 marketing and acquisition budget includes €23,400 for the fractional CMO, €27,000 for the acquisition specialist and €7,200 for social media management. The social media role costs €400 per month. Revised market tests and German localization reduce the reserve further.

Campaigns start in May 2027. The first three months allow for expensive learning, with paid-media CAC around €45 per payer and blended warm-up CAC of €37. Later HR acquisition assumes €5 CPM, 4% CTR, 6% click-to-trial conversion and 10% initial trial-to-paid conversion.

The source KPI table reports €25.23 channel CAC, €20.91 contribution LTV and 0.83× LTV/CAC, excluding the €1,500 early waitlist test. The monthly schedule includes that test: €47,875.06 divided by about 1,838 new payers gives €26.05 CAC. Channel CAC excludes the acquisition specialist, fractional CMO and founder time. Total marketing, including fractional CMO and social media, divided by the source’s estimated 1,838 new payers is approximately €58.42; that is a broader spend proxy, not a cohort-matched CAC.

Simple payback divides CAC by monthly contribution and ignores churn. The sensitivity above also shows a geometric retention calculation. It assumes constant monthly churn of 1 / paid lifetime; trial and settlement delays add calendar time.

The closing target of 1,201 follows the rounded calculation 775 × 0.75⁶ + 9,895 × 10.74%. It applies churn to the opening base, then adds H1 conversions as a block. New H1 cohorts are not individually aged within that half-year.

The 2027 model uses 775 premium at year end and an average balance across the year to derive €33,443 subscription revenue. Campaigns begin only in May. The monthly plan schedules acquisition and spending, but does not roll forward active paid subscriptions, trial completion, refunds, tax and settlements into a monthly cash forecast.

The targets are retained as supplied. A dated monthly billing and retention schedule is still needed before treating them as a cash forecast.

€45,000 for a Full stack dev & AI expert at 50% capacity, €4,000 for English localization, €6,500 for German localization after a positive demand signal and €6,000 for servers. Ivan leads technology.

€107,350 includes fractional CMO €23,400, acquisition staff €27,000, social media €7,200, HR paid media €24,650, DACH and UK/Ireland tests €11,000 and supporting channels €14,100.

Independent review, pilot support, supervision and analysis across the 100-family programme.

The first six months allocate €111,131 including €3,159 of reserve. Planned spend excluding that reserve is about €107,972. The decision gate protects the remainder if the programme needs to stop or change.

At a constant average operating spend of €18,917.89 per month, the €350,000 round funds 18 months above the reserve without relying on revenue. This is a budget sensitivity, not a bank cash forecast. The workbook’s total-capital runway uses seed plus an unraised €1 million growth round and is not the seed runway.

Investor materials are available in English, German and Croatian. This language choice does not change the initial Croatian family pilot. Regional content and country-level acquisition budgets still need an explicit allocation: Serbia and Slovenia were discussed, while the workbook’s market table also includes Bosnia and Herzegovina and Montenegro.

In September–December 2027, a €1,500 landing-page and waitlist test measures Austrian and German interest. German localization (€6,500) is conditional on a positive signal. H1 2028 budgets €7,000 for DACH and €2,500 for UK/Ireland. In 2029, DACH is the base case, with priority switching to UK/Ireland if measured results are better. Parallel English tests cover the Nordics, Netherlands, Singapore and UAE. Conditional entry follows in 2030 against the DACH CAC benchmark. Japan and South Korea remain 2032+ with separate localization and a local partner.

Sven Špehar Rajević · CEO · Co-founder

Economist with 20+ years in business development, marketing and project management, including eight years in IT, mainly SaaS. Has piloted Harmonia emotional-resilience workshops with 100+ children. Mother of one. Licensed Relax Kids trainer.

Tihana Starčević · Chief Clinical Officer · Co-founder

Psychologist with 10+ years working with children and families, including SOS Children’s Village and Osijek University Hospital. Training in CBT psychotherapy; author of Steps Through the Storm. Mother of one. Stepmother of three.

Ivan Lozančić · CTO

Co-founder and CEO of Gauss: 60+ engineers, clients in 20 countries and platforms including Gauss Box and ORKA ERP. Deloitte Technology Fast 50 CE recognition. Owner of Gaudeamus private gymnasium. Leads engineering and AI architecture. Technical leadership is distinct from the funded day-to-day engineering role. Father of three.

Budgeted delivery and growth roles

The Full stack dev & AI expert role will take over all beta refinements, including bug fixes and AI integration, under Ivan’s technical leadership.

A Full stack dev & AI expert is budgeted at 50% capacity for €2,500 per month. Ivan provides technical leadership; Ante and Krešo remain external contributors/advisors. The fractional CMO sets channel strategy and oversight; a separate acquisition specialist executes campaigns. Social media management is budgeted at €400 per month. All three roles sit inside marketing and acquisition. Named candidates are not presented as signed appointments.

Advisors

Krešimir Vrhovčak: 20+ years in business analysis, software development and team leadership. Software architect with experience preparing and managing R&D projects.

Ante Magzan: Founder of Afekt; co-founder of Iskon, T-portal, MAXtv and HUB385. Mentor at Nuqleus.

The funding target is €350,000. The proposed investor stake is 15%; this is a proposal, not evidence of agreed terms. At 15%, €350,000 implies €2,333,333 post-money and €1,983,333 pre-money.

These are prior management planning allocations, not verified issued shares or signed grants. The ESOP is already within the 100% pre-round total; the calculator defaults to no additional pool creation.

The source scenario turns EBITDA positive in 2030 and projects €52.68 million revenue in 2035. It assumes longer paid lifetime after seed, improved conversion, optimized acquisition and later DACH and English-market scale. None of those improvements is an observed outcome.

Full-year 2028 includes H2 beyond the seed period. The annual P&L and detailed round budget have different cost scope and timing; a reconciliation is required. Cumulative accounting profitability does not establish repayment to investors, available cash or realized returns.

The analysis distinguishes planning targets from calculated checks. Monthly billing, retention and cash timing need to be validated before larger investment commitments.

Follow-on conversations are planned from November 2027. DACH and UK/Ireland tests in H1 2028 should inform larger market commitments. A waitlist alone does not establish paid conversion or retention.

Financial projections and budgets are based on the Harmonia business plan. Product strategy, team biographies and the pilot design are supplied by the founders.

Financial figures are planning estimates. Customer acquisition, conversion and retention assumptions will be tested with families and market campaigns.

External references: Croatian Bureau of Statistics · Good Inside · Joy · MissPoppins.

Product illustrations show the planned experience. Competitor descriptions and demographic context are attributed to the linked sources.